www.wiisfi.com Open in urlscan Pro
104.21.86.226  Public Scan

Form analysis
0 forms found in the DOM

Text Content

Wi is Fi


Understanding Wi-Fi 4/5/6/6E/7 (802.11 n/ac/ax/be)
Make your own educated Wi-Fi upgrade decisions
Version 10.9n (updated October 8, 2024)


Chapters

00. Introduction 01.Executive Summary 02.A quick overview of Wi-Fi 03.★Wi-Fi's
weak link: your client 04.★Client 'PHY' speed is key 05.Understanding Wi-Fi
Overhead 06.Router Marketing Hype 07.MIMO - a wireless revolution 08.Wi-Fi 1/2/3
- legacy 802.11 09.Wi-Fi 4 - 802.11n (HT) 10.Wi-Fi 5 - 802.11ac (VHT) 11.Wi-Fi 6
- 802.11ax (HE) 12.Wi-Fi 6E - 802.11ax in 6 GHz 13.Wi-Fi 7 - 802.11be (EHT)
14.DFS channels in 5 GHz 15.Router/Wi-Fi setup tips 16.★How to improve Wi-Fi
speeds
17.Wi-Fi Range Extenders 18.Wireless Access Points 19.Mesh Wi-Fi Networks 20.A
Reality Check 21.★Recommendations 22.Wrapping it all up


Extras

E1. Ethernet / PoE E2.BufferBloat E3.New home construction tips E4.The cost of
'always on' E5.MoCA - Ethernet over Coax E6.Wi-is-Fi Windows Tools
E7.Recommended Cable Modem E8.Powerline Networking / PLC E9.POTS (Telephone) to
Ethernet

 

Appendices

A. Troubleshooting B.Router Specifications C.Netgear 'Mode' D.Throughput testing
E.PHY is asymmetric F.PHY speed tables G.mW, dBm, dB H.Maximizing Wi-Fi range
I.Signal vs distance J.Channel width vs range K.SINR / SNR / Noise floor
L.Router Deep Dive M.WiGig - 802.11ad N.Beware tri-band hype O.Extending to
Outbuilding P.Router/AP Placement Q.What 'Stream' means R.Learn more S.Spectral
Masks T.Terminology V.Version history W.Call for changes / rants X.Supporting
wiisfi.com Y.Contact the author


0. Introduction
This paper details how Wi-Fi works in the United States. While most of this
paper also applies to other countries, there will be subtle Wi-Fi differences
(supported channels, Tx power allowed, etc) for other countries, for which
Wikipedia has the details.

> Faster navigation TIP: As you scroll through this paper, headers are 'sticky'
> and remain docked at the top of the web browser window. Clicking anywhere on a
> header takes you back to the top of this paper.


1.95 Gbps real throughput -- Intel AX210 client to TP-Link AX80
(2402 PHY -- 5 GHz band 160 MHz channel 114)

So you just got fast Internet. Congratulations. But now how are you going to
obtain that fast speed wirelessly to your client devices? Because if you can't,
are you then wasting money (every month!) on an ISP speed that can't actually be
fully utilized?

★ Learn about Wi-Fi and how to obtain real Wi-Fi throughput in excess of 1 Gbps
today...

Like 1.95 Gbps in real Wi-Fi throughput (seen upper right).

1. Executive Summary


TP-Link Archer AX80


The best router/AP (Access Point) value today: A mid-range Wi-Fi 6 router
supporting (1) 4×4 MIMO [§7], (2) all DFS [§14] channels, (3) 160 MHz (HE160)
wide channels, and (4) beamforming is a great value today (as of September 2024;
one example is seen right) for the needs of most people, either (a) as a single
centralized main router, or (b) as an additional 'access point' node in a
growing network. See the Recommendation [§21] chapter far below for details and
other recommendations.

Wi-Fi speeds vs. broadband speeds: Wi-Fi speeds have lagged behind ever
increasing Internet speeds. As a result, there has been a very rapid switch in
Wi-Fi from Wi-Fi 4 [§9] (2.4 GHz 802.11n) to Wi-Fi 5 [§10] (5 GHz 802.11ac), to
Wi-Fi 6 [§11] (5 GHz 802.11ax), and now to Wi-Fi 6E [§12] (802.11ax extended
into 6 GHz) in an attempt to keep up.

> And now the next generation Wi-Fi 7 [§13] is just starting to trickle out.

Router Manufacturers' Marketing Hype: Don't be fooled by the marketing hype [§6]
of router manufacturers advertising outrageously high aggregate (all Wi-Fi bands
maximum speeds added together) Gbps wireless speeds (like "11 Gbps"), and for
client devices that the industry does not yet make (4×4 MIMO client devices).
What really matters is the realistic speed that your Wi-Fi client devices [§3]
can actually achieve today (often that means a 2×2 MIMO limit), for a single
Wi-Fi band.

> The dirty little secret in the router industry is that virtually ALL routers
> (of the same Wi-Fi generation) have the SAME maximum PHY speed to YOUR 2×2
> MIMO client device. And likely very similar throughput speeds when standing
> right next to the router. So the real differentiator is how the router
> performs 'at distance' for your devices (DFS, MIMO, etc), at your location
> (contention, interference, etc).



Wi-Fi's first weakest link -- your client devices: Wi-Fi throughput to a Wi-Fi 5
wireless client device using an 80 MHz channel will likely max out at around 650
Mbps (±60 Mbps) for 2×2 MIMO (and around 1 Gbps for Wi-Fi 6) no matter what new
router is used when right next to the router; and slower at distance. And the
far majority of ALL wireless client devices today (smartphones, tablets,
laptops, etc) are still only 2×2 MIMO. So your client device [§4] is almost
certainly causing 'slow' Wi-Fi PHY speeds (and maybe not your existing
AP/router).


> For Wi-Fi 6 client devices, throughput can be around 1 Gbps 1201 Mbps PHY
> speed for an 80 MHz channel, provided your 2×2 client device is very close to
> the router, and even faster speeds (and an extended range for obtaining 1
> Gbps) are possible with 160 MHz channels.



Wi-Fi's second weakest link -- distance/obstacles: If you are very close to a
Wi-Fi access point and line-of-sight, you CAN expect top Wi-Fi speeds but
limited to the capabilities of your client device. However, as your 'distance
from the router' increases (which honestly, is very typical), there are more
obstacles (walls, floors, etc), and it is 100% normal that Wi-Fi throughput
drops off quickly. In Wi-Fi, you really do always want 'full bars'.

> Fundamental theorem of Wi-Fi: Wi-Fi speed decreases very quickly as 'distance
> from the router' increases.



Installing an Access Point is often a GREAT way to maximize Wi-Fi speeds: Wi-Fi
can work great, but only if you are physically close to the router or an access
point. The closer you are to the Wi-Fi signal source, the better. Which is why
installing an access point [§18] (most often a regular router configured as an
access point), wired/Ethernet back to the main router, and placing it where
needed the most, is an easy and very effective way to dramatically improve Wi-Fi
speeds for far away 'at range' client devices.

So, upgrade Wi-Fi, or not?: The only question that really matters when upgrading
Wi-Fi is: What are YOUR client PHY speeds [§4] and throughput [§D] now and what
will they be after an upgrade? Because, if (the majority of) client speeds will
not increase after an upgrade (especially for 'at range' client devices), what
is the point in spending money on a new router that won't actually improve Wi-Fi
speeds for your client devices?

★ The goal of this paper: This paper was written to help everyone understand
current Wi-Fi technology better, so that YOU can make educated Wi-Fi upgrade
decisions -- because there is WAY too much hype [§6] out there (especially about
Wi-Fi speeds) -- and router manufacturers are directly to blame.

So, let's dig in and learn more about Wi-Fi...

2. A quick overview of Wi-Fi



Why is it called "Wi-Fi": See Here's Why It's Called 'Wi-Fi' (huffpost.com) --
"The origin story is all about marketing". Yes, "Wi-Fi" actually did
(originally) mean "Wireless Fidelity" because that is what the Wi-Fi alliance
(wi-fi.org) actually placed on their website (see right, a capture from May 10,
2000). But the claim today is that "Wi-Fi" has no specific meaning (a tiny bit
of revisionist history) -- and that "Wi-Fi" is just a 'brand' name. And the
branding has worked! Today, virtually everyone has heard of (and uses the term)
"Wi-Fi".

> Other examples of brand names (mentalfloss.com) now used by most as 'generic
> terms' due to wildly successful branding (eg: Jet Ski, Jacuzzi, Kleenex, etc).

First steps: Wi-Fi 1 came out in 1997 and supported PHY speeds of 1 Mbps to 2
Mbps in the 2.4 GHz band. A couple of years later, in 1999, Wi-Fi 2 increased
2.4 GHz band PHY speeds to 11 Mbps.

The big change in 2002: And then everything changed in 2002 with the
introduction of Wi-Fi 3 [§8] into the 2.4 GHz band, increasing PHY speeds from
11 Mbps to 54 Mbps. Wi-Fi 3 introduced the core technologies of Wi-Fi that are
STILL very much used today, like QAM modulation.

Dramatic speed increases: But modulation and coding has only changed slightly
over all Wi-Fi generations. What accounts for the dramatic speed increases seen
in Wi-Fi over the last 20 years is increasing channel widths, increasing MIMO,
and other improvements:

> Typical Wi-Fi 'client device' maximum throughput¹ Version Sym Bits QAM Enc
> MIMO MHz PHY Mbps Wi-Fi 1 802.11 11.0 2 - 1/11 - 22 2 1 Wi-Fi 2 802.11b 11.0 8
> - 1/8 - 22 11 5 Wi-Fi 3 802.11a/g 12.0 6 64 3/4 - 20 54 30 Wi-Fi 4 802.11n
> 14.4 6 64 5/6 2×2 20 144 115 40 300 240 Wi-Fi 5 802.11ac 14.4 8 256 5/6 1×1 80
> 433 375 2×2 80 866 750 Wi-Fi 6 802.11ax 17.2 10 1024 5/6 2×2 80 1201 1000 160
> 2402 1900 Wi-Fi 7 802.11be 17.2 12 4096 5/6 2×2 320 5765 (todo)
> 
> ¹ Sym=Millions of symbols per second; Bits=bits per symbol; MHz=channel width;
> Mbps=Max throughput expected

TIME: All Wi-Fi devices (in the same Wi-Fi generation) are sending/receiving the
same 'symbols', using the same amount of time to send each symbol. See "Sym/sec"
(symbols per second) in the table above. But what does change is the
modulation+coding (MCS) within those symbols -- or, how many 'bits of
information' is sent in each time slot. The closer a device is to an access
point, the clearer symbols can be 'heard' -- and that translates to more bits
that can be sent per symbol. And if you are too far away from an access point,
maybe only one bit can be 'heard' per symbol. So frankly, Wi-Fi works best when
you are physically close to a Wi-Fi access point.

Digital Modulation: The technique that Wi-Fi 3 used to transmit symbols (bits of
information), OFDM, remained largely unchanged until Wi-Fi 6 [§11], when a new
technique was introduced, OFDMA.

MCS Encoding: Each new generation of Wi-Fi (so far) has increased the maximum
amount of bits that can be conveyed every symbol but that requires client
devices to be ever closer to the router. See "Bits/Sym" (bits per symbol) in the
table above.

Spectrum/Bands: Wi-Fi started in the 2.4 GHz band with only 70 MHz of spectrum!
And then later expanded into the 5 GHz band (with 560 MHz of spectrum).
Supporting the 'need for speed' (increasing channel widths), the FCC recently
allowed Wi-Fi to expand into 1200 MHz of spectrum in the 6 GHz band (Wi-Fi 6E
[§12]) for supporting 320 MHz channel widths.

In Wi-Fi, spectrum/channels are shared: It is remarkable that Wi-Fi actually
works as well as it does, because every device using a particular channel
(spectrum) must cooperate and SHARE the time on that channel with every other
device using the same channel (you and close neighbors). It is exactly like
using a Walkie-Talkie -- everyone on the same channel can either talk (Tx) or
listen (Rx), but can not do both at the same time. So Wi-Fi works best when all
Wi-Fi access points (for you and close neighbors) are evenly spread out across
all available channels (not everyone using the same channel).

The big 'gotcha' -- only maximum 'common functionality' can be used: So just
because a new router supports a crazy high speed does not mean that your client
device (camera, tablet, phone, laptop, etc) will all of a sudden support that
new fast speed. Instead, your client device is often the weak link limiting
factor [§3] and may support a smaller channel width, a lower MIMO level, or an
older version of Wi-Fi. The 'lowest common denominator' of capabilities between
any two devices is what is used.

> Key 'weak link' example: In mid-2023, Ring introduced a brand new $60 Ring
> Indoor camera (amazon.com) that only supports (the now very old and slow)
> Wi-Fi 4. So even with the latest and greatest new Wi-Fi 7 router, this Ring
> camera can only communicate to that new router at a maximum PHY speed of 72.2
> Mbps in only the 2.4 GHz band -- Source (fccid.io).

The 'weak link' in Wi-Fi is often YOUR client device...

3. Wi-Fi's weakest link - YOUR client device!


YOUR client device often limits Mbps speeds, not the router/AP
(maximum PHY speed when standing right next to the router)

In summary: Each client device has its own 'maximum speed' at which it can
communicate with a router. Wi-Fi can only operate as fast as the least capable
device in a two-way 'conversation', which almost always is your client device
(camera, tablet, phone, laptop, etc), and not the router.

> KEY Wi-Fi concept: Client device capabilities often limit Wi-Fi speeds, not
> the router/AP.

> Analogy: Just because the speed limit on a highway is 65 does not mean that
> your (speed limited) moped can go that fast. Your actual speed will be the
> speed limit of the moped. Similarly, in Wi-Fi, your speed is often limited by
> your client device.

The weakest link in Wi-Fi is YOUR client device: You have 1 Gbps Internet, and
just bought a very expensive AX11000 class router with advertised speeds of up
to 11 Gbps, but when you run a speed test from your iPhone XS Max (at a distance
of around 32 feet), you only get around 450 Mbps (±45 Mbps). Same for iPad Pro.
Same for Samsung Galaxy S8. Same for a Wi-Fi 5 laptop computer. Same for most
wireless clients. Why? Because that is the speed expected from these Wi-Fi 5
(2×2 MIMO) devices! This chapter explains in great detail exactly why that is.

> Another example: The Galaxy Tab A8 (a 10.5" tablet), even when connected to
> the latest and greatest Wi-Fi 7 router, is limited to a PHY speed of 433 Mbps
> (and max throughput around 300 Mbps) -- because this intentionally
> 'affordable' class device only supports 1×1 MIMO.

You may safely skip to the next chapter [§4] for a shortcut if this chapter is
too detailed/technical for you. The rest of this chapter is a "deep dive" into
everything that limits speed when a Wi-Fi 5 client device communicates with a
Wi-Fi 5 router.



  AC5300 4×4 Router to 2×2 Client
(at a distance of 32 feet)

5300 ➤ 2166 ➤ 1083 ➤ 866 ➤ 650 ➤ 455  


  AC5300 4×4 Router to 4×4 Client
(at a distance of 32 feet)

5300 ➤ 2166 ➤ 1733 ➤ 1300 ➤ 910  




AC5300 rating: How did your router even get a 'rating' of 5300 Mbps in the first
place? Router manufacturers combine/add the maximum physical network speeds for
ALL Wi-Fi bands (usually 2 or 3 bands) in the router to produce a single
aggregate (grossly inflated) Mbps number. But your client device only connects
to ONE band (not all bands) on the router at once. So, '5300 Mbps' is all
marketing hype [§6].

The following details how the grossly inflated speed of 5300 Mbps is reduced
down to a 'real-world' speed of only 455 Mbps...

◆ 5300 ➤ 2166: Maximum ONE band speed: The only thing that really matters to you
is the maximum speed of a single 5 GHz band (using all MIMO antennas). You find
out by looking at the 'tech specs' for an AP/router. 5300 is just 1000 + 2166 +
2166, where 1000 is the 2.4 GHz band speed and 2166 is the 5 GHz band speed.
2166 also is a tip-off that this router is a 4×4 MIMO router (by looking for
'2166' in the speed table, right).

> More on bands in the Beware tri-band marketing hype chapter far below.

◆ 2166 ➤ 2166: Realistic 80 MHz channel width: Router manufacturers cite speeds
for 2.4 GHz using 40-MHz channel widths, but a 20-MHz channel width is much more
realistic (that cuts cited speeds in half). For 5 GHz 802.11ac, speeds are
typically cited for an 80-MHz channel width, which all AC clients are required
to support. But if cited speeds are for a 160-MHz channel width (that is
starting to happen for the new Wi-Fi 6 routers), cut the cited speeds in half
(as most clients won't support that).

◆ 2166 ➤ 1083: Client 2×2 MIMO: Which MIMO column do you use in the Wi-Fi speed
table (right) -- The MIMO of the router or the MIMO of the client device? You
must use the minimum MIMO common to both devices (often the client). So if you
have a 4×4 router, but use a 2×2 client (like the Apple iPhone XS Max or Samsung
Galaxy S8) to connect to it, maximum speeds will be instantly cut in half (2/4)
from cited router speeds.

> Wi-Fi specifications for iPhone (apple.com) or iPad (apple.com). Virtually all
> newer iOS devices are 2×2 MIMO and older iOS devices are 1×1 (no MIMO).

◆ 1083 ➤ 866: Client 256-QAM: You can only use the maximum (common) QAM
supported by both the router and the client. Router manufacturers may cite
speeds for 1024-QAM (which the router DOES support), but you will only get that
if your clients supports that QAM (many do not) and you are very close to the
router (sometimes only just feet away). So reduce to a much more realistic
maximum of 256-QAM 5/6.

◆ 866 ➤ 650: 32 feet from router (Modulation/Coding): Router manufacturers love
to cite the maximum PHY speed possible, which you will only get when you are
very close (just feet) to the router. But as you move farther away from the
router, speeds gradually decrease. The 'distance' issue is represented by rows
in the PHY speed table (seen upper right). At just 32 feet away from the router
(a very typical distance), 64-QAM 5/6 was actually observed, so use that. For
more details, see the next chapter [§4].

> Analogy: Understanding Modulation/Coding: Imagine that once a second (the
> 'symbol rate' for this channel), you hold up your arms in various positions to
> convey a message to someone else. If you were only ten feet away from that
> person, the number of arm positions reliably detected would be very high. But
> now move 100 feet away. The number of arm positions reliably conveyed would be
> reduced. Now move 500 feet way. The number of arm positions reliably conveyed
> might be reduced to just 'did the arm move at all'. The same thing happens in
> Wi-Fi. If you are close to the AP/router, a large number of bits can be
> conveyed 'at once' (within each symbol). But as you move away, a smaller and
> smaller number of bits can be reliably conveyed 'at once'. So
> 'modulation/coding' is simply how much information can be conveyed at once,
> and is directly related to distance from the AP/router.

◆ 650 ➤ 455: Wi-Fi overhead (MAC efficiency): What is the overhead at the
network level? All of the speeds we have discussing so far are for PHY
(physical) network speeds. But due to Wi-Fi protocol overhead, speeds at the
application level are around 60% to 80% the physical network level. So use 70%
as a fair estimate of throughput you can expect to see. 70% of 650 is 455 Mbps.

> Just Google 802.11ac MAC efficiency (google.com) to understand this issue. In
> short, there are 'housekeeping' packets that MUST be sent at the SLOWEST
> possible modulation, and that takes time and slows everything down (along with
> other issues, see understanding Wi-Fi overhead [§5])
> 
> Analogy: You are on a road going 60 mph, but every 100 feet you must slow down
> to 1 mph for 1.5 feet. Do the math - your average mph is deceptively much
> lower than you might think. Because what matters is not the (minimal) distance
> traveled at the slow speed, but the TIME that it takes.

◆ 455 ➤ ???: Interference/Contention: So, the final number is 455 Mbps for a 2×2
device (at a fair distance away from the router), but only if your device gets
exclusive use of ALL time left in the Wi-Fi channel. But there may (or may not)
be other Wi-Fi users (either local, or even neighbors on the same spectrum)
which will decrease your speed by some unknown amount.

Results: 2×2 MIMO devices get a realistic (maximum) download speed of 455 Mbps
(±45 Mbps) at around 32 feet, which is dramatically lower than the '5300 Mbps'
advertised by router manufacturers.

A lesson learned: The critical factors that greatly impact and determine maximum
real-world speed for a single client are: (1) lowest common MIMO level, (2)
lowest common channel width, and (2) MAC efficiency.

> An analogy for all of the above: What if I built a three lane toll road from
> Washington, DC to New York, NY, and sold passes for chauffeured rides with
> speeds "up to 74 mph" (AC5300). But after you pay for a ride, you discover
> that the speed limit is 30 mph (AC2166) on two of the lanes and 14 mph on the
> third lane. So you take the 30 mph lane, but find out your chauffeur is only
> driving at 10 mph (MIMO level and QAM) and worse yet, every 100 feet the
> chauffeur slows down to 1 mph for 10 feet (MAC efficiency). Your average speed
> is 5 mph. And yet, that is exactly what router manufacturers are doing to you
> -- AC5300 is really only 455 Mbps for most wireless devices -- just like 74
> mph is really 5 mph.


4. Client 'PHY' speed is the key (limiting) factor
YOUR client device is the key (limiting) factor for the speed (and maximum
distance) at which your device connects to a router (a modern router is rarely
the limit; for technical details, see prior chapter [§3]). Stay in this chapter
for a fast shortcut.

> KEY Wi-Fi concept: Client device capabilities often limit Wi-Fi speeds, not
> the router/AP.

For a modern 2×2 MIMO [§7] client device, expect 'top PHY speeds' standing right
next to a modern router of:
 * Wi-Fi 1 [§8], 2 Mbps
 * Wi-Fi 2 [§8], 11 Mbps
 * Wi-Fi 3 [§8], 54 Mbps
 * Wi-Fi 4 [§9], 144 Mbps (20 MHz), or 300 Mbps (40 MHz) -- 64-QAM, 2×2 MIMO
 * Wi-Fi 5 [§10], 866 Mbps (80 MHz), or 1733 Mbps (160 MHz) -- 256-QAM, 2×2 MIMO
 * Wi-Fi 6 [§11], 1201 Mbps (80 MHz), or 2402 Mbps (160 MHz) -- 1024-QAM, 2×2
   MIMO
 * Wi-Fi 6E [§12], 2402 Mbps (160 MHz) -- 1024-QAM, 2×2 MIMO

and then according to the fundamental theorem of Wi-Fi: Wi-Fi speed decreases
very quickly as 'distance from the router' increases.

> IMPORTANT: After you find PHY speed (below), it is best to run speed tests
> [§D] to validate actual throughput.

The limit: With a new modern Wi-Fi 6 router, it is virtually never the router
that has the speed limit, but rather, it is the client device (that is NOT as
capable as the router) that limits speeds. For example:




Your device (not the router) is almost certainly limiting Wi-Fi speeds


The problem (of finding maximum speed): So how do you find the maximum
(realistic) wireless speed of a client to an AP/router? You could just run a
speed test, but if the speed is not what you expected, where is the problem --
the client, the router, the Internet, interference, elsewhere, or is the
speedtest accurate?

The solution: Go to your wireless device and find the PHY speed (the maximum raw
bitrate between the device and your AP/router) and take 70% of that PHY speed to
estimate maximum application speed (the next chapter [§5] explains why the
overhead is so large). Then lookup the PHY speed number in the PHY speed tables
[§F] to then find which MIMO level is currently being used.

> Expect throughput anywhere from 60% to 80% of PHY speed. So use 70% (±10%) as
> a fair estimate.
> 
> 
> If you don't see 60% to 80% of the PHY speed in a throughput test, then
> either: (1) the PHY speed you have is not accurate, or (2) there is contention
> on the channel, like another device is actively using the channel.
> 
> > UPDATE: The more testing I do, the more it seems like the throughput speed
> > test is more informative about what actual PHY speed was, than the other way
> > around. Namely, throughput speed divided by around 0.7 to 0.8 is a great
> > estimate of what average PHY speed was. Why? Because for me, there are times
> > that reported PHY speed is great, but as soon as I saturate the channel with
> > a speed test, reported PHY takes a dip (during the test), and then after the
> > test, reported PHY goes back to being great. The lesson learned is to
> > scrutinize and collect reported PHY speed during a speed test (not before
> > and not after).
> 
> New client devices often outperform older devices: In a real-world test, with
> Wi-Fi 5 client devices reporting a 2×2 866.6 Mbps PHY link, I measured 461
> Mbps (53%) download speeds on one (very old) computer, 540 Mbps (62%) on a
> second computer, and 673 Mbps (78%) on a third (brand new) computer. All
> tested just feet away from the same R7800 router. And when testing Wi-Fi 6E
> devices running in Wi-Fi 5 mode, I get 700 Mbps (81%). So brand new client
> hardware seems to perform much better than years old client hardware.
> 
> PHY speed: When any Wi-Fi device reports a PHY speed to you, take that value
> with a huge grain of salt. Due to how Wi-Fi works, a device is often
> constantly adjusting PHY speeds adapting to noise and interference, so the
> single PHY speed you see is often a best case maximum PHY speed. Namely a
> device could be using PHY speeds of 78, 104, and 52 Mbps every single second,
> but will report a PHY speed of 104 Mbps (the maximum) to you.
> 
> PHY speed also reveals: (1) channel width, (2) modulation/coding (distance
> from router), and (3) minimum MIMO level support. Please note that the PHY
> speed displayed is not a static value, but changes over time, depending upon
> distance from AP/router, interference, etc.
> 
> PHY speed tables: If you don't find your PHY speed in the PHY tables in this
> paper below [§10], look up the speed in the full PHY speed tables
> (docs.google.com).

TIP: Before looking up PHY values on your device (below), cause some Internet
activity. You want an up-to-date PHY value displayed, and not an old stale value
(which can happen with no Internet activity). And when possible (eg: Windows),
lookup the PHY value at the same time that a speed test is actively running.


Windows 10/11 (Wi-is-Fi way): Run the WI.BAT tool [§E6]. Best because the
information is constantly refreshed -- especially helpful during speed tests.
Also, the tools [§E6] section gives some commands to run manually from a Windows
'command prompt'.

Windows 11 (new way): Right-click on the Wi-Fi icon in the taskbar. Click on
"Network and Internet Settings". Click on "Properties" and scroll down to "Link
speed (Receive/Transmit)". You will see something like this:

> 

Windows 10 (new way): Go to the 'Settings' app, click on 'Networking &
Internet', click on the 'View your network properties' link and find the
transmit/receive speed under your 'Wi-Fi' adapter. However, I suspect that
sometimes transmit/receive are just a single value displayed twice instead of
two actual speeds.




Windows PHY Speed

Windows 11/10/8/7 (legacy mode): In the Windows "Control Panel", search for and
then click on "Network and Sharing Center", then click on the named wireless
connection (which opens a 'status' dialog), and look for the 'Speed' (example
seen right).

> Example: Lookup 702 Mbps speed (right) in the PHY tables far below [§10] and
> it is not found. So, go to the full PHY speed tables and you will find various
> matches, but only one makes logical sense: 80 MHz channel, 2×2 MIMO, 256-QAM.
> 
> The PHY speed reported by Windows appears to actually be the maximum of both
> the Tx PHY and the Rx PHY speeds. Some tests showed the speed reported as the
> Rx PHY speed. Other tests showed the speed reported as the Tx PHY speed. Also,
> if Windows is getting the PHY speed from the Wi-Fi driver, this observation
> could be very Wi-Fi device (and vendor) specific.

Mac: (1) Hold the option/alt key down and click the Wi-Fi icon in the menu bar
and look for the "Tx Rate". (2) Run the "Network Utility" (under Applications /
Utilities; or use Spotlight to find) and look for the Wi-Fi "Link Speed". More
info on finding Link Speed on a Mac (tweaking4all.com).

iOS (iPhone/iPad): Not known within iOS itself (tell me if you know how).
However, to find the maximum PHY speed and MIMO level for your iOS device, visit
the Wi-Fi specification details for iPhone (apple.com) or iPad (apple.com), All
modern iOS devices are 2×2 MIMO. Here are some tips I received from readers of
this paper:

> Reader TIP #1: if you happen to use Apple's AirPort Wi-Fi base station:
> "Install Apple's 'AirPort Utility' and then open it. Click on the Wi-Fi base
> station. Click on 'wireless clients' and then click on your iOS device and
> then 'connection'. This will give you the iOS device 'PHY' connection speed."
> 
> Reader TIP #2: Consider using the "nOversight" app (free version), which shows
> Tx/Rx rates, plus a lot of other detailed Wi-Fi information.



Android (BEST way): Install the "WiFiAnalyzer (open-source)" (play.google.com)
Android app to get a ton of detailed Wi-Fi information.




Android PHY Network Speed

Android (may no longer work): Go into "Settings / Connections / Wi-Fi", click on
the connected Wi-Fi network, and find the 'Network Speed' (example right). Or,
it may also be called 'Link Speed'. But it looks like this method no longer
works -- Android removed displaying the speed, so use the app method (above).

> Example: Lookup 585 Mbps (right) in the PHY tables [§10], and you will find it
> in multiple columns, but given the client, what makes the most sense is: 80
> MHz channel width, 2×2 MIMO, and 64-QAM.
> 
> It appears that Android reports the "Tx PHY" as the 'link speed'. This is
> preliminary and needs more research.

Kindle: Under Settings, click on "Wireless", then click on the connected Wi-Fi
network, and look for "Link speed".

Chromebook: Open "crosh" on your Chromebook (CTRL-ALT-T) and type "connectivity
show devices" and look for the Link Statistics Transmit Bitrate. You should see:
(1) the Mbps transmit bitrate, (2) the MCS index number, (3) the channel width
in MHz, and (4) the number of spatial streams. Type "exit" to exit/close the
crosh window.

OR, find client device PHY speeds in the router's web interface or app: Instead
of looking at client devices, an alternative is to look at your router's web
interface, or app, for client device PHY speeds:

 * 
   
   TP-Link TIP: In the TP-Link router web interface, under "Network Map", click
   on 'Clients' to see a list of all clients, along with "Tx/Rx Rate(Mbps)" PHY
   speeds (example seen right). Very helpful information to see all devices in a
   single list, along with PHY information! Best when client devices are
   actively using the Internet.
 * Netgear TIP: In the Netgear 'Nighthawk' router app, click on 'Device
   Manager', then click on a client device, and a "Link Rate" will be displayed.
   But Netgear displays a 'link rate' that is slightly too small. To correct (to
   bps), multiply by 1024/1000 thanks to reader Matthew S. for pointing out that
   correction.
 * Comcast TIP: If you use a Comcast provided cable modem / gateway device,
   connecting to the http administration interface, signing in, and clicking on
   the 'View Connected Devices' button will take you to a page that shows the
   "RSSI Level" (in dBm) for all Wi-Fi connected devices. Very helpful for
   quickly spotting devices 'in trouble'.

Wi-Fi 7 Devices DeviceMIMO  OnePlus 11 5G2×2EHT320 Google Pixel 82×2HE160


Wi-Fi 6E Devices DeviceMIMO  Apple iPhone 15 Pro2×2HE160 Apple iPad Pro (high
end)2×2HE160 Dell Laptops (high end)2×2HE160 Samsung Galaxy S22 / S232×2HE160
Google Pixel 6/72×2HE160


Wi-Fi 6 Devices DeviceMIMO  Apple iPhone 11, 12, 13, 142×2HE80 Apple iPad Pro
(low end)2×2HE80 Apple iPad Air2×2HE80 Dell Laptops (high end)2×2HE160 Samsung
Galaxy S202×2HE80


Wi-Fi 5 Devices DeviceMIMO  Apple iPhone X,8,7,6s2×2VHT80 Apple iPhone 61×1VHT80
Apple iPad + Air22×2VHT80 Apple iPad Air1×1VHT80 Dell Laptops (high
end)2×2VHT160 Dell Laptops (low end)1×1VHT80 Fire TV (gen 2 and later)2×2VHT80
Galaxy S9,S8,S7,S62×2VHT80 Google Pixel 5,4,3,2,12×2VHT80 MacBook Pro
(some?)3×3VHT80


 

Wi-Fi 4 Devices DeviceMIMO  Ring Indoor Cam 20231×1HT20

MOST client devices today are 'stuck' at 2×2 MIMO: As can be seen from the
tables (right), most client devices today are STILL only 2×2 MIMO [§7]. Why
haven't devices switched to 4×4? Because (1) there is (currently) no compelling
need for that speed today (there is no app that 'requires' 400 Mbps to function)
and more importantly (2) the increased speed is not worth the tradeoff in
greatly reduced run time for battery powered devices.

> Supporting 4×4 MIMO takes more power, and for battery powered devices, runtime
> is FAR more important.
> 
> You will see the spec sheets for many modern phones that MIMO is 4×4, but look
> closely and notice that this is only for cellular, not Wi-Fi.
> 
> However, of note is that many (non-Apple) Wi-Fi 6/6E devices support 160-MHz
> channels (HE160), which instantly doubles throughput vs Wi-Fi 5 using 80-MHz
> channels (VHT80). So speed is increasing dramatically via wider channels, not
> increased MIMO levels.

You can expect a maximum PHY speed of 866 Mbps, and around 650 Mbps (±60 Mbps)
throughput, from a Wi-Fi 5 (802.11ac) 2×2 client device. It is noteworthy to
point out that Dell apparently had a 3×3 laptop in the past, but today, I can't
find anything other than 2×2 laptops on Dell's website.

A final warning: This discussion about 'the' PHY speed of your device is
slightly over simplified, as for every Wi-Fi device, there is actually a Tx
(transmit) PHY speed and a Rx (receive) PHY speed, and those two speeds are
almost always different (asymmetric). But even when different, the two speeds
are relatively close to each other, so the asymmetry is rarely noticed. See the
PHY speed is asymmetric [§E] appendix below for more details.

> A final wrench in the PHY puzzle: And PHY speed is not 'constant'. Unless you
> are right next to the router (with a fantastic signal strength and PHY speed
> is highest possible speed), PHY speed is actually constantly changing up and
> down between MCS levels, adapting to changing signal strength conditions. It
> is not uncommon with a single second to see 3 to 4 different MCS levels (PHY
> speeds) used. Are you highly technical? Then use the Router deep dive [§L]
> appendix below to determine exactly what MCS indexes are being used for both
> Tx PHY and Rx PHY.

Another client device limitation: Range: The maximum distance at which a device
can connect to an AP/router is (almost always) determined NOT by the power
output of the AP/router (around 950 mW is typical), but the power output of the
client device (around 50 mW to 250 mW typical), as client devices almost always
operate at lower power levels than the AP/router. The implication of this is
that the Tx PHY speed from a client device to an AP/router is almost always
lower (hits the limit sooner) than the Tx PHY speed from the AP/router to the
client. Full details [§H].

5. Understanding Wi-Fi overhead
Throughput to PHY Ratio (TPR): The TPR for a Wi-Fi connection is the ratio
between (a) measured 'throughput speed' and (b) rated 'PHY speed':

> 

Under ideal conditions, expect a TPR around 70% (±10%). The rest (up to 100%) is
'overhead'. So why is overhead 20% to 40%?

> KEY Wi-Fi concept: Expect TPR (throughput to PHY ratio) to be around 70%
> (±10%) -- or anywhere from 60% to 80%.

For example, obtaining a throughput of 1955 Mbps for a PHY speed of 2402, means
a TPR of 1955/2402, or 81% (or around 19% overhead), a great result. And seeing
a TPR of 50% is a huge tip-off that the Wi-Fi connection is having issues, like
interference, contention, etc.

> What I am beginning to realize is that the PHY speed reported by Wi-Fi devices
> is often a 'best case' maximum value (and may not accurately reflect the
> actual PHY speed used). The MCS Spy tool (duckware.com) has been invaluable in
> detecting and visualizing this during a throughput speed test. PHY speed also
> can fluctuate a lot up and down, and is not a 'fixed' single value. Also, the
> newer the device the better. There appears to be some issues with older
> 802.11ac devices (not achieving top rated speeds).
> 
> 
> 
> With both a modern client device and a modern router, and no one else using
> Wi-Fi -- so NO contention -- I regularly measure a TPR of 80%.

Wi-Fi overhead can be surprisingly 'large': So, if your smartphone connects to
your AP/router at a PHY speed of 702 Mbps, why doesn't your smartphone get that
full speed? Instead, 70% of your PHY speed (70%×702=491 Mbps) is a fair estimate
of actual (maximum) Mbps seen, but why?

> PHY speed in Wi-Fi is exactly like the speed limit (sign) on a local road. You
> can go that fast some of the time but clearly not all the time. Because you
> must take into account known slow downs: stop signs, turns, traffic lights,
> traffic, school zones, weather conditions, etc. And in Wi-Fi, there are a lot
> of slow downs that also add up.

TIP: The efficiency of TCP/IP over Ethernet, with a MTU of 1500, is 1460/1538,
or (1500-20-20) / (1500+38)), which translates to a maximum possible
(application level) speed of 949.28 Mbps for Gigabit Ethernet (50.72 Mbps
overhead). Details (wikipedia.org)

First, there is TCP/IP and Ethernet overhead: On wired Ethernet, you can expect
around 5% overhead for TCP/IP and Ethernet, or 95% throughput at the application
level. As a ballpark figure, assume something very similar for Wi-Fi. Just
remember that part (around 5%) of the total overhead you are seeing in Wi-Fi is
actually coming from TCP/IP and Ethernet protocol overhead, and not Wi-Fi
itself.

Management transmissions must be sent at the 'slowest' possible modulation: In
order to guarantee that ALL devices on a channel (AP and clients) can
receive+decode management transmissions, those transmissions must be transmitted
at the slowest possible modulation -- so that devices that are furthest away
from the AP (and hence, running at the slowest speed) can receive and
successfully decode those transmissions. For example, 802.11 'Beacon Frames'
(typical send rate is once every 102.4 ms). And this 'slow' speed can be as slow
as 1 Mbps (2.4 GHz band) or 6 Mbps (5 GHz band). When compared to 433 Mbps and
866 Mbps, that 'slow' speed is a hit.

> SSID overhead: The overhead per SSID (on one channel) can be anywhere from 3%
> to incredibly high. See this article (archive.org), using a 802.11b 1 Mbps
> beacon rate, for details.
> 
> A striking analogy: You are on a road going 120 mph for 5240 feet, and then 1
> mph for 40 feet -- What is your average mph for the entire mile (5280 feet)?
> -- MUCH lower than you might initially expect. The answer is
> 5280/(5240/120+40/1), or 63 mph! It is not the distance that you slowed down
> that is important, but rather the time you spend slowed down that really
> matters (as compared to the time you spend going fast).

Half Duplex: There is no separate download spectrum and upload spectrum in Wi-Fi
whereas Ethernet is full duplex - can send and receive at the same time.
Instead, there is only a common spectrum (channel) that ALL Wi-Fi devices
(router and clients) operating on that channel must use in order to transmit
(and receive!).

> So when you are running an ISP download throughput speed test, your device is
> mostly receiving, but it is also transmitting (acknowledging data sent)! Using
> the MCS Spy tool, a PC downloaded at 120 Mbps, but was uploading at 1.8 Mbps
> at the same time. See Router deep dive [§L] appendix below. This is simply due
> to how TCP/IP works. And almost always, the client transmits back to the AP at
> a slower MCS than the MCS the router uses to transmit to the client. So
> because Wi-Fi is half duplex, there may be around 1% to 3% (relative)
> 'overhead' simply due to how TCP/IP works (acknowledgements).
> 
> Analogy: A Wi-Fi channel is just like a Walkie-Talkie, where you can either
> talk, or listen, but not both at the same time. OR, just like a narrow bridge,
> where cars (Wi-Fi traffic) can either go one way or the other, but not both
> ways at the same time.
> 
> Consider the impact on a Wi-Fi channel of a video conference call that is both
> transmitting and receiving 5 Mbps 1080p HQ video streams. 10 Mbps of channel
> capacity is used.

CSMA/CA: The Wi-Fi spectrum is a shared resource. So how does a device know that
it is OK to transmit? Wi-Fi uses something called CSMA/CA (wikipedia.org) , or
Carrier-Sense Multiple Access with Collision Avoidance. So any device on a
channel that wants to transmit must first 'sense' that the spectrum is
available/unused. And to ensure 'fairness' to all Wi-Fi stations that want to
transmit, all 'want to transmit' stations wait a random amount of time before
transmitting (if the spectrum is still unused at that point, transmit, and hope
for no collisions). More info (archive.org).

> And if you have a lot to transmit, that 'wait for a random amount of time'
> over and over adds up. But that random wait is necessary to ensure 'fairness'
> to other Wi-Fi devices.
> 
> CSMA/CA works very well when there are not many devices all wanting to
> transmit at the same time (which IS typical in Wi-Fi, which is why Wi-Fi
> mostly works so well). But overhead can increase dramatically if there are too
> many devices all wanting to transmit at the same time (due to collisions; see
> below).

Collisions/Retransmissions: When multiple devices want to transmit at once (as
the channel gets busy), the possibility of collisions (more than one device
transmitting at the same time) increases, causing that entire transmission to be
lost, and a future retransmission. Or a transmitted packet just did not make it
(too much interference).

> From a test AP, there were 1,519,932 packets transmitted and 48,878 packets
> retransmitted. So, around 3% of the data packets had to be retransmitted.

Acknowledgements: Every Wi-Fi packet sent must be 'acknowledged' at the Wi-Fi
level (to confirm receipt). To accomplish this, each sent packet has a little
bit of an extra reserved space (a 'time window') appended to the end of the
packet, for the receiver to transmit back (during the empty 'time window') an 'I
got it' acknowledgement (to the sender).




Wi-Fi hidden node problem

Hidden Node Issue: There is something called the Hidden Node Problem
(wikipedia.org) that can (potentially) cause a large number of collisions in
Wi-Fi -- where device 'A' and device 'B' can both hear transmissions from the
AP, but device 'A' and device 'B' can NOT hear each other's transmissions. So
both device 'A' and device 'B' might transmit at the same time (as seen at the
AP, a 'collision') and both transmissions are lost (at the AP).

> A mitigating factor is that even if your network has the hidden node problem,
> the hidden nodes will not impact each other, unless they attempt to use Wi-Fi
> and transmit at the exact same time. If both hidden nodes are sporadically
> using Wi-Fi, the problem will not happen that often.

Coexistence with 802.11 a/b/g/n: For an 80 MHz 802.11ac channel to properly
coexist with older 20 MHz radios operating within the channel, there is a
'request to send' and 'clear to send' exchange before each real message is sent.
And that slows everything down. More information (computerweekly.com)

Beamforming overhead: The sounding frames for beamforming adds a tiny bit of
overhead. How much overhead this causes needs to be researched.

CRITICAL: Don't forget that Wi-Fi is a shared resource: After all of the above
(which assumes you have the Wi-Fi channel all to yourself), if you are unlucky
enough to have a router set to the same channel as your neighbor (and your
neighbor is using Wi-Fi), you are sharing spectrum/time/bandwidth with your
neighbor!

> How is Wi-Fi spectrum shared? By bandwidth? By time? By something else? In
> general, by TIME -- if 'N' users all want to use Wi-Fi at the same time, on
> average, they will all get to use the channel '1/N' of the time. For example,
> if two users want to use the same channel, and first user at a PHY of 6 Mbps,
> and the second user at a PHY of 866 Mbps, the first user will get to use the
> channel 50% of the time (so 6/2, or around 3 Mbps), and the second user will
> get to use the channel the other 50% of the time (so 866/2, or around 433
> Mbps).
> 
> This 'a channel is shared' concept is easy to gloss over and not fully
> understand. But if you see (in a Wi-Fi analyzer app) that your wireless router
> and other wireless routers (neighbors) on the same channel (or overlapping
> channels), you are sharing the SAME Wi-Fi spectrum.
> 
> A striking example: A laptop in the same room as an AP (on the second floor of
> a house) with an unobstructed view of the AP gets real-world Tx average
> throughput of 92 Mbps -- measured using TxRate (duckware.com). But when the
> laptop is moved downstairs (so now the Wi-Fi signal has to go through
> walls/floors), Tx throughput increases to a rock solid 116 Mbps. But how is
> that possible? -- throughput should have decreased! The surprising answer is
> that downstairs, neighboring Wi-Fi signals were greatly attenuated, so the
> laptop transmitted more often. And yes, that means the laptop WAS actually
> transmitting at the same time as a far away station but all that did is raise
> the noise floor for those transmissions, which were still successfully
> received.

A final caveat: PHY speed is a very complicated thing. Tx PHY and Rx PHY can not
only be asymmetric (more details in appendix below [§E]) but also be highly
variable. The 'link speed' your device reports to you is a highly
over-simplified 'computed' single number. You should only use that speed as a
'ballpark' figure of actual PHY speeds used. Or, when you run a throughput test
and attempt to calculate the 'overhead' at the PHY level, that 'overhead' is
only an estimate.

> A prime example: A Windows laptop with an 'older' 802.11ac 2×2 MIMO four feet
> from the router reports (an expected) 'speed' of 866.6 Mbps (MCS9). A
> throughput tests shows download speeds of 475 Mbps. That is a MAC efficiency
> around 55%. But the MCS Spy tool (see Router deep dive [§L] appendix below)
> clearly shows that the router is transmitting to the PC using only MCS7 (650
> Mbps), which is actually a much better MAC efficiency of around 75%. There is
> still the problem of why MCS9 is being reported and MCS7 used, but MAC
> efficiency is much better than it initially appears. Take the PHY speed
> reported by your client device with a 'grain of salt'.

Learn More:
 * Wi-Fi Overhead, Part 1: Sources of Overhead (cwnp.com)
 * Wi-Fi Overhead, Part 2: Solutions to Overhead (cwnp.com)
 * RTS/CTS enhanced with bandwidth signaling (cisco.com)


6. Cutting through router marketing hype



The bottom line: The AX#### naming convention (AX6000, AX11000) and AC####
naming convention (AC1900, AC2600, AC5300, AC7200) used in the router industry
(where the #### is a maximum combined Mbps) is nothing more than marketing
hype/madness.

The naming convention implies (incorrectly) that the larger the number, the
better and faster the router -- and the faster Wi-Fi will be for your wireless
devices. Also, speeds are cited for hypothetical wireless devices that DO NOT
EXIST -- can you actually name a single smartphone, tablet, or laptop computer
that has 4×4 MIMO for Wi-Fi?

> KEY Wi-Fi concept: There is a lot of marketing hype in the claims made by
> router companies.


iPhone XS Max

Wi-Fi 5 Example: Seen upper right are the specifications for an AC4000 (4000
Mbps) class router. But realistically, what speed can YOU expect from your
"iPhone XS Max", a 2×2 MIMO device, at a reasonable distance of 32 feet?

Bands/MIMO: AC4000 is 750+1625+1625. So what do those numbers mean? It is the
'maximum' speeds (best modulation possible with highest MIMO) of all 'bands' in
the router added together as follows:
 1. Band One: 750 is the maximum PHY speed for MIMO 3×3 in the 2.4 GHz band, but
    for an unrealistic 1024-QAM (see PHY table far below). A much more realistic
    PHY speed for a 2×2 MIMO wireless device is 300 Mbps.
 2. Band Two: 1625 is the maximum PHY speed for MIMO 3×3 in the 5 GHz band, but
    for an unrealistic 1024-QAM (see PHY table far below). A much more realistic
    PHY speed for a 2×2 MIMO wireless device is 650 Mbps (about 32 feet from the
    router).
 3. Band Three: same as band two.

MAC Overhead: Take the 5 GHz PHY speed (for one 5 GHz band, not both bands, so
650) and multiple by 70% to get an estimate of the Mbps speeds that you will see
within speed test applications running on your wireless device.

Conclusion: At 32 feet, you will get a maximum speed of around 455 Mbps (±45
Mbps) from your iPhone XS Max from this '4000 Mbps' router. With a second AC
band, you 'might' get up to 455 Mbps from another wireless device at the same
time. However, see also the tri-band router [§N] appendix below.

> So upgrading to a faster router will increase your iPhone XS Max speeds,
> right? No! What about an AC5400 4×4 tri-band router? Same speed. What about an
> ultra-fast Wi-Fi 6 AX6000 8×8 router, marketed as being 4x faster than Wi-Fi
> 5? Same speed. Understand router manufacturers' marketing hype.

A faster router only gets you half way there. But in order to get the high
advertised speeds from a router (for only one band; not the published aggregate
number), you need a 4×4 MIMO client wireless device that the industry does not
yet make. Virtually all wireless client devices today are still 2×2 MIMO -- so
the maximum speeds for 2×2 MIMO are what you should realistically expect no
matter how powerful the router. The only known exceptions to this general rule
are an older Dell laptop that did have 3×3 MIMO (but I can't find any new Dell
laptops that do now) and some MacBook Pros that have 3×3 MIMO. If you know of
any other exceptions, please let me know [§Y]


7. MIMO - a wireless revolution


4×4 MIMO illustration

MIMO: What is partly driving the dramatic increase in wireless (Wi-Fi, cellular,
etc) capacity in the last few years is MIMO, an acronym for Multiple Input,
Multiple Output (wikipedia.org), or spatial multiplexing, or spatial streams --
by using multiple antennas all operating on the same frequency at the same time.
Most smartphones today are capable of 4×4 cellular MIMO -- so they are
(potentially) four times as fast as a single antenna phone. But MIMO for Wi-Fi
is stuck at 2×2 MIMO for most wireless Wi-Fi (client) devices.

> Analogy: Think of MIMO as adding 'decks' to a multi-lane highway. More lanes
> (capacity) are added without using more land (spectrum). 2×2 MIMO is a highway
> with one more highway deck above it. And 4×4 MIMO is a highway with three more
> highway decks above it.

What is the big deal: The reason MIMO is such a huge deal is because it is a
direct capacity multiplier (×2, ×3, ×4, ×8, etc) while using the SAME (no more)
spectrum. This is accomplished by simply using more antennas (but both the
router and client must have the additional antennas).

> MIMO adds more capacity without using more spectrum! Regardless of the MCS
> level used (distance between router and router), 2×2 MIMO can transmit twice
> as much data as 1×1 MIMO can, and 4×4 MIMO can transmit four times as much
> data as 1×1 MIMO -- in the same amount of time. This is why MIMO is such a big
> deal -- a direct capacity multiplier.
> 
> The huge caveat, of course, is that BOTH the transmitter and receiver must
> support MIMO. And if each supports different levels of MIMO, the minimum MIMO
> level common to both devices will be used. For example, a 2×2 MIMO tablet
> connecting to an 8×8 MIMO router will only use 2×2 MIMO. But as a very
> significant bonus, the 'extra' antennas (if there is a mismatch in MIMO levels
> between the client and router) do not go unused, but are used for 'diversity'
> and 'beamforming', which extends range, and improves speed at range.
> 
> NOTE: The 'number of visible antennas' does NOT (necessarily) indicate MIMO
> level. It may, but it may not. Sometimes antennas are shared between Wi-Fi
> bands. Sometimes antennas are dedicated to just a single Wi-Fi band so 2× or
> 3× as many antennas as MIMO level. Some routers use a mix of both internal and
> external antennas. Some Wi-Fi 6E routers are moving to all internal antennas.

1×1 MIMO 2×2 MIMO 4×4 MIMO


Example: On a single 80 MHz 802.11ac Wi-Fi 5 channel operating at a PHY of 433
Mbps:
 * 1×1 MIMO yields a PHY of 433 Mbps
 * 2×2 MIMO yields a PHY of 866 Mbps (most wireless clients are 2×2)
 * 3×3 MIMO yields a PHY of 1300 Mbps
 * 4×4 MIMO yields a PHY of 1733 Mbps (most higher end routers are 4×4)
 * 8×8 MIMO yields a PHY of 3466 Mbps

all on the same 80 MHz channel.

Notation: You might see the MIMO level written as T×R:S, where 'T' is the number
of transmit antennas, 'R' is the number of receive antennas, and 'S' (an
optional component) is the number of simultaneous 'streams' supported. If the
'S' component is missing, it is assumed to be the minimum of 'T' and 'R'. OR,
some devices will just say '2 streams' (for 2×2:2) or 'quad stream' (for 4×4:4).

Diversity: If there are more antennas than streams (eg: 2×2 client to 4×4
router), the 'extra' antennas can then be used to improve link quality, and
increase range. With multiple antennas receiving the same transmitted signal,
the receiver can recombine all of the received signals into a better estimate of
the true transmitted signal.

> FCC documents discuss that the 'maximum' gain when doubling antennas is
> 10×log(NANT/NSS) dBi, which for a 2×2 client to a 4×4 access point, would
> result in a diversity gain of 'around' 3 dBi. OR a 4×4 access point to a 1×1
> client means a diversity gain 'around' 6 dBi.
> 
> This explain why you really do want a 4×4 MIMO router, even though there may
> only be 1×1 and 2×2 client devices connecting to it!

Comcast XB6 gateway - 8×8 MIMO

Beamforming: This Wi-Fi technology uses multiple antennas to 'focus' the
transmitted RF signals more towards a device (instead of just broadcasting the
signal equally in all directions). The end result is a slightly stronger signal
(in the direction of the device), which typically causes a slightly higher
modulation to be used, which in turn increases Mbps speed by a little bit.

> A very nice explanation of beamforming! (youtube.com)

> KEY Wi-Fi concept: It is easy to overlook and miss, but beamforming and
> diversity are the key reasons why you want a 4×4 MIMO router even though most
> clients are still only 2×2 MIMO. The extra antennas are actually used and
> offer significant value (a stronger signal, which translate to better connect
> speeds for far-away users)!

Client MIMO: Almost all battery powered wireless devices are stuck at 2×2 MIMO
for Wi-Fi, and this seems unlikely to change anytime soon. The extra power
requirements of 4×4 MIMO causing reduced run times is just not worth the
tradeoff (yet). But for devices with lots of power (like a PC on AC power), you
can buy 4×4 MIMO adapters.

Must a client device with MIMO always use MIMO? No, it does not have to. I have
a Dell laptop with an "Intel Wi-Fi 6E AX210" card that I have documented
flipping back and forth between 1×1 and 2×2 data rates (keeping the MCS level
the same) depending upon conditions. This needs more research (is this a bug or
a feature).

A final note: You will only get the dramatic speed benefits of MIMO if you have
a client device (phone, tablet, TV, computer, etc) that actually supports MIMO.
Most client devices today (September 2024) are STILL (at best) 2×2 MIMO. It is
very rare to see a (battery powered) client device that supports MIMO higher
than that.

> AX 'Stream' Deception: Router vendors' are now being incredibly deceptive when
> it comes to advertising in their new "AX" class of routers. Details [§Q].
> Netgear is using "spatial streams" to describe their new AX routers
> (netgear.com), but this is NOT the same thing as "spatial streams" in MIMO in
> the 802.11ax standard -- which is what most people will (wrongly) conclude --
> and that is outright deceptive, and Netgear knows it (because when I mentioned
> this in a Netgear forum post, a Netgear moderator deleted my post). Netgear
> claims their new RAX80 4×4 four-antenna (four spatial streams) router is "8
> streams" (netgear.com). So, do your research, and buyer beware.
> 
> Netgear's "spatial stream" logic is provably wrong. The maximum number of
> 'streams' in a router can not be larger than the number of antennas in the
> router. because "In the T×R configuration the maximum number of spatial
> streams is limited by the lesser of either T or R". Source (cisco.com).
> 
> A pattern emerges: Router vendors are incredibly 'creative' in their marketing
> of new routers. They are constantly figuring out creative ways to make new
> hardware sound 'so much better' than older similar hardware.

Learn More:
 * EXCELLENT: 802.11ax Frequently Asked Questions (aerohive.com via archive.org)
 * MIMO: Why multiple antennas matter (cisco.com)
 * FAQ: MIMO, MRC, Beamforming, STBC, and Spatial Multiplexing (huawei.com)
 * MIMO Communications (youtube.com)


8. Wi-fi 1/2/3 -- Legacy 802.11
Historical note: There were many wireless products that existed well before the
802.11 specification. For example, WaveLAN (wikipedia.org) in 1990, and others.
But the "802.11" specification created a single standard so that products from
different vendors could interoperate with each other. And thankfully, there was
a Wi-Fi Alliance ultimately formed to build consensus and move Wi-Fi forward as
an industry.

Reference: A brief look at past legacy Wi-Fi generations (and while not official
names, Wi-Fi 1, Wi-Fi 2, and Wi-Fi 3):

> GenSpecYearPHY SpeedsUnofficial nameBand First 802.11 1997 2 Mbps Wi-Fi 1 2.4
> GHz Second 802.11b 1999 11 Mbps Wi-Fi 2 2.4 GHz Third 802.11a 2001 54 Mbps
> Wi-Fi 3 5 GHz Third 802.11g 2002 54 Mbps Wi-Fi 3E 2.4 GHz

802.11 (Wi-Fi 1): PHY data rates of 1 or 2 Mbps with 11,000,000 symbols/sec,
using direct sequence spread spectrum (DSSS) with three non-overlapping 22 MHz
channels in 2.4 GHz (1, 6, 11).

802.11b (Wi-Fi 2): PHY data rates of 1, 2, 5.5, or 11 Mbps with 11,000,000
symbols/sec, using direct sequence spread spectrum (DSSS) with three
non-overlapping 22 MHz channels in 2.4 GHz (1, 6, 11). How rates are calculated
(blogspot.com).



PHY speed 2 Mbps
5.5×
11 Mbps


★ The PHY speed increases in Wi-Fi 2 (802.11b): Most Wi-Fi 2 client devices saw
a 5.5× maximum PHY speed change from 2 Mbps to 11 Mbps, from:
 * 450% - a bitrate encoding change increased speeds from 2 Mbps to 11 Mbps

Throughput: For an 11 Mbps PHY, expect: (1) throughput speed tests [§D] to peak
at 'around' 6.5 Mbps when very close to the router, and (2) speeds to decrease
quickly with distance and obstacles.


Symbols: A key concept in Wi-Fi: All Wi-Fi devices within the same generation of
Wi-Fi transmit data via the SAME 'symbols' -- or essentially, a very tiny slot
in time. And that is it! End of story. However, client devices very close to a
router can 'hear' those symbols very clearly, and client devices far away from
the router can barely 'hear' those symbols. So Wi-Fi leverages this fact and
always encodes the maximum amount of information that can be successfully
'heard' -- and this takes place for each client device, individually. The result
is the "Modulation + Coding" tables you see in this (and following) chapters.
The tables define how data will be encoded in Wi-Fi, and the MCS levels directly
correlate to how well symbols can be heard (distance from the router). More
Information [§F]

> KEY Wi-Fi concept: Wi-Fi works BEST (fastest) when you are close to an access
> point.


--------------------------------------------------------------------------------

The START of modern Wi-Fi...

--------------------------------------------------------------------------------


802.11a/g
PHY Speeds
20 MHz channel
800ns guard interval Modulation
+ CodingMbps 0BPSK 1/26.0 1 3/49.0 2QPSK 1/212.0 3 3/418.0 416‑QAM 1/224.0 5
3/436.0 664‑QAM 2/348.0 7 3/454.0


More PHY tables

802.11a (Wi-Fi 3): PHY data rates 6 Mbps to 54 Mbps (see table right) with
12,000,000 symbols/sec, using orthogonal frequency-division multiplexing (OFDM)
with 12 non-overlapping 20 MHz channels in 5 GHz (36, 40, 44, 48, 52, 56, 60,
64, 149, 153, 157, 161), but some channels (52-64) had DFS restrictions. Details
(motorolasolutions.com).

But 802.11a really never 'took off' (in the consumer market) since initial
802.11a devices worked only in the 5 GHz band (did NOT support existing 802.11b
clients in the 2.4 GHz band) and were expensive (as compared to 802.11b
products).

> The router industry learned a very hard lesson about the 'consumer market' --
> that any new router/AP must also be fully backward compatible (must support
> most, if not all, of the old client devices out there)!
> 
> New routers today support ALL prior generations of Wi-Fi back to 802.11b.
> 
> Note that 64-QAM 5/6 is not a part of Wi-Fi 3. That arrives in the next Wi-Fi
> generation, Wi-Fi 4.



Linksys WRT54G
(a big hit back in the day)

802.11g (Wi-Fi 3E): Wi-Fi 3 802.11a technology in 5 GHz was moved/extended back
into the 2.4 GHz band. PHY data rates 6 Mbps to 54 Mbps (see table upper right)
with 12,000,000 symbols/sec, using orthogonal frequency-division multiplexing
(OFDM) with three non-overlapping 20 MHz channels in 2.4 GHz (1, 6, 11) -- see
the next chapter [§9] for channel details. Wi-Fi 3 also could revert to 802.11b
mode to support older clients -- so 802.11g was highly successful. And it worked
incredibly well considering that typical residential broadband Internet download
speeds back then were around 3 Mbps. It is remarkable that today you can still
today buy a brand new Linksys WRT54GL router (802.11g).

> A note about channels: In the U.S. there are 11 overlapping Wi-Fi channels in
> 2.4 GHz. The only way to get non-overlapping channels is for all routers/AP to
> cooperate and set their channels to either 1, 6, or 11. But when I use a Wi-Fi
> analyzer, I see routers operating on other channels all of the time. Be a nice
> neighbor and only use channels 1, 6, or 11. See the drawing in the next
> chapter [§9] for more details.


PHY speed 11 Mbps
4.9×
54 Mbps


★ The PHY speed increases in Wi-Fi 3 (802.11g): Most Wi-Fi 3 client devices saw
a 4.9× maximum PHY speed change from 11 Mbps to 54 Mbps, from:
 * 391% - The introduction of OFDM (vs DSSS) increased speeds from 11 Mbps to 54
   Mbps

Throughput: For a 54 Mbps PHY, expect: (1) throughput speed tests [§D] to peak
at 'around' 30 Mbps when very close to the router, and (2) speeds to decrease
quickly with distance and obstacles.


Learn More:
 * Wi-Fi (802.11) PHY Data Rates (blogspot.com) -- Understand how PHY rates are
   computed
 * Different Wi-Fi Protocols and Data Rates (intel.com)


9. Wi-Fi 4 (802.11n) 2.4 GHz and 5 GHz (HT: High Throughput)
★ The big advance in Wi-Fi 4 was the introduction of MIMO [§7] (multiple
antennas), instantly doubling (for 2 antennas) or tripling (for 3 antennas)
throughput over the prior Wi-Fi 3 (but both client/AP must implement MIMO).





802.11n in 2.4 GHz is a legacy wireless band that has been replaced by much
faster Wi-Fi in the 5 GHz and 6 GHz bands. This chapter is provided for
reference only. You should be using 'newer' Wi-Fi for all of your 'new' wireless
Internet devices. Only use 2.4 GHz when you are forced to -- by a device that
does not support newer versions of Wi-Fi (like many low bandwidth IoT devices
only support Wi-Fi 4).


> GenSpecYearPHY SpeedsNew Name Fourth802.11n 200672 to 217 Mbps Wi-Fi 4

2.4 GHz Wi-Fi channels ChannelMHz center20 MHz channel 124122402-2422
224172407-2427 324222412-2432 424272417-2437 524322422-2442 624372427-2447
724422432-2452 824472437-2457 microwave ovens: 2450 MHz ±50 MHz 924522442-2462
1024572447-2467 1124622452-2472 12not available
in the U.S. 13 14

217 Mbps speed: The 217 Mbps maximum PHY speed is for a 20 MHz channel to a 3×3
MIMO client. However, a much more realistic maximum PHY speed is 144 Mbps for a
20 MHz channel to a 2×2 client.

> 802.11n is called "HT" for High Throughput

Spectrum: There is ONLY 70 MHz of spectrum (2402-2472 MHz) available for Wi-Fi
to use in the U.S. in the 2.4 GHz band, supporting only three non-overlapping
20MHz channels.

> Let that sink in. Only 70 MHz of spectrum in 2.4 GHz must be shared between
> you, your family, and a bunch of neighbors.

There are eleven OVERLAPPING 2.4 GHz Wi-Fi channels: In the US, Wi-Fi routers
allow you to set the 2.4 GHz Wi-Fi channel anywhere from 1 to 11. More
information (wikipedia.org). So there are 11 Wi-Fi channels, right? NO! These
eleven channels are only 5MHz apart -- and it actually takes a contiguous 20MHz
(and a little 5 MHz buffer between channels) to make one 20MHz Wi-Fi channel
that can actually be used. Because of this, in the US, these restrictions result
in only three usable non-overlapping 20MHz Wi-Fi channels available for use (1,
6, or 11; seen right).

> Technically, channels 12 and 13 could be used in the US, but under significant
> restrictions that in reality, cause all routers/clients to elect to not touch
> those channels. Details (wikipedia.org)

The THREE non-overlapping channels: You CAN set the Wi-Fi channel to any channel
and it will work. However, if you don't select 1, 6, or 11, the 20 MHz channel
you create will almost certainly impact TWO other 20 MHz (neighbor) channels
operating on 1, 6, 11. And more importantly, maybe TWO neighbor channels will
impact your one channel. Not good. If your AP/router uses channel 2, 3, 4, 5, or
7, 8, 9, 10, that is an error to fix! So be a nice neighbor and only use one of
the three non-overlapping channels: 1, 6, or 11!




2.4 GHz Wi-Fi has only THREE non-overlapping channels

> More Info: Video about channels 1/6/11 from MetaGeek (youtube.com)
> 
> An analogy: A router using channel 2, 3, 4, 5, or 7, 8, 9, 10 (instead of
> channel 1, 6, or 11) is exactly like a car on a three lane Intestate driving
> directly over the dashed lane divider lines and refusing to move over fully
> into a lane. That 'car' gets to its destination, but in the process, then
> impacts TWO lanes on the Interstate, instead of just one lane.

Screenshot from 'RF-Explorer Handheld Spectrum Analyzer'

A small (intentional) gap between channels: Notice the very small 5 MHz gap
between channels 1, 6, and 11. This is very intentional and designed so that
(hopefully) traffic on one channel does not interfere with traffic on an
adjacent channel.

The gap is needed because there IS a small amount of 'bleed-over' of signal past
the limits of the channel width and this tiny gap between adjacent channels
helps to greatly minimize the impact of that bleed over. See Spectral Masks
[§S]. This subtle 'bleed over' can be clearly seen in a capture from a spectrum
analyzer (seen right) of Wi-Fi devices operating on channel 1 (left hump) and 6
(right hump). Notice that (a) the edges of each channel are not a vertical
drop-off, but rather slant down, and that (b) the valley between humps does NOT
go all the way down to the noise floor (the 'floor' is seen as small vertical
lines on the X-axis in the graph).

> But what about the 'four-channel plan' overseas using channel 1, 5, 9, 13?
> That can only work well when adjacent channels are not actively being used!.
> There is now NO 5 MHz space between channels, so the bleed over [§S] from the
> sideband signal in adjacent channels will be even more pronounced -- and be
> even more impactful. When Wi-Fi is heavily used, it is best to stick with only
> using channels 1, 6, 11.

Symbols/sec: 802.11n uses either 14,444,444 symbols/sec (400ns short GI), or
13,000,000 symbols/sec (800ns long GI), for a 20 MHz channel.

Shared spectrum: All Wi-Fi devices on the same spectrum must SHARE that
spectrum. Ideally, all Wi-Fi devices decide to operate on either channel 1, 6,
or 11 -- the only true non-overlapping channels. Then all devices operating on a
channel share that channel. But I have seen routers operate on channel 8, which
means that router is being a 'bad neighbor' and interfering with 20 MHz channels
operating on 6 and 11.

Protocol Overhead: Each 20MHz Wi-Fi channel has maximum PHY bitrate of around
72Mbps, but due to Wi-Fi protocol overhead, you may only get to use around 60%
to 80% of that.

> In a very 'clean' Wi-Fi environment, I have seen throughput [§D] around 54.2
> Mbps for a PHY speed of 72.2 Mbps, which comes out to 75% efficiency -- pretty
> good. Another time, when I was just feet from the router, I measured a peak
> throughput of around 118 Mbps for a PHY speed of 144.4 Mbps (82% efficiency)
> -- very good.

802.11n PHY Speeds (Mbps)
20 MHz channel, 400ns guard interval Modulation
+ CodingMIMO 1×12×23×34×4 0BPSK 1/2 7.2 14.4 21.6 28.8 1QPSK 1/214.4 28.8 43.3
57.7 2 3/421.6 43.3 65.0 86.6 316‑QAM 1/228.8 57.7 86.6115.5 4 3/443.3
86.6130.0173.3 564‑QAM 2/357.7115.6173.3231.1 6 3/465.0130.0195.0260.0 7
5/672.2144.4216.6288.8 -256‑QAM 3/486.6173.3260.0346.6 - 5/696.2192.5288.8385.1
-1024‑QAM3/4108.3216.6325.0433.3 - 5/6120.3240.7361.1481.4

256-QAM and 1024-QAM are non-standard




802.11n PHY Speeds (Mbps)
40 MHz channel, 400ns guard interval Modulation
+ CodingMIMO 1×12×23×34×4 0BPSK 1/215 30 45 60 1QPSK 1/230 60 90120 2 3/445
90135180 316‑QAM 1/260120180240 4 3/490180270360 564‑QAM 2/3120240360480 6
3/4135270405540 7 5/6150300450600 -256‑QAM 3/4180360540720 - 5/6200400600800
-1024‑QAM3/4225450675900 - 5/62505007501000

256-QAM and 1024-QAM are non-standard
More PHY speed tables



Understanding channel widths: The standard Wi-Fi channel width is 20 MHz. So a
40 MHz channel is TWO 20 MHz channels put together (2× capacity).

> Analogy: Think of channel width as how many 'lanes' you can use at once on a
> multi-lane highway. 20 MHz is a car using a single lane. 40 MHz is a 'wide'
> load trailer using two full highway lanes.

Channel bonding / 40MHz channels: This is the biggest marketing rip-off ever (in
2.4 GHz). Routers can then advertise 2x higher speeds, even though in virtually
all circumstances, you will only get 1/2 of the advertised speed (only be able
to use a 20 MHz channel)! For example, The Netgear N150 (implying 150Mbps),
which is the result of taking TWO 20MHz Wi-Fi channels and combining them into
one larger 40MHz channel, doubling the bitrate. This actually does work, and
works well BUT ONLY in 'clean room' testing environments (with NO other Wi-Fi
signals). However, for Wi-Fi certification, the required 'good neighbor'
implementation policy prevents these wider channels from being used in the real
world when essentially the secondary channel would interfere with neighbors'
Wi-Fi -- which unless you live in outer Siberia, you WILL 'see' neighbors' Wi-Fi
signals and the router will be required to automatically disable channel
bonding.

> Great article on the subject: "Bye Bye 40 MHz Mode in 2.4 GHz"
> (smallnetbuilder.com)
> 
> I am curious if this issue had anything to do with why Netgear stopped getting
> their routers 'Wi-Fi Certified'?
> 
> Or, if there is a single 20 MHz only client that connects to the AP, the AP
> will (should) drop from 40 MHz operation to 20 MHz operation, disabling
> channel bonding. This situation is actually VERY likely to happen (for
> example, my daughter's very inexpensive laptop that is only two years old, but
> only supports 20 MHz channels).
> 
> Virtually all (inexpensive) IoT devices only support 20 MHz channels.
> 
> Also, in the real world, things are MUCH more complicated, because many
> routers don't always follow 'good neighbor' standards try to force and use 40
> MHz channels when they should only be using 20 MHz channels.
> 
> Of note is that 40 MHz channels in the 5 GHz band for 802.11n does work --
> very well.

256-QAM and 1024-QAM HYPE: These are non-standard extensions to 802.11n, so most
client devices will never be able to get these speeds. And even if you have a
device that is capable of these speeds, are you close enough to the router to
get these speeds? Understand that advertised speeds in these ranges are mostly
marketing hype. See Broadcom TurboQAM (archive.org) and NitroQAM (archive.org).

> The reason why 256-QAM and 1024-QAM are included in the PHY tables here is for
> reference/convenience -- because these PHY tables ARE ALSO the PHY speed
> tables for 802.11ac for 20 MHz and 40 MHz channel widths. The PHY speeds for
> an 80 MHz channel is far below in the next chapter.

Interference: The entire 2.4 GHz Wi-Fi band is plagued by interference (a victim
of the success of that spectrum), or other devices using the SAME frequency
range. For example, cordless phones, baby monitors, Bluetooth, microwave ovens,
etc. Microwave ovens operate at 2450 MHz ± 50 MHz. Source (archive.org), which
is the entire Wi-Fi space, and very likely impacting two of the Wi-Fi channels,
and in some cases, even all three Wi-Fi channels (zdnet.com).

> Major interference (in red) caused by a Microwave Oven (for a couple seconds)
> 
> Microwave ovens are licensed in the entire ISM (Industrial, Scientific and
> Medical) band from 2.4 GHz to 2.5 GHz, which covers all 2.4 GHz Wi-Fi
> channels. Example (fccid.io).
> 
> How bad the interference is totally depends upon the specific microwave. Some
> microwaves are very bad, while others seem to have very little impact. At one
> house, using the microwave oven causes Wi-Fi clients to disconnect from the
> AP, while in another house, using the microwave oven only causes a slight
> slowdown in bandwidth to Wi-Fi clients.
> 
> Years ago I was testing a Wi-Fi security camera (base station and cam
> connected via 2.4 GHz), and just happened to use the microwave oven, and
> noticed the cam was unable to record any video. I learned my lesson and
> immediately returned the camera.
> 
> Is your house near a busy road? If so, you are likely getting some
> interference from all the cars driving by that are operating a 'hotspot'
> (likely always enabled, although maybe not with Internet activated). And the
> worst part is, you can NOT plan for that channel usage, because the cars are
> mobile!

Proprietary beamforming: Some 802.11n devices did support 'beamforming', but
these were proprietary extensions that required matching routers and clients
(one vendor's implementation would not interoperate with a second vendor's
implementation).

The BOTTOM LINE: The 2.4 GHz band is just WAY too crowded. Today, it is a victim
of its own success. Use a modern dual-band (2.4 and 5 GHz) router/AP and switch
over to the 5 GHz band -- for all devices that support 5 GHz. All quality
devices made in the last few years (phones, tablets, laptop computers, TVs, etc)
will absolutely support 5 GHz for Wi-Fi.

> The real impact of overcrowding: In a beach resort community, with homes very
> close to each other, a Wi-Fi analyzer app shows well over fifteen 2.4 GHz
> networks within range. At night, Wi-Fi performance (actual throughput) on the
> 2.4 GHz band was horrible (even with great PHY speed) due to contention
> (sharing bandwidth) with many neighbors. However, performance on the 5 GHz
> band was excellent.

A final warning: I am glossing over the fact that 802.11n can also operate in
the 5 GHz band, using 20 MHz and 40 MHz channels (but not 80 MHz channels and
not 256-QAM), because 802.11ac is so common place today. Just be aware that
802.11n using 5 GHz is possible using 'dual-band 802.11n' Wi-Fi devices -- don't
assume a Wi-Fi device operating in 5 GHz is 802.11ac (it may only be 802.11n).
There are still brand new dual-band 802.11n routers and devices (smartphones,
doorbell cameras, etc) being sold today that are 802.11n dual-band (and not
802.11ac)!





So why is 2.4 GHz Wi-Fi 4 not considered 'legacy' Wi-Fi: Frankly, Wi-Fi 4 is
'too old' and should be considered legacy (and not used much anymore)! But the
surprising fact is that many brand-new IoT devices (especially 'battery'
devices) being sold today come with only 2.4 GHz Wi-Fi 4 support.

> Example: In mid-2023, Ring introduced a new $60 Ring Indoor camera
> (amazon.com) that supports ONLY 2.4 GHz (and no MIMO). And since it is plugged
> in all the time to an outlet, that seems very short-sighted (especially when I
> can find smart plugs for $8 on Amazon that support 5 GHz Wi-Fi). If the 2.4
> GHz Wi-Fi band still works OK for you, then no problem. But if 2.4 GHz is too
> congested at your physical location, then this camera will not work well for
> you. Buyer beware.

PHY speed 54 Mbps
2.7×
144.4 Mbps


★ The PHY speed increases in Wi-Fi 4 (802.11n): Most Wi-Fi 4 2×2 MIMO client
devices saw a 2.7× maximum PHY speed change from 54 Mbps to 144.44 Mbps, from:
 * 8.3% - 54 Mbps becomes 58.5 Mbps by using 52 subcarriers out of 64 (instead
   of just 48)
 * 11.1% - which then becomes 65 Mbps by reducing the guard interval (GI) from
   800ns to 400ns
 * 11.1% - which then becomes 72.2 Mbps via a new QAM (64-QAM 5/6) modulation
 * 100% - which then becomes 144.44 Mbps (100%) with 2×2 MIMO [§7]
 * not counting 40 MHz channels in 2.4 GHz because in reality it can rarely be
   used
 * not counting MIMO past 2×2 because that is not seen in battery-powered client
   devices

HT20 Throughput: For a 144.4 Mbps PHY (2×2 MIMO client; 20 MHz channel), expect:
(1) throughput speed tests [§D] to peak at 'around' 115 Mbps when very close to
the router, and (2) speeds to decrease quickly with distance and obstacles.


Learn More:
 * Understanding High Speed 802.11n Wireless Networks in Depth (cisco.com)


10. Wi-Fi 5 (802.11ac) 5 GHz (VHT: Very High Throughput)
★ The big advance in Wi-Fi 5 was moving into the 5 GHz band and the introduction
of 80 MHz channels, instantly quadrupling throughput over the prior Wi-Fi 4 with
20 MHz channels (but both client/AP must implement 80 MHz channels).



Wi-Fi's current 'state-of-the-art' is Wi-Fi 6 and 6E (next chapters). Most 'new'
devices today support at least Wi-Fi 6 and maybe even Wi-Fi 6E (especially
higher end devices/phones).




Wi-Fi 5: The fifth generation of Wi-Fi is 802.11ac (2013) on 5 GHz. It provides
a maximum PHY speed of 3.4 Gbps on an 80 MHz channel using 8×8 MIMO (and fully
backward compatible with prior Wi-Fi generations). However, a much more
realistic maximum PHY speed is 1.7 Gbps on an 80 MHz channel using 4×4 MIMO.
Wi-Fi 5 has now been 'officially' replaced by 802.11ax Wi-Fi 6 (see next
chapter).

> GenSpecYearPHY SpeedsNew Name Fifth802.11ac2013433 to 1733 MbpsWi-Fi 5

5 GHz Wi-Fi channels (U.S.) Channel #20 MHz
center20 MHz
channel 160804020 5042383651805170-5190 4052005190-5210 464452205210-5230
4852405230-5250 58545252605250-5270 5652805270-5290 626053005290-5310
6453205310-5330 GAP (160 MHz) 11410610210055005490-5510 10455205510-5530
11010855405530-5550 11255605550-5570 12211811655805570-5590 12056005590-5610
12612456205610-5630 12856405630-5650  13813413256605650-5670 13656805670-5690
14214057005690-5710 14457205710-5730 GAP (5 MHz) 16315515114957455735-5755
15357655755-5775 15915757855775-5795 16158055795-5815 17116716558255815-5835
16958455835-5855 17517358655855-5875 17758855875-5895

More info from Wikipedia

1733 Mbps speed: The 1733 Mbps maximum PHY speed is for an 80 MHz channel to an
4×4 client. You can find 4×4 Wi-Fi cards for your PC. However, a much more
realistic maximum PHY speed (for 'on battery' devices) is 866 Mbps for an 80 MHz
channel to a 2×2 client, and in the real-world, a PHY speed of 780 Mbps is
reasonable.

> 802.11ac is called "VHT" for Very High Throughput

Spectrum: There is 560 MHz of spectrum (5170-5330, 5490-5730, 5735-5895 MHz)
available for Wi-Fi to use in the U.S., supporting seven non-overlapping 80 MHz
channels. If a device is labeled as supporting 802.11ac, you KNOW it also
supports 80 MHz channels.

> BEWARE: Many entry-level low-end routers only support 180 MHz of the 5 GHz
> spectrum (not all 560 MHz).
> 
> NOTE: One 80 MHz channel in 5 GHz has more spectrum than ALL 2.4 GHz channels,
> combined!

Channels: Channels in 5 GHz are messy and complicated. The 5 GHz Wi-Fi band has
seven 80 MHz channels (see table right, bolded numbers; 42, 58, 106, 122, 138,
155, 171) BUT ONLY if you have an AP that supports ALL of the DFS channels [§14]
(the channels in red).

> Channel Use Restriction: 16 (seen in red, right) of the 25 channels (or 64%)
> come with a critical FCC restriction (DFS - dynamic frequency selection) to
> avoid interference with existing devices operating in that band (weather-radar
> and military applications). Very few 'consumer-grade' access points support
> ALL of these 'restricted' channels, whereas many 'enterprise-grade' access
> points DO support these channels. More on this later in this chapter. 802.11h
> defines (1) dynamic frequency selection (DFS) and (2) transmit power control
> (TPC).
> 
> Channel 144: This channel was added as part of FCC changes in 2014. So this
> channel will be problematic for older devices that don't recognize this
> channel. Worst of all is that some brand new devices also mess up and don't
> support channel 144, so it is best to avoid selecting 144 as a 'primary'
> channel in most routers -- because if you do, a small subset of clients will
> not be able to connect to your router. Devices that don't recognize 20 MHz
> channel 144, also by definition don't recognize 40 MHz channel 142 and 80 MHz
> channel 138 (so a client device may have limited channel width when connecting
> to an AP using primary channels 132, 136, or 140).
> 
> > For example, Ring video doorbell cams that operate in 5 GHz don't understand
> > that channel 144 exists. The cam will NOT connect to an AP on channel 144,
> > and will only connect to an AP on channel 140 using a 20 MHz channel (not 40
> > MHz).
> 
> 120/124/128: Terminal Doppler Weather Radar (TDWR): If you are 'near' a major
> metropolitan airport, you might not be able to use 20 MHz channels 120, 124,
> or 128 (and hence 80 MHz channel 122) due to use of Terminal Doppler Weather
> Radar operating within 5600-5650 MHz at a peak power of 250,000 watts. TDWR
> locations and frequencies (wispa.org). See also Terminal Doppler Weather Radar
> (wikipedia.org) and a Weather Radio Channels (cisco.com) on the TDWR issue.
> Channels affected are in dark red (above right).
> 
> 
> 
> New channels: In mid December 2019, the FCC voted (fcc.gov) to move forward on
> allocating an additional 45 MHz to the end of U-NII-3 in 5.9 GHz to Wi-Fi
> (seen in yellow in table right). This results in three NEW 20 MHz channels
> (169, 173, 177). See also FCC 19-129 (govinfo.gov). This also creates two
> additional 40 MHz channels (167, 175), one new 80 MHz channel (171), and one
> new non-DFS 160 MHz channel (163).
> 
> But, the big problem is that these new channels are for INDOOR USE ONLY. See
> FCC guidelines (fcc.gov). Also, given that Wi-Fi 6E now exists (with 1200 MHz
> of new spectrum), the value of these new channels in 5 GHz is GREATLY
> diminished -- because existing Wi-Fi 5 devices don't know about these new
> channels (and will fail to connect to a router using them) and new devices
> capable of recognizing the new channels will just use the Wi-Fi 6E channels
> instead.
> 
> 
> Center Frequency TIP: The 'center frequency' (in MHz) of any 5 GHz channel
> number is simply that channel number multiplied by five (and added to 5000).
> For example, the center frequency of channel 60 is 5000 + 60×5 = 5300 MHz. Or,
> reverse to compute channel number from center frequency.

Understanding 160/80/40/20 MHz channel selection: Your router will NOT present a
list of the 160/80/40 MHz channels to you (eg: 42, 155). Instead, your router
presents a list of ALL 20 MHz channels supported, and you select one channel as
the 'primary' channel (and 20 MHz channel support). Then to support 160/80/40
MHz channel clients, the router just automatically selects the appropriate
160/80/40 MHz channels as per the table seen upper right.

> Channel 165: ONLY select channel 165 when the router is configured for 20 MHz
> channel widths. Because if you select channel 165 when the router is
> configured to use 160/80/40 MHz channel widths, there are actually NO
> available 160/80/40 MHz channels -- NONE! Wi-Fi clients will ONLY be able to
> connect to 20 MHz channel 165! This behavior was first noticed on a Netgear
> R7800 router.
> 
> Other routers are smart enough to not show channel 165 (unless the router is
> configured to only use 20 MHz channels).
> 
> This 'automatic' selection of the appropriate 160/80/40 channel from a single
> 20 MHz channel that you select totally sidesteps the problem of one
> (misaligned) wide channel straddling two other wide channels.

Symbols/sec: 802.11ac uses either 14,444,444 symbols/sec (400ns short GI), or
13,000,000 symbols/sec (800ns long GI), for a 20 MHz channel -- this is the same
as the prior 802.11n.





Range: It is true that the range/distance of 5 GHz is reduced (in 'on paper'
calculations) as compared to 2.4 GHz (around 6 dB difference at same distance),
but counterintuitively, that can be a significant benefit when it comes to
actual throughput. The problem with 2.4 GHz is too much range (interference) --
I always see the SSID of lots of neighbors (red highlight right), and that is a
very bad thing because it means that I am sharing spectrum and bandwidth with my
neighbors (or if not outright sharing a channel, increasing the 'noise floor' so
your throughput suffers). With 5 GHz the number of neighbor's networks I can see
is dramatically reduced (green highlight right). Then, 5 GHz uses a much wider
channel width (80MHz vs 20MHz) and with a "wave 2" 4×4 MIMO access point with
beamforming, you will see actual useable bandwidth greatly increased.

> With 5 GHz, neighbors can (often times) be on the same channel and typically
> not interfere with each other (nearly as much as 2.4 GHz), because with
> reduced range, neighbors can't see as many neighbors Wi-Fi anymore. Of course,
> all of this depends upon how 'close' your neighbors are.

Protocol Overhead: The Mbps seen at the application level will be around 60% to
80% of the Mbps at the Wi-Fi (PHY) level. This is just due to Wi-Fi protocol
overhead (see chapter on PHY client speed [§4] far above).

FCC Channel Plan: Here is the 5 GHz 802.11 Channel Plan (fcc.gov) from the FCC
itself, also seen below. Of note is that on April 1, 2014 the FCC changed the
rules (fcc.gov) for usage in the 5 GHz band, to increase availability of
spectrum for Wi-Fi use. Summary of the new rules (fcc.gov). Channel 144 was
added (but older 5GHz clients will not be aware of this), power levels for
channels 52 to 64 were increased, and other miscellaneous changes.




Helpful FCC reference documents: Here are some helpful documents RE spectrum
usage:
 * 802.11 Channel Plans New Rules v02 (fcc.gov) - This is just the chart above
 * Code of Federal Regulations (ecfr.gov) - The actual Federal Code governing
   the Wi-Fi spectrum
 * Straddle channels (fcc.gov) - treatment of channels that straddle U-NII bands

Transmit Power: Channels 149-165 allow for both router/client to transmit at
1000 mW. Channels 36-48 allow for the router to transmit at 1000 mW (and clients
at 250 mW). For all other DFS channels, both the router/client can transmit at
250 mW. However, this does NOT necessarily mean that channels 149-165 are the
best channels to use (because everyone wants to use them). The 'reduced signal
strength' for the other channels can actually be a huge advantage, because it
means there is a much higher likelihood that you will NOT see neighbors Wi-Fi
channels (as frequently as 2.4 GHz channels), which translates directly to less
interference (the channel is all yours) and higher Wi-Fi speeds.

> Many residential routers have a transmit power around 995 mW. Many (battery
> powered) Wi-Fi clients have a transmit power anywhere from 90 mW to 250 mW.
> Client devices often transmit at power levels below the maximum power level
> permitted. More information [§H]

802.11ac PHY Speeds (Mbps)
80 MHz channel, 400ns guard interval Modulation
+ CodingMIMO 1×12×23×34×48×8 0BPSK 1/2 32 65 97 130 260 1QPSK 1/2 65 130 195 260
520 2 3/4 97 195 292 390 780 316‑QAM 1/2130 260 390 5201040 4 3/4195 390 585
7801560 564‑QAM 2/3260 520 78010402080 6 3/4292 585 87711702340 7 5/6325 650
97513002600 ↕ typical real-world Modulation/Coding at distance ↕ 8256‑QAM 3/4390
780117015603120 9 5/6433 866130017333466 -1024‑QAM3/4487 975146219503900 -
5/65411083162521664333

1024-QAM is non-standard



802.11ac PHY Speeds (Mbps)
80 MHz channel, 800ns guard interval Modulation
+ CodingMIMO 1×12×23×34×48×8 0BPSK 1/2 29 58 87 117 234 1QPSK 1/2 58 117 175 234
468 2 3/4 87 175 263 351 702 316‑QAM 1/2117 234 351 468 936 4 3/4175 351 526
7021404 564‑QAM 2/3234 468 702 9361872 6 3/4263 526 78910532106 7 5/6292 585
87711702340 ↕ typical real-world Modulation/Coding at distance ↕ 8256‑QAM 3/4351
702105314042808 9 5/6390 780117015603120 -1024‑QAM3/4438 877131617553510 -
5/6487 975146219503900

1024-QAM is non-standard
More PHY speed tables



Another big thing is beamforming / more antennas: After playing around with a
new 4×4 "wave 2" router (as compared to a 2×2 "wave 1" router), wow! A very
noticeable increase in speeds at range. 802.11ac beamforming really works.

> Your mileage will vary depending upon construction materials. In one home
> (single level; sheetrock with aluminum studs), I saw a dramatic increase in
> speeds at range. But at an older second home with very thick brick walls,
> range improved just a little.

256-QAM: This modulation requires a very good SNR (signal to noise ratio), that
is very hard to get with entry level routers. With a consumer-grade 802.11ac 2×2
"wave 1" AP I never got 256-QAM, even feet from the router. However, with a much
higher quality 802.11ac 4×4 "wave 2" AP, I now regularly see 256-QAM 3/4 being
used (at 25ft, through two walls).

1024-QAM HYPE: This modulation is a non-standard extension to 802.11ac, so most
client devices will often never be able to get these speeds. And even if you
have a device that is capable of these speeds, are you close enough to the
router to get these speeds? Understand that advertised speeds in these ranges
are marketing hype. See Broadcom NitroQAM (archive.org).

802.11ac Wave 2: The next generation (wave 2) (wi-fi.org) of 802.11ac is already
here. With feature like: (1) four or more spatial streams, (2) DFS 5 GHz channel
support, (3) 160 MHz channels, and (4) MU-MIMO. Cisco Wave 2 FAQ (cisco.com).

> Buyer beware: Not all 'wave 2' products will support the restrictive 5 GHz DFS
> channels! Wi-Fi certification for 'wave 2' only 'encourages' (wi-fi.org)
> devices to support this -- so NOT required.
> 
> 160 MHz channels: Support for 160 MHz channels in some routers reduces MIMO
> support. For example, in Netgear's R7800, there is 4×4 MIMO support for 80 MHz
> channels, but for 160 MHz channels, MIMO is reduced to 2×2.
> 
> MU-MIMO issues: There are a lot of issues with MU-MIMO. So it may or may not
> work for you. MU-MIMO (1) sometimes disables client MIMO (smallnetbuilder.com)
> (where a 2×2 client switches to 1×1; Broadcom chipset) (2) requires spatial
> diversity (physical distance) (archive.org) between clients (3) has
> significant sounding overhead (youtube.com) (4) a client device must be
> MU-MIMO aware (many are not) (5) only works with high SNR (very strong
> signals) and (6) works best with completely stationary clients. For more
> details, read "A MU-MIMO Reality Check" (networkcomputing.com). Aruba Networks
> says "Experience from 802.11ac MU-MIMO in real-world deployments revealed some
> limitations". Source (arubanetworks.com). More info (youtube.com).

Interference: It is a lot less common to find devices that use the 5 GHz band
(vs the 2.4 GHz band), causing interference for Wi-Fi, but it is still possible.
Just Google 'Panasonic 5.8 GHz cordless phone' for a cordless phone that uses
the upper 5 GHz channels 153 - 165. FCC info on Panasonic phone (fccid.io).

Minimum Sensitivity (dBM) for each MCS: Here is a graph of information that
comes from the IEEE spec. Note that each time you double channel width, that
there is a 3 dB 'penalty' [§J]:


A final warning and caveat regarding 802.11n in 5 GHz: I have glossed over the
fact that 802.11n can operate in the 5 GHz band, so DO NOT ASSUME that just
because a device operates in 5 GHz that the device must be 802.11ac. That is NOT
necessarily true. For example, the Motorola E5 Play (very low end) smartphone
does NOT support 802.11ac, but does support dual-band 802.11n, so it connects to
the 5 GHz band, but only using 20/40 MHz channels (in 1×1 mode), not the 80 MHz
channels of 802.11ac, and not using 256-QAM.

> Another example: An older Dell laptop using Centrino Advanced-N 6230 dual-band
> (intel.com) Wi-Fi. The laptop 'sees' the 5 GHz SSID being broadcast from a
> 802.11ac router, but when the laptop connects to the router, it is only doing
> so using 802.11n, 2×2 MIMO, and 40 MHz channels (max PHY of 300 Mbps; no
> 256-QAM)

PHY speed 144.4 Mbps
6.0×
866.6 Mbps


★ The PHY speed increases in Wi-Fi 5 (802.11ac): Most Wi-Fi 5 2×2 MIMO client
devices saw a 6.0× maximum PHY speed change from 144.4 Mbps to 866.6 Mbps, from:
 * 350% - 144.44 Mbps becomes 650 Mbps by increasing channel widths from 20 MHz
   to 80 MHz
 * 33.3% - which then becomes 866.6 Mbps via a new 256-QAM modulation
 * not counting 160 MHz channels because support was rare in most (battery)
   client devices.

VHT80 Throughput: For an 866.6 Mbps PHY (2×2 MIMO client; 80 MHz channel),
expect: (1) throughput speed tests [§D] to peak at 'around' 750 Mbps when very
close to the router, and (2) speeds to decrease quickly with distance and
obstacles.

> VHT160: For a 1733.3 Mbps PHY (2×2 MIMO client; 160 MHz channel), expect: (1)
> throughput speed tests [§D] to peak at 'around' 1300 Mbps when very close to
> the router, and (2) speeds to decrease quickly with distance and obstacles.


Learn More:
 * O'Reilly '802.11ac: A Survival Guide' by Matthew S. Gast (cmu.edu) -- 154
   page book
 * 802.11ac In-Depth (arubanetworks.com)
 * Technical White Paper: "802.11ac: The Fifth Generation of Wi-Fi" (cisco.com)
 * 802.11ac (wikipedia.org)
 * Clients List (mikealbano.com) - lists common Wi-Fi clients and the channels
   they support
 * United States Frequency Allocations Chart (ntia.doc.gov) - Who is using what
   parts of the radio spectrum
 * Eero 160MHz over 5GHz FAQ regarding Apple devices
 * 802.11ac and Wi-Fi Radio Fundamentals (youtube.com)


11. Wi-Fi 6 (802.11ax) 2.4 GHz and 5 GHz (HE: High Efficiency)
★ The big advance in Wi-Fi 6 was (1) efficiently transmitting to a large number
of users at the same time (but only for new Wi-Fi 6 clients, not prior Wi-Fi
generation clients) and (2) 1024-QAM modulation.

> Note that 160 MHz channel support in battery powered Wi-Fi 6 client devices is
> MUCH more common than in Wi-Fi 5 devices -- So while 160 MHz channels are NOT
> a new feature of Wi-Fi 6, as a practical matter, Wi-Fi 6 may be the first
> version of Wi-Fi that regularly delivers on 160 MHz channels for client
> devices, and allows 1 Gbps throughput to be regularly achieved.

"The bottom line is until Wi-Fi 6 / 802.11ax clients reach critical mass, the
benefits of 11ax are minimal and will have low impact." Source (cisco.com). The
key reason why: Wi-Fi 6 was designed from the ground up to provide speed
improvements (HE: High Efficiency) to a group of Wi-Fi 6 clients as a whole, NOT
an individual Wi-Fi 6 client!





Wi-Fi 6: The sixth generation of Wi-Fi is 802.11ax (2019). It provides a maximum
PHY speed of 9.6 Gbps on an 160 MHz channel using 8×8 MIMO. The 802.11ax
modulation (OFDMA) is NOT backward compatible with any prior version of Wi-Fi --
so you need Wi-Fi 6 clients to take advantage of Wi-Fi 6 router features.
However, any Wi-Fi 6 router will be able to revert back to Wi-Fi 4/5 to support
your older devices (with NO speed advantage over Wi-Fi 5).

> GenSpecYearPHY SpeedsNew Name Sixth802.11ax2019600 to 4804 MbpsWi-Fi 6

802.11ax PHY Speeds (Mbps)
80 MHz channel, 800ns guard interval Modulation
+ CodingMIMO 1×12×23×34×48×8 0BPSK 1/2 36 72 108 144 288 1QPSK 1/2 72 144 216
288 576 2 3/4108 216 324 432 864 316‑QAM 1/2144 288 432 5761152 4 3/4216 432 648
8641729 564‑QAM 2/3288 576 86411522305 6 3/4324 648 97212972594 7 5/6360
720108014412882 ↕ typical real-world Modulation/Coding at distance ↕ 8256‑QAM
3/4432 864129717293458 9 5/6480 960144119213843 101024‑QAM
3/45401080162121614323 11 5/66001200180124014803



802.11ax PHY Speeds (Mbps)
80 MHz channel, 1600ns guard interval Modulation
+ CodingMIMO 1×12×23×34×48×8 0BPSK 1/2 34 68 102 136 272 1QPSK 1/2 68 136 204
272 544 2 3/4102 204 306 408 816 316‑QAM 1/2136 272 408 5441088 4 3/4204 408 612
8161633 564‑QAM 2/3272 544 81610882177 6 3/4306 612 91812252450 7 5/6340
680102013612722 ↕ typical real-world Modulation/Coding at distance ↕ 8256‑QAM
3/4408 816122516333266 9 5/6453 907136118143629 101024‑QAM
3/45101020153120414083 11 5/65671134170122684537


More PHY speed tables



4804 Mbps speed: The 4804 Mbps maximum PHY speed is for an 160 MHz channel to an
4×4 client. However, a much more realistic maximum PHY speed is 1201 Mbps for an
80 MHz channel to a 2×2 client (840 Mbps throughput), and for a realistic
distance away from the router, a PHY speed of 864 Mbps (600 Mbps throughput).

> 802.11ax is called "HE" for High Efficiency

The goal of Wi-Fi 6: The primary goal of Wi-Fi 6 is 'high efficiency' (HE). In a
nutshell, Wi-Fi 6 adds 'cellular'-like technology into Wi-Fi. This was
accomplished by changing to the OFDMA modulation scheme and changing the Wi-Fi
protocol to directly support many users at once. The result is greatly improved
overall (aggregate) capacity in highly 'dense' (lot of devices) environments
(like schools, stadiums, convention centers, campuses, etc).

> Multi-user support is baked into OFDMA: This is a critical concept to fully
> understand about Wi-Fi 6. In Wi-Fi 5, 'multi-user' was accomplished via
> MU-MIMO using multiple antennas. HOWEVER, in Wi-Fi 6, there is a SECOND (and
> now primary) 'multi-user' method 'baked' into the protocols called MU-OFDMA.
> Don't confuse MU-OFDMA with MU-MIMO! (archive.org) Also, see this interesting
> MU-OFDMA vs MU-MIMO (archive.org) article.

MU-OFDMA (Multi-User OFDMA): The efficiency gains in 802.11ax primarily come
from using OFDMA in 'dense' (lots of users) environments -- breaking up a
channel into smaller Resource Units (RU) -- where each RU is (potentially) for a
different user. There are up to 9 users per 20 MHz channel (so up to 36 users
per 80 MHz channel). So, 802.11ax has high efficiency multi-user transmission
built into the protocol, meaning that the user must be 'Wi-Fi 6' to take
advantage of this. Capacity to a large number of users at once (as a whole)
should dramatically increase (the design goal of 802.11ax was a 4x improvement).

> This multi-user support is a big deal, and will greatly improve Wi-Fi for all
> -- but it will take many YEARS before most clients are 802.11ax. So don't
> expect to see Wi-Fi 6 benefits for YEARS.

But what about peak speed to ONE user: Please note that 'peak' speed for one
user using the entire channel at distance changes very little (around 11%
improvement over 802.11ac). So, if you are looking for much higher Mbps download
speeds (benefiting just one user), 802.11ax is not the solution (eg: PHY speed
at 256-QAM 3/4 in 802.11ac of 780 Mbps changes to 864 Mbps in 802.11ax).
Instead, find a way to increase the MIMO level (or channel width) of the one
user.

> The goal of every prior version of Wi-Fi was dramatically increasing 'peak'
> speeds (for one user). And by looking at Wi-Fi generation Mbps speeds, you can
> see this: 2 -> 11 -> 54 -> 217 -> 1733 -> 2401, except for the last jump,
> which is Wi-Fi 6. Instead, by changing to MU-OFDMA in Wi-Fi 6, there will be
> dramatic (overall) capacity gains to a dense set of users (as a whole), but
> only when (all) clients fully support Wi-Fi 6.


Keep all of the marketing hype in perspective: In order to take advantage of
Wi-Fi 6 improvements, you need client devices that support Wi-Fi 6 and you must
be very close to the router. Until this happens, Wi-Fi 5 will do just fine in
most homes. Most of the speed advances in 802.11ax (MU-OFDMA) will NOT
materialize until ALL client devices are 802.11ax, which will take a LONG time.
So an 802.11ax AP used today will actually be operating in (revert back to)
802.11ac (Wi-Fi 5) mode for many clients.


1024-QAM: This higher order QAM is now officially part of the standard, but you
will need to be very close to the router/AP to actually obtain this QAM. Also,
this modulation can only be used when a client is using an entire 20-MHz (or
wider) channel -- so NOT available for small RU's. In order to achieve 1024-QAM,
you will need an excellent signal (be very close to the router). Note that each
time you double channel width, that there is a 3 dB 'penalty':


Channels: The channels in Wi-Fi 6 are exactly the same as the available channels
in Wi-Fi 4 and Wi-Fi 5. However, since there is so much more spectrum in 5 GHz
than 2.4 GHz, what matters the most for Wi-Fi 6 are the available channels in 5
GHz.

Channel Width: Unlike 802.11ac, which required clients to support 80 MHz
channels, 802.11ax permits 20 MHz channel only clients. This was done to better
support low-throughput low-power IoT devices (eg: those devices powered by
battery) that would take a range/power hit using wider channel widths.

160 MHz Channel WARNING: There are currently only two 160 MHz channels in the 5
GHz band, and they both intersect with DFS channels. 160 MHz channels MAY be
used, but the router, upon sensing a DFS conflict, may automatically revert the
router back to a non-DFS 80 MHz channel.

> Support for 160 MHz channels in some routers reduces MIMO support. For
> example, in Netgear's RAX120, there is 8×8 MIMO support for 80 MHz channels,
> but only 4×4 MIMO support for 80+80 channels.

Bands: Technically, 802.11ax does also operate in 2.4 GHz, but since there are
NO 80 MHz channel there, most people (especially home installations) will stay
in 5 GHz. It has been said that 802.11ax is in 2.4 GHz mainly for the benefit of
IoT device support, but it remains to be seen if that will happen at all -- as
most low power IoT devices stuck with Wi-Fi 4 and never even implemented Wi-Fi
5.

> 6 GHz spectrum: The FCC opened up the 6 GHz and for Wi-Fi (but this requires
> new hardware). See Wi-Fi 6E [§12] in the next chapter. This will make 160 MHz
> channels commonplace.

WPA3: For a device to be Wi-Fi 6 'certified', it was announced (wi-fi.org) that
WPA3 support is a mandatory feature.

TWT: Target Wake Time (TWT) is new feature that is supposed to improve battery
life for IoT devices by better scheduling a time in the future when the device
needs to wake up. However, a warning: some people are having issues with TWT and
are forced to turn TWT off to get their devices to work. It seems this feature
will take time to mature and work properly.

Works better outdoors: 802.11ax changed symbol timings (from 3.2µs to 12.8µs;
and increased GI times), which allows for Wi-Fi to operate much better in
outdoor environments, where signal reflections take more time and can cause
problems. The increased timings (symbols+GI) account for these reflections.

Symbols/sec: With OFDMA, 802.11ax uses either 17,205,882 symbols/sec (800ns
short GI), or 16,250,000 symbols/sec (1600ns medium GI), or 14,625,000
symbols/sec (3200ns long GI), for a 20 MHz channel.

HERE COMES THE HYPE: Manufacturers are touting incredible speed claims regarding
802.11ax (example seen immediately below). However, we know that an 802.11ac 2×2
client at 256-QAM 3/4 has a PHY speed of 780 (see table above chapter). And with
802.11ax (and everything else the same -- like same distance from router), then
PHY speed is 864 (see table immediately above). YES, that is better by a little
(11%), but not nearly as much as you are led to believe.


Very deceptive router manufacturer speed comparison
The above "2.3X" above is comparing 'apples to oranges' -- different channel
widths and different modulation+coding, and combining the total of two bands
(2.4 GHz and 5 Ghz). When you compare 'apples to apples' the raw PHY speed
advantage of 802.11ax over 802.11ac is 38%, and 'at distance' only 11%.

> Analogy: It should be painfully obvious by now that router manufacturers are
> selling you on hype. They are selling you on a 'dragstrip' (the router), where
> you can 'legally' go '1000 mph' -- and that sounds fantastic, so you buy the
> dragstrip (router). But then you step back and realize that (1) all the
> vehicles (Wi-Fi devices) you own don't go over 120 mph, (2) you can buy faster
> cars but they are not legal for you (desktops have faster Wi-Fi than
> smartphones), and (3) 1000 mph was obtained by adding the speeds of multiple
> cars together (aggregating multiple Wi-Fi bands).

Is an upgrade from Wi-Fi 5 to Wi-Fi 6 worth it? The conventional wisdom is that
Wi-Fi 6 offers very little advantage over Wi-Fi 5. And that is 100% true 'on
paper' for client devices 'at distance' that only use 80 MHz channels.

> "The bottom line is until Wi-Fi 6 / 802.11ax clients reach critical mass, the
> benefits of 11ax are minimal and will have low impact". Source (cisco.com).
> 
> "For [most enterprise customers], we recommend installing 802.11ac wave 2
> access points today, because of the sheer value of 802.11ac wave 2". Source
> (cisco.com).
> 
> Consumer Reports concludes (consumerreports.org) that "there is very little
> point in buying a new Wi-Fi 6 router, especially if your smartphone, TV,
> laptop, etc. only support Wi-Fi 5".

However, for whatever reason, support in Wi-Fi 6 for 160 MHz channels in client
devices is now very common (in Wi-Fi 5, it was very rare), which means that
actually regularly obtaining 1 Gbps throughput in the 5 GHz band is now very
possible.

> Also, if your client device is in the same room as a Wi-Fi 6 wireless router
> (so not 'at distance'), you may see a modest speed boost using Wi-Fi 6 over
> Wi-Fi 5.

Regardless of what I and others say, be informed with the facts (and not hype)
and make your own (fully educated) upgrade decisions. Look at your PHY speed
before and after a router upgrade and decide for yourself if the change was
worth it.

UPDATE March 2022: When Wi-Fi 6 CAN deliver the goods: If you have a brand new
Wi-Fi 6 client device and a brand new Wi-Fi 6 router supporting DFS channels,
and are using both in the same room (so both devices are very close to each
other) there is a high likelihood that the two devices will negotiate an initial
160 MHz channel width (Windows laptops and Android; not iOS). Throughput can be
as high as 80% of the 2402 Mbps PHY speed (or around 1900 Mbps) -- which is very
nice! However, this only happens when the client device and router are very
close to each other (in my testing, four feet away) and only if 1024-QAM can
actually be used -- and once you start adding distance or walls, the two Wi-Fi 6
devices will 'slow down' significantly (use lower QAM levels) and communicate
with each other at much closer to Wi-Fi 5 speeds.

> Warning: There are currently NO Apple Mac/iPhone/iPad client devices that
> support HE160 in Wi-Fi 6 in the 5 GHz band. Apple currently only supports 160
> MHz channels with Wi-Fi 6E in the 6 GHz band.
> 
> WARNING: Pay close attention to PHY speeds, because the ability to
> successfully use QAM-1024 (or not) will impact throughput results, which is
> mostly 'out of your control' (just the RF environment at your location).

PHY speed 866.6 Mbps
2.8×
2402 Mbps


★ The PHY speed increases in Wi-Fi 6 (802.11ax): Most Wi-Fi 6 2×2 MIMO client
devices, newly supporting 160 MHz channels, saw a 2.8× maximum PHY speed change
from 866.6 Mbps to 2402 Mbps, from:
 * 11% - 866.6 Mbps in 802.11ac becomes 960.8 Mbps via the switch to OFDMA,
 * 25% - which then becomes 1201 Mbps via a new 1024-QAM modulation,
 * 100% - which then becomes 2402 Mbps by using a 160 MHz channel width (instead
   of 80 MHz), but this requires a router with DFS channel support.

HE160 Throughput: For a 2400 Mbps PHY (2×2 MIMO client; 160 MHz channel),
expect: (1) throughput speed tests [§D] to peak at 'around' 1900 Mbps when very
close to the router, and (2) speeds to decrease quickly with distance and
obstacles.

> This comes with the huge caveat that 160 MHz channels are actually usable for
> you in 5 GHz (that there are no DFS channel conflicts). Technically, the
> increase from 160 MHz channels belongs under Wi-Fi 5, but I am now moving that
> here under Wi-Fi 6 (as of December 2023) -- because these 'speed increase'
> sections are for what most 2×2 client devices experience. Support for 160 MHz
> channels in Wi-Fi 5 for battery powered client devices was rare. But now in
> Wi-Fi 6, support is very common (but still not in Apple devices).

--------------------------------------------------------------------------------


PHY speed 866.6 Mbps
1.4×
1201 Mbps


Most Wi-Fi 6 2×2 MIMO client devices, still only supporting 80 MHz channels (eg:
Apple), only saw a modest 1.4× maximum PHY speed change from 866.6 Mbps to 1201
Mbps. What a mess -- speeds in Wi-Fi depend heavily upon client device
capabilities.

HE80 Throughput: For an 1201 Mbps PHY (2×2 MIMO client; 80 MHz channel), expect:
(1) throughput speed tests [§D] to peak at 'around' 1000 Mbps when very close to
the router, and (2) speeds to decrease quickly with distance and obstacles.


Learn More:
 * 802.11ax High Level Overview (youtube.com) -- Video by Ruckus
 * White paper on 802.11ax (arubanetworks.com)
 * Technical White Paper: "802.11ax: The Sixth Generation of Wi-Fi" (cisco.com)
 * White Paper: IEEE 802.11ax, The Sixth Generation of Wi-Fi (broadcom.com)
 * Dual-Band 11AX (quantenna.com)
 * 802.11ax (wikipedia.org)
 * 802.11ax Resource Units (wikipedia.org)
 * What is 802.11ax Wi-Fi (ruckusnetworks.com)
 * High Efficiency Wi-Fi- 802.11ax - WLPC US Phoenix 2017 (youtube.com)


12. Wi-Fi 6E -- Wi-Fi 6 Extended into the 6 GHz band
★ The big advance in Wi-Fi 6E is a TON more spectrum/channels (the entire 6 GHz
spectrum) -- adding 14 new 80-MHz channels! This makes 160 MHz channels actually
usable and very commonplace.





Wi-Fi 6E: Wi-Fi 6E (High Efficiency Wi-Fi 6 Extended into the 6 GHz band) has
the potential to be a game changer. It adds 1200 MHz (5925 MHz - 7125 MHz) of
new spectrum to Wi-Fi. So, to be clear, Wi-Fi 6E is NOT a new version of Wi-Fi
protocols, but rather it only moves existing Wi-Fi 6 (802.11ax) into a very
large section of new spectrum.

> Wi-Fi 6 in the 5 GHz band DOES have 160 MHz channels, but usage is not
> guaranteed since 160 MHz channels overlap with DFS channels. Meaning that you
> can attempt to use the channels, but that the router may detect a DFS conflict
> and automatically revert back to a non-DFS 80 MHz channel.
> 
> There is only 560 MHz of spectrum currently available to Wi-Fi in 5 GHz (and
> only 70 MHz in 2.4 GHz). So adding an additional 1200 MHz in 6 GHz is a very
> welcome and significant jump in spectrum.
> 
> Apple has finally started to offer HE160 in Wi-Fi 6E, but only for the highest
> top-of-the-line Mac/iPhone/iPad models, and only in 6 GHz (not in 5 GHz).

Wi-Fi 6E spectrum GHzMHzNameAFC? 5.925-6.425500U-NII-5YES
6.425-6.525100U-NII-6no 6.525-6.875350U-NII-7YES 6.875-7.125250U-NII-8no

The big deal: TONS of new spectrum! The additional spectrum allows for 14
additional 80 MHz channels (or seven additional 160-MHz channels) in Wi-Fi,
which means the chances of sharing spectrum with another device/neighbor will be
greatly reduced. You should then have your own 160 MHz channel all to yourself,
potentially doubling throughput (vs an 80 MHz channel).

> Many entry-level Wi-Fi 5 routers (with no DFS support) only support 180 MHz of
> spectrum. But I expect entry-level Wi-Fi 6E routers (with no AFC support) to
> support all 1200 MHz of spectrum.
> 
> Also of note is that Wi-Fi currently has no 160 MHz channel that is not
> subject to DFS restrictions, meaning that currently, actually being able to
> use a 160 MHz channel today in 5 GHz is hit or miss. This new spectrum should
> (hopefully) make it much easier to find and actually use multiple 160 MHz
> channels.

The gotcha: New hardware (routers/clients) will be required. Current Wi-Fi 6
devices don't support Wi-Fi 6E!

Two power modes: Wi-Fi 6E routers have two different power modes. "Low-power"
mode and "Normal-power" mode. Expect most routers to only support "Low-power"
mode (for now).



6 GHz Wi-Fi channels (U.S.) Channel #20 MHz
center20 MHz
channel 320160804020 311573159555945-5965 559755965-5985 11959955985-6005
1360156005-6025 23191760356025-6045 2160556045-6065 272560756065-6085
2960956085-6105 4739353361156105-6125 3761356125-6145 434161556145-6165
4561756165-6185 55514961956185-6205 5362156205-6225 595762356225-6245
6162556245-6265 957971676562756265-6285 6962956285-6305 757363156305-6325
7763356325-6345 87838163556345-6365 8563756365-6385 918963956385-6405
9364156405-6425 111103999764356425-6445 10164556445-6465 10710564756465-6485
10964956485-6505 11911511365156505-6525 11765356525-6545 12312165556545-6565
12565756565-6585 15914313513112965956585-6605 13366156605-6625
13913766356625-6645 14166556645-6665 15114714566756665-6685 14966956685-6705
15515367156705-6725 15767356725-6745 17516716316167556745-6765 16567756765-6785
17116967956785-6805 17368156805-6825 18317917768356825-6845 18168556845-6865
18718568756865-6885 18968956885-6905 20719919519369156905-6925 19769356925-6945
20320169556945-6965 20569756965-6985 21521120969956985-7005 21370157005-7025
21921770357025-7045 22170557045-7065   22722570757065-7085 22970957085-7105
 23371157105-7125

More info from Wikipedia and here

Low-power mode: In 'low-power' mode, access points are permitted to use the
entire 1200 MHz of spectrum with no AFC restrictions, but range is less (and is
an unknown right now until tests are performed on real hardware), and use is
restricted to indoor use only.

> Wi-Fi 6E access points in low-power mode are permitted to operate at 24 dBm
> EIRP (6 dB BELOW 5 GHz DFS power levels), and Wi-Fi 6E clients at 18 dBm EIRP
> (6 dB BELOW that of the AP). Many Wi-Fi 5 clients today already operate
> 'around' this power level, so Wi-Fi 6E range will be affected by the slightly
> higher operating frequencies, and the 6 dB power difference (below DFS).
> 
> A change in Wi-Fi 6E, for the 6 GHz spectrum, is that clients will now operate
> at a constant 5 dBM/MHz PSD (Power Spectral Density).

Normal-power mode: In normal power mode, access points are only permitted to use
850 MHz of spectrum (see table upper right), but are required to use something
called AFC (see below), which requires the access point to report its geo
location (GPS), as well as serial number to a centralized database. It remains
to be seen if customers will accept this 'invasion of privacy'.

> Wi-Fi 6E access points in normal-power mode are permitted to operate at 36 dBm
> EIRP (the same power levels of 5 GHz U-NII-1 power levels), and Wi-Fi 6E
> clients at 30 dBm EIRP (6 dB BELOW that of the AP). Most Wi-Fi 5 devices
> already operate below these levels, so Wi-Fi 6E range will be affected only by
> the slightly higher operating frequencies.

Automated Frequency Coordination (AFC): The FCC docs (fcc.gov) extensively
discuss an 'Automated Frequency Coordination' (AFC) system to avoid conflicts
between existing licensed use (point to point microwave) and new unlicensed
devices (access points). It appears that the FCC has settled (govinfo.gov) (May
26, 2020) on a centralized AFC system whereby an access point must contact the
AFC "to obtain a list of available frequency ranges in which it is permitted to
operate and the maximum permissible power in each frequency range". But in order
for this to work properly, the access point MUST report its geo-location (eg:
GPS location), as well as antenna height above the ground, to the centralized
AFC system. The FCC will also require the 'FCC ID' of the access point, as well
as the serial number of the access point.

> Privacy mitigating factors: An access point can operate in 'low power mode'
> and then NOT be subject to AFC (but then signal range WILL suffer) OR, the
> access point can reduce the GPS quality and then report a larger general
> 'area' to the AFC instead of an exact location (but then frequencies and power
> levels that can be used might be reduced).

A major concern: Range: A major concern is what range will be for Wi-Fi 6E
devices. Based upon raw specifications, range will be reduced over what is
possible in 5 GHz.

> In some of my speed testing 'at distance', 5 GHz channels are providing much
> better real Mbps throughput than 6 GHz channels. This needs more research.

Another concern: the spectrum is already heavily used: The 6 GHz spectrum that
the FCC wants to open up (for unlicensed Wi-Fi use) is already being "heavily
used by point-to-point microwave links and some fixed satellite systems" by
existing licensed services. Source (fcc.gov). So, it remains to be seen how many
channels can actually be used in real-life with AFC for normal-power devices.

> Incumbent Services: The FCC did not just have 1200 MHz of spectrum laying
> around unused. Instead, this spectrum is heavily used by 'incumbent services',
> such as:
>  1. Fixed Microwave Services (FS): You have probably seen these towers around
>     with a 'dish' pointed in a fixed (horizontal) direction. Just Google
>     microwave tower (google.com).
>  2. Fixed Satellite Services (FSS): Ground to satellite communication (and
>     vice-versa).
>  3. Radio Astronomy: The study of celestial objects at radio frequencies.
>  4. Other miscellaneous services: Mobile services, etc.
> 
> Interesting: Video showing FS usage in the US (youtube.com)

So how well Wi-Fi 6E actually works for you will literally depend upon your
physical location (and nearby incumbent services).

Best use case: The first wave of Wi-Fi 6E devices are operating in only
'low-power' mode (as no AFC is required and the entire 1200 MHz can be used; but
restricted to indoor use only), but range will be reduced. When combined with
effective range decreasing with channel width, the best use case for Wi-Fi 6E
160 MHz channels is between two devices in the same room.

> The thinking is that with Wi-Fi 6E and 160 MHz channels, a 2 Gbps PHY
> connection with reliable 1 Gbps actual throughput becomes commonplace (instead
> of hit or miss) when are you in the same room as the access point -- with only
> a 2×2 MIMO client device.

Interesting observations about Wi-Fi 6E from this FCC doc (govinfo.gov):
 * Low-power access points can use the entire 1200 MHz spectrum, but use is
   restricted to indoor use only, and range will be limited. These low-power
   devices cannot be weather resistant, must have permanently attached
   integrated antennas, cannot be battery powered, and must be labeled "for
   indoor use only".
 * Normal-power access points can be used outdoors, but must use AFC, and are
   restricted to using U-NII-5 or U-NII-7.
 * Client devices are prohibited from being used as a mobile hotspot.
 * Access points are prohibited in moving vehicles such as cars, trains, ships,
   or small aircraft (but with an exception for large passenger aircraft
   operating over 10,000 feet, but may only use U-NII-5).
 * Access points are prohibited on ships and oil platforms.
 * Use is prohibited on unmanned aircraft systems.
 * Devices using AFC must report geo-location, geo-location accuracy, antenna
   height above ground, device FCC ID, and device serial number to a centralized
   AFC database, which then returns frequencies and power levels that may be
   used. The device must contact the AFC database at least once per day (failure
   means stop working; with one day grace period)
 * Assumes that there will be 17 dB in signal loss when the signal from an
   indoor access point travels outside through a building's walls.
 * Sets a maximum channel width of 320 MHz.
 * Client devices are prohibited from transmitting anything until the device
   hears something from an access point (so no probe requests).
 * Client devices must operate at 6 dB below the power level of the access point
   power level.
 * An underlying AFC presumption is that access points are at a fixed location
   (not mobile nor moving around).

Well, that was fast: Wi-Fi 6E was announced (wi-fi.org) on January 3, 2020. Days
later, Broadcom announced chipsets (broadcom.com) supporting Wi-Fi 6E in 6 GHz.
Then on April 24, 2020, the FCC moved forward (fcc.gov) in supporting this.
Summary (fcc.gov). First Wi-Fi 6E routers started shipping in 2021 Source
(dongknows.com). Today, virtually all new high-end client devices phones,
laptops, etc fully support Wi-Fi 6E.



PHY speed 2402 Mbps
0×
2402 Mbps


★ The PHY speed increases in Wi-Fi 6E: There is technically NO speed increase in
Wi-Fi 6E over Wi-Fi 6. However, Wi-Fi 6E will make 160 MHz channels much more
commonplace for 2×2 MIMO client devices.

> However, if you did not see the 2× speed change (from 80 MHz to 160 MHz
> channels) in Wi-Fi 6 (eg: Apple devices), you should now see that speed
> increase with Wi-Fi 6E (but only when using the 6 GHz band).

Throughput: Same as Wi-Fi 6 [§11]. However, expect speeds to be impacted even
more by 'distance from router'. Wi-Fi 6E excels when you are very close to the
router. But, when not, you will likely see better throughput using Wi-Fi 5/6 in
the 5 GHz band.


Learning More:
 * FCC Opens 6 GHz Band to Wi-Fi and Other Unlicensed Uses (fcc.gov)
 * FCC Technical Report and Order 20-51 (fcc.gov)
 * Power Spectral Density and Wi-Fi 6E (juniper.net)


13. Wi-Fi 7 -- 802.11be (EHT: Extremely High Throughput)



Yes, there are early Wi-Fi 7 routers for sale on Amazon, right now -- and
phone/tablets are starting to support Wi-Fi 7. BUT, just be patient.

> It will take time for Wi-Fi 7 implementations to mature, become stable, and
> actually implement all Wi-Fi 7 features. I see Wi-Fi 7 routers being sold
> today that:
>  * promise to implement MLO in a future firmware version
>  * state "This router may not support all the mandatory features as ratified
>    in the IEEE 802.11be specification"
> 
> So frankly, buyer beware!

> GenSpecYearPHY SpeedsNew Name Seventh802.11be2023(todo)Wi-Fi 7

★ The big changes that are expected to be introduced in Wi-Fi 7 are:
 * 320 MHz channels contiguous (320 MHz) or non-contiguous (160+160 MHz)
 * 240 MHz channels contiguous (240 MHz) or non-contiguous (160+80 MHz)
 * allows 16×16 MIMO / spatial streams (vs prior 8×8 limit) -- but only expect
   2×2 in virtually all client devices, and 4×4 in most mid to higher end
   routers
 * 4096-QAM modulation (vs prior 1024-QAM) -- but only obtainable if you are
   VERY close to the router, and are line-of-sight (no obstacles)
 * Multi-Link Operation, or MLO, uses multiple bands/channels at the SAME time
   -- for example using 2.4 GHz, 5 GHz, and 6 GHz all concurrently.
 * Preamble Puncturing - if part of a 'wide' channel is being used by an old
   legacy client device, the rest of the channel can still be used by a modern
   Wi-Fi device
 * plus other potential new features (as the Wi-Fi 7 specification has not yet
   been 'ratified' -- expected in early 2024)

> 802.11be is called "EHT" for Extremely High Throughput

TP-Link has a nice web page (tp-link.com) explaining all new Wi-Fi 7 features
(scroll down to the "How Does WiFi 7 Work" section).




TP-Link BE550

Prices are just starting to drop: Prices are finally starting to drop! TP-Link
is selling a very solid entry level 2×2 MIMO tri-band Wi-Fi 7 router now for
around $220 to $250, the BE550 (amazon.com), seen right, with all 2.5 Gbps
Ethernet ports (WAN+LAN). A very good value.

Please note that Wi-Fi 7 routers will only benefit Wi-Fi 7 clients (not Wi-Fi
4/5/6 clients). And once you do actually have Wi-Fi 7 clients, there is very
little point in upgrading your router to Wi-Fi 7 if you are not currently
noticing any speed issues. Give Wi-Fi 7 technology time to mature.

The 'distance' issue: The benefit for Wi-Fi 7 will be mainly for devices that
are very 'near to' (in the same room) as the router/AP -- expect a maximum PHY
speed of around 4.8 Gbps for a 320 MHz 2×2 channel.

> Expect speeds to drop very quickly with distance.

Symbols/sec: 802.11be uses either 17,205,882 symbols/sec (800ns short GI), or
16,250,000 symbols/sec (1600ns medium GI), or 14,625,000 symbols/sec (3200ns
long GI), for a 20 MHz channel -- the same as the prior 802.11ax.

4096-QAM HYPE: I have some strong reservations about how often 4096-QAM in Wi-Fi
7 can actually be used. Because even today with Wi-Fi 6E, with 1024-QAM, I
sometimes find that being just feet from the router is not good enough to even
use 1024-QAM so there is no hope to use 4096-QAM, which requires an even better
signal. Maybe I just need to be patient and wait for the second generation of
Wi-Fi 7 devices?

> Is 4096 QAM even possible? "The answer is a mixed bag. Possible? Yes, as Wi-Fi
> 7 Extremely High Throughput does add the features needed, it is possible to
> achieve this in the wild. Is it probable? Probably not in many scenarios.
> While very cool and very fast, it's also fragile and we expect to only see
> this level of modulation in a small number of instances." Source
> (ruckusnetworks.com)
> 
> So we are being 'sold' something that -- for all practical purposes -- we can
> rarely use! For now, 1024-QAM is a reasonable maximum QAM.
> 
> Q: Can any reader point me to an online review and Wi-Fi 7 speed test that
> shows 4096-QAM actually being successfully used?



Mesh network systems Wi-Fi 7 MLO conundrum: Mesh network systems often use a
backhaul operating simultaneously on a different channel/band. But Wi-Fi 7 MLO
is already using all antennas for MLO. It will be interesting to see how mesh
systems ultimately deal with this.

> I see one mesh network system dealing with this by adding a non-MLO 240 MHz
> backhaul channel in 5 GHz (quad-band). Don't expect mesh systems to deliver
> MLO throughput speeds all the way from client device to the main router from a
> satellite mesh node. The speed advantages that Access Points have over mesh
> systems will become much more obvious with Wi-Fi 7.

PHY speed 2402 Mbps
?×
???? Mbps


★ The PHY speed increases in Wi-Fi 7 (802.11be): The speed increases came from:
 * 100% - doubling channel width from 160 MHz to 320 MHz but only for 6 GHz band
 * 20% - via a new 4096-QAM modulation but only when VERY close to the router
 * MLO - (todo)
 * this section is a work-in-progress -- waiting to see what becomes 'common' in
   client devices (especially phones)

EHT Throughput: to be determined, but expect very high PHY speeds and great
throughput (3+ Mbps) right next to the router, which then decrease VERY quickly
with distance and obstacles.


14. DFS channels in 5 GHz
This chapter applies to all versions of Wi-Fi operating in the 5 GHz Wi-Fi band
(does not apply to 2.4 GHz and 6 GHz bands).

DFS: DFS stands for Dynamic Frequency Selection. The 5 GHz band is already used
by other (non Wi-Fi) 'incumbent' services, so the FCC allows a Wi-Fi router to
use a 5 GHz channel only when the router detects that the channel is 'free'. And
if the router later detects an 'incumbent' service using the channel, the router
must immediately stop using the channel (and switch to another channel).

In a nutshell: First, you need DFS channels if you want 160 MHz channel support.
Next, if you buy a router that does not support DFS channels, you are limited to
only having TWO 80 MHz channels available in 5 GHz (instead of six/seven
channels), greatly increasing the likelihood of sharing that channel with others
(a close neighbor) -- meaning that you are sharing bandwidth. If your router
supports DFS channels, your likelihood of being on your own channel all by
yourself is much higher -- meaning all channel bandwidth is yours.

> UPDATE: Beware that some very inexpensive routers might only support a SINGLE
> 5 GHz channel. Just do your research before a purchase!

The 5 GHz 80-MHz channels Channel #MHzInfo 4236+40+44+485170-5250OK
5852+56+60+645250-5330DFS 106100+104+108+1125490-5570DFS
122116+120+124+1285570-5650DFS, TDWR 138132+136+140+1445650-5730DFS (1)
155149+153+157+1615735-5815OK 171165+169+173+1775815-5895Indoor (2)

Background: There are SEVEN core 80-MHz Wi-Fi channels in 5 GHz. Two channels
can always be used (green highlight, right), and one is new as of 2019 (and for
indoor use only). But, for the other four DFS channels to be used, a router must
include special processing to avoid interference with existing usage (weather
radar and military applications; red highlight, right) and pass FCC
certification tests.

> (1) Many Netgear Wi-Fi 5 routers do not support CH 144, and as such, can't
> support 80 MHz channel 138. This flaw was fixed in Netgear Wi-Fi 6 routers.
> 
> (2) The FCC added 20 MHz channels 169, 173, and 177 in 2019 (for indoor use
> only), which also creates a new 80 MHz channel 171. But don't expect support
> in very many devices (yet) for the new channels.

Why DFS support is important: Support for all channels becomes critically
important to avoid interference (sharing bandwidth) with a neighbor's Wi-Fi.
Ideally, every AP/router (yours and neighbors) should be on a unique/different
Wi-Fi channel.

> Also, this is especially important if you can see several other 5 GHz AP's,
> which happens when you (1) have close neighbors like in an apartment building,
> or (2) want to install multiple AP/routers. So, only consider AP/routers that
> support ALL the DFS channels.

160 MHz channels: DFS channel support is needed in order to support contiguous
160 MHz channels in 5 GHz. If you need 1 Gbps throughput speeds, then DFS
channel support becomes a 'must have' feature.

Range: An AP/router for DFS channels has a transmit power limitation of 250 mW
(vs 1000 mW for non-DFS channels). However, this rarely limits range to clients,
as virtually all client devices already transmit at less the 250 mW for ALL 5
GHz channels (so the client device limits range, not the AP/router). More
information on range [§H].

Avoid AP/routers with NO DFS channels: It is very common to find
'consumer-grade' routers that support NONE of the DFS channels (they only
support TWO channels). Buyer beware. Also beware brand new routers with NO DFS
channel support, as the vendor may not release a firmware update that adds
support for these DFS channels (don't buy a device on the hope that DFS support
will be added later via a firmware update).

Some 'consumer-grade' AP's DO support some DFS channels: Some consumer grade
routers DO support some or all of the DFS channels. Just do your research.
Support is becoming more common.

> Netgear ALERT: Most Netgear routers don't support 80 MHz channel 138. But this
> is slowly changing. The R7800 is a rare exception, supporting channel 138, but
> only via firmware 1.0.2.68. Also, it appears that Netgear is finally 'aware'
> of the issue as some of the newer 'AX' hardware also supports channel 138.

Some business-grade AP's DO support 5 GHz DFS channels: Some business-grade 5
GHz devices DO support the DFS channels, so you get the full advantage of a LOT
more channels in 5 GHz.

> Most of the Netgear business access points (Netgear ProSafe Access Points) do
> NOT support the restricted 5 GHz channels. But I did find ONE that did. Just
> do your research.

Many Enterprise-grade AP's DO support 5 GHz DFS channels: According to this data
sheet (ubnt.com) ALL of the Ubiquiti UniFi AC models (802.11AC Dual-Radio Access
Points) are DFS certified.

> For example, I was in a Drury Hotel and from my room, I could see the Drury
> SSID on channels 48, 64, 100, 104, 108, 140. So the hotel was clearly using
> DFS certified 5 GHz access points -- successfully.

Beware some 'best router' reviews: Watch out for 'best router' reviews online
that select a 'best overall' router that do NOT support ANY DFS 5 GHz channels
(only TWO channels supported).




FCC Operating Frequencies show DFS support

How to research DFS support for any router/AP (check the FCC filings):
 1. Google 'fccid.io' and 'FCC' and the router vendor name and model.
 2. Look for a google search result to the fccid.io web site with 'FCC' in the
    link and click on the link. If not on the 'home page' for the router, there
    will be link near the top of the page home page for that device.
 3. As best you can, confirm that you found the correct router by looking at the
    'external photos' document. Once confirmed, go back to the previous (home)
    page.
 4. In the resulting home web page, look for the "Operating Frequencies" section
    (example seen right).
 5. Look for frequencies that cover the DFS channel range (highlighted in yellow
    right). [Frequency ranges are usually based upon '20 MHz center' values]. If
    so, that router/AP HAS DFS channel support. Otherwise, there is NO DFS
    channel support. For the RAX80, notice that it appears that DFS channels are
    supported (except for channel 144), which then also excludes channels 142
    and channel 138 -- because the GHz range stops at 5.7 GHz instead of 5.72
    GHz).
 6. Look for the 'DFS Test Report' and see if the device is a master or a slave,
    or both (see immediately below). You are looking for 'master' (router)
    support.

DFS Master/Slave: When looking at FCC filed documents, look for and open up the
"Test Report (DFS)". The report will then talk about the EUT (Equipment Under
Test) being certified as a 'Master' or a 'Slave' (or both). Master means a
router/AP (broadcasts a SSID) and Slave means a device that connects to a Master
(Wi-Fi client). A device is not allowed to use any DFS channels unless the
proper paperwork is filed with the FCC.

> Netgear was plain lazy: Netgear got the R6700v3 certified as a DFS Master
> (fccid.io), or Wi-Fi router mode, but failed to get the router certified as a
> DFS Slave (Wi-Fi client mode). This matters if you use the R6700v3 as a
> 'wireless bridge' (to connect 'ethernet only' devices to your main Wi-Fi
> router), because all of a sudden, in that mode, the R6700v3 no longer supports
> DFS channels -- meaning that if you bought the R6700v3 to connect to your main
> router (broadcasting/using a DFS channel), the R6700v3 will NOT work!

Warning: Just because a router allows for DFS channels does not mean DFS
channels can be used: Be aware that when a DFS channel is selected, the router
MUST look for conflicts on that frequency, and if a conflict is found, the
router must automatically change the channel (likely to a non DFS channel). You
won't know until you try. Often times, one or two of the DFS channels can not be
used (but the other DFS channel can). And each physical location is different.
You won't know until you try.

> I have even selected a DFS channel and seen it work for weeks, only for the
> router to then all of a sudden auto select a non-DFS channel (meaning the
> router detected a 'conflict'). Was this a real radar signal detected, or a
> false alarm (most likely)? You just need to be patient finding a DFS channel
> that works long-term for you.
> 
> Often times, when a router automatically switches to a non-DFS channel, that
> change is temporary -- as simply power cycling the router will cause the
> router to once again use the (configured) DFS channel.

Warning: Not all Wi-Fi clients are DFS capable! All of the above is discussing
DFS support in routers, because that is where ALL of the hard work takes place
(like scanning for radar, etc). Wi-Fi clients have it easy -- just follow the
lead of the router. And yet, it is possible that a Wi-Fi client never got DFS
certified, and therefore is NOT permitted to use DFS channels, and can NOT
connect to a router using any DFS channel.

> A Wi-Fi client not supporting DFS channels is very rare -- and is definitely
> incredible laziness on the part of the device manufacturer. Often times, you
> will never actually notice, because the problem device will just connect to
> the router's slower 2.4 GHz band, ignoring the fast 5 GHz DFS band.
> 
> 
> 
> 
> 
> Multiple readers have told me that most Roku devices do not support DFS
> channels! And Roku's own support article confirms this. That is crazy laziness
> on Roku's part, and is not at all customer friendly.


15. Router/Wi-Fi setup tips
Setup: Web Browser vs App: There is no right or wrong answer. Just use whatever
method works best for you. I personally setup new routers via the web browser
interface because I don't want the hassle of having yet one more app installed
on my phone. Just be aware that some companies (Netgear) seem to add more
features in their apps vs web browser interface.

Placement: Take the time to correctly place your main router [§P] in your home.




Netgear R7800 wireless setup

SSID: SSID is simply the Wi-Fi network NAME. When you connect to a Wi-Fi network
in a client, you must select this network name (called SSID). At home, you
typically will only have one router with that ONE network name.

However, if you add another Wi-Fi access point [§18], you want it to use the
SAME network name (and password + security), as this allows for Wi-Fi roaming.
Your wireless devices simply connect to the strongest Wi-Fi signal with a
matching SSID name.

> You can use different SSID names, but then you don't get Wi-Fi roaming.

2.4 GHz and 5 GHz SSID names: There is a big debate -- should your 2.4 GHz band
network and 5 GHz band network have the SAME SSID names, or be 'split', using
different names (often with a "-5G" appended to the 5 GHz band SSID name)? If
named the same, your client device selects which band to connect to. If named
differently, then you must select which band to connect to.

> The problem with the 'same name' technique is that some client devices are
> 'dumb' and incorrectly connect to the 2.4 GHz band instead of the 5 GHz band
> (and then speeds are much slower than they should be). I had this problem with
> a laptop that would initially start out connected to the (fast) 5 GHz band,
> but after about 10 minutes, it would then switch to the (much slower) 2.4 GHz
> band (for unknown reasons). Yes, the 5 GHz RSSI signal was weaker, but
> throughput from 5 GHz was MUCH better. I fixed by appending a "-5G" to the 5
> GHz band SSID, and connecting to that SSID.
> 
> > TIP: Under Windows, look at your Wi-Fi device in the 'Device Manager' --
> > there may be an option called 'Preferred Band' that is set to 'auto'. Change
> > to attempt to force your device to use a particular Wi-Fi band.
> 
> The bottom line: Be practical. Just use whatever SSID naming method works best
> for you and your devices.
> 
> Router "Smart Connect" TIP: If you want to 'split' the SSID names for your
> Wi-Fi bands, first turn the 'Smart Connect' feature OFF (as that feature 'on'
> forces the SSID band names to be the same).

Disabling 'SSID Broadcast' is NOT a form of security: Do not think that
disabling 'SSID Broadcast' will improve the security of your wireless network.
It will not. Because anyone with the right tools (techgenix.com) can still see
your network and find out the network name.

> Because any client device connecting to your network must include the SSID
> name in the 'I want to connect' request, that any Wi-Fi sniffer can capture
> and then see.

BSSID: This is the MAC address of the AP that your client actually connects to
(because you can't tell which AP you connected to from only the SSID). This is
very useful when you have more than one AP using the same SSID, because the
BSSID identifies the unique AP that you actually connected to (a must for
debugging).




80 MHz 5 GHz channel 42 is made by
combining four 20 MHz channels
(only one considered 'primary')

Channel: ALL Wi-Fi access points (in your house and visible neighbors networks)
covering the same frequency must share the Wi-Fi bandwidth. Because of this,
assign channels 1, 6, 11 (2.4 GHz) and 42, 58, 106, 122, 138, 155 (5 GHz) to
your APs in a manner to best avoid conflicts (with yourself and neighbors). Try
to set/use a DFS channel that is stable (does not kick you off the channel after
some time). There is nothing special about channel selection. Any channel can be
selected. However, to maximize throughput, you want Wi-Fi usage spread around
(not everything running on the same channel).

> Analogy: A channel is like a lane on an Interstate highway. All cars (AP) can
> use the same lane (channel), but that is slow and inefficient (lanes go
> unused). Everything works best when cars (AP) use all lanes (channels) -- as
> evenly as possible.
> 
> Could you put 10 APs in your house and configure them to all use the same SSID
> and the same channel? Yes, and it would work -- albeit slowly. But that would
> not be the best and most efficient way to use the available spectrum, since
> all 10 APs are attempting to use the same 'lane' of a superhighway. Instead,
> distribute all 10 APs across all available lanes (channels) of the
> superhighway. Just remember that 2.4 GHz Wi-Fi has overlapping channels and
> that the only real (non-overlapping) channels available are 1, 6, 11.
> 
> Channel Planning: A full discussion on channel planning is beyond the scope of
> this paper, but in short, always try to leave a full channel of unused
> spectrum between actively used channels. In other words, if you have a main
> router on 80 MHz channel 42, never put another nearby AP on 80 MHz channel 58.
> Instead, you would select a higher channel for the nearby AP. If possible, you
> want a 'gap' between active channels to avoid Adjacent Channel Interference
> (ACI). See the appendix on spectral masks [§S] showing the impact of a
> sideband ACI signal on throughput.
> 
> Beware the 'automatic' channel: Some AP/routers improperly set the 2.4 GHz
> channel to some channel other than one of the three non-overlapping channels.
> This can be corrected by not using 'automatic' and instead manually selecting
> either channel 1, 6, or 11.

TIP: Static IP Addresses: In your main router, assign a fixed (static) IP
address to IoT devices that are always connected to your network. This actually
works around some known connectivity bugs in IoT devices (eg: Ring cameras) that
go into a 'deep sleep mode' in order to extend battery life, but this sometimes
causes the router to not 'find' the device (the Ring camera does not respond to
the routers ARP request to 'find' the camera). Assigning a fixed known IP
address allows the router to always 'find' the camera, even when the camera is
in 'deep sleep' mode.





DNS Servers: I configure all of my routers to use Google's Public DNS
(google.com) servers at 8.8.8.8 and 8.8.4.4. In your router, the setting for DNS
servers is usually found in the 'Internet Setup' section.

> Another fast DNS service is provided by CloudFlare, with DNS servers 1.1.1.1
> and 1.0.0.1.
> 
> TIP: And there are many more public DNS servers. Just Google "public dns
> servers" (google.com).

The default DNS setting is typically 'automatic' (so, use the ISP's DNS
servers). But the problem with an ISP's DNS servers is that they (1) can be
slow, (2) often improperly redirect on DNS errors (like 'server ip can not be
found') to some 'self-promotion' web page (and this can cause some software
programs that reply upon 'not found' DNS replies to fail -- not good), and (3)
frankly, sometimes work incorrectly -- violate TTL (time to live) rules. I trust
Google more than the ISP's to properly follow 'time to live' rules.

> I actually ran into this 'TTL' problem while at a vacation property. Long
> story short is that something I depended upon was failing and that I tracked
> down a problem to the ISP incorrectly caching a DNS entry WAY past the TTL
> expiration time. When I used the ISP DNS servers, I received an old stale DNS
> entry. When I used Google's DNS servers, I got a fresh, correct DNS entry. I
> don't trust ISP DNS servers anymore to 'follow the rules'. They can save money
> by breaking the TTL rules (caching information longer than allowed).

When you make this change, all devices locally on your network will (eventually)
automatically use the new DNS servers (except for those devices that manually
override the 'get automatically from the router' behavior).

Possibly turn UPnP OFF: There have been so many security vulnerabilities with
"Universal Plug and Plug" in routers over the years, that the first thing you
might want to do is turn UPnP off. Then just see if everything in your network
still works (it will for most people). If so, great. But if not, then consider
turning UPnP back on (or manually fixing what stopped working).

Change the router password! Change the 'administrator' password on your router!
You don't want a guest (or hacker) gaining easy access to your router and making
changes. I am surprised how often I visit a vacation home only to find the
router set to default credentials (often 'admin/password'), which opens up the
router to unauthorized changes.

> Xfinity gateway's are the worst! They always seem to be 'wide open' via the
> HTTP interface (often http://10.0.0.1). The default username and password is
> admin/password. Are people using the Xfinity app (which requires your account
> username and password) and unaware that the 'web interface' to their gateway
> is still wide open using the default username and password?

Do NOT touch 'Enable WMM': "Enable WMM" is ON by default on ALL routers, because
it is actually needed for any speed past 54 Mbps. Turn if off if you want to see
what I mean (your Wi-Fi speeds will tank). More Information
(smallnetbuilder.com)

Firmware updates: If your router does not update firmware automatically, stay on
top of keeping the firmware up-to-date, as there are security vulnerabilities
fixed (and new firmware releases) all the time.

16. ★How to improve Wi-Fi speeds
Are Wi-Fi speeds 'at range' not as fast as you would like? So, how do you go
about improving Wi-Fi speeds?

> This chapter assumes that your client device, when right next to a 'modern'
> router, is capable of good Wi-Fi speeds. If you have an older client device
> that is not capable of good Wi-Fi speeds, the rest of this chapter won't help
> you because you first need to upgrade your client device, not the router.

FIRST, follow these initial steps:
 1. Reboot ALL Internet connected devices. First, start with your modem, router,
    and Wi-Fi network (extenders, AP, mesh, etc) and then wait 5 minutes (for
    all Wi-Fi bands to be fully operational, especially DFS channels). Next,
    reboot all other Wi-Fi connected devices (game consoles, smart TV's,
    tablets, phones, cameras, etc). In some rare cases, this may fix your
    problem and you are then done.
    
    
 2. Check for proper placement [§P] of your existing Wi-Fi devices, especially
    your main router, and any mesh nodes.

THEN, if problems persist, you are left with fixing either (a) the easy way
immediately below; buy new hardware or (b) the hard way farther below; try to
fix what you have.



The EASY way: The 'easy' way to improve Wi-Fi speeds often involves buying new
hardware (from a vendor with a great return policy). Follow the advice for the
first question where you answer 'yes':

> Q: Are you already using a mesh Wi-Fi network system?
> 
> > If yes, improve your existing Mesh network: Either (a) move your mesh nodes
> > around -- see placement [§P], or (b) add another mesh node, or (c) if your
> > mesh network is old (Wi-Fi 4 or 5), first consider upgrading to a new Wi-Fi
> > 6 (or better) mesh network [§19].
> > 
> > If that does not fix the issue, then you MUST find a way to place the mesh
> > nodes where they are needed most AND add/use a wired/Ethernet backhaul for
> > your mesh network. That WILL 100% fix the problem. Either (a) directly
> > extend Ethernet, or (b) indirectly extend Ethernet via MoCA [§E5] or other
> > similar technology (see farther below in this chapter).
> 
> Q: Are you already using a Wi-Fi extender?
> 
> > If yes, remove any Wi-fi extender(s): Either (a) replace your router and
> > extender(s) with a modern Wi-Fi 6 (or better) mesh network [§19], and go
> > back to the start of this chapter, OR (b) remove the Wi-Fi extender and
> > continue with the next questions (try a Router+AP solution)...
> 
> Q: Are you already using a 4×4 MIMO and Wi-Fi 6 (or better) main router?
> 
> > If yes, then add an Access Point: Add another recommended [§21] 4×4 MIMO
> > Wi-Fi 6 (or better) router -- but configured as an access point [§18] --
> > precisely where it is needed most and wired/Ethernet back to your main
> > router. That WILL 100% solve the problem.
> > 
> > If unable to run an Ethernet cable from the router to the AP location (or
> > you just don't want to use Access Points), then replace your router with a
> > modern Wi-Fi 6 (or better) mesh network [§19], and go back to the start of
> > this chapter.
> 
> Q: Are you willing to spend (around) $200?
> 
> > If yes, then UPGRADE your main router: Try a brand new recommended [§21] 4×4
> > MIMO Wi-Fi 6 (or better) router, and go back to the start of this chapter.
> > 
> > For most people, a great modern router centrally located [§P] in the home
> > will be good enough. Please note that you really do want a 4×4 MIMO router,
> > as it will have both a range and signal quality advantage over cheaper 2×2
> > MIMO routers.
> 
> Otherwise, if unwilling to buy new hardware, your only option is to consider
> improving your current Wi-Fi setup the 'hard way' -- see the rest of this
> chapter below.


The HARD way: The 'hard way' is to try to improve the Wi-Fi network that you
already have.

> This assumes that your network is capable of fast speeds, but is currently not
> obtaining fast speeds. If you are not using modern Wi-Fi devices, the gains in
> speed you can achieve will be minimal.

Wi-Fi is a TIME shared resource. So, the goal for improving things is to get
every Wi-Fi client to use that resource in as little time as possible,
especially those few devices that are 'heavy users'. So, target the 'heavy
users' first with the goal being to 'free up Wi-Fi time' for other Wi-Fi users:
 1. Use Ethernet whenever possible
 2. Improve wireless PHY speeds

Consider these options...



1) Use Ethernet whenever possible

> For any device using Wi-Fi now (smart TV, game console, desktop PC, etc) that
> has a RJ45 jack, try to switch over to using Ethernet instead of Wi-Fi for
> that device. Your goal is to free up Wi-Fi for use by devices that can only
> connect wirelessly.
> 
> a) Best v1: Go direct wired: Gigabit ethernet via Cat 5/5e/6/etc is still the
> gold standard of speed and reliability. If you have a Wi-Fi device that also
> has ethernet/RJ45 (smart tv, game console, etc), find a way to run a wired Cat
> 5/6 from your main router to the device. Expect 1000 Mbps PHY and 949 Mbps
> throughput from 1 Gigabit Ethernet.
> 
> > TIP: If you are out of ports on your main router, add a Gigabit switch. Be
> > aware that while 1GbE switches are very common, there ARE switches that
> > support 2.5GbE (Cat5e), 5GbE (Cat6), and even 10GbE (Cat6a) speeds.
> > Interesting video (youtube.com) on cheaply adding 10 Gigabit Cat6a/RJ45 to
> > your home network.
> > 
> > TIP: If you don't have Ethernet jacks in your house, investigate if your
> > home's phone jacks were wired using 'structured wiring' techniques -- where
> > Cat5 was run to all phone jack locations and back to a single structured
> > wiring panel. If so, you likely have a golden opportunity to convert some of
> > the unused phone line locations to Ethernet.
> 
> b) Best v2: Convert some telephone jacks to Ethernet: It may be possible to
> convert some telephone jacks to Ethernet [§E9] -- everything depends upon what
> cable is 'in the walls'.
> 
> c) Best v3: MoCA 2.5 Ethernet: If you can't use/run Ethernet wire, but have
> access to RG6 (CATV) in every room, MoCA adapters [§E5] may be for you. This
> adds Ethernet, but over existing coax wiring in a home.
> 
> d) Fair: Wireless bridge: Many high-end routers can be configured as a
> 'wireless bridge', meaning that they use wireless to connect to the main
> router, and provide that Internet to ONLY all of the wired Ethernet LAN ports
> (does NOT allow for wireless clients). Best when a 4×4 bridge is connected to
> a 4×4 router. Under good conditions, expect 1500 Mbps PHY and 750 Mbps
> throughput (Wi-Fi 5).
> 
> > Yes, you are trading one Wi-Fi (on the device), for another Wi-Fi (on the
> > bridge). But the PHY speed of Wi-Fi on the bridge is (hopefully) several
> > times faster than Wi-Fi on the devices you replaced. Only use this option if
> > PHY speeds go up 2x or more.
> > 
> > Another easy way to do this is to connect your PC wired/Ethernet to a mesh
> > node that has a 4x4 MIMO wireless backbone. You are then likely exchanging
> > your 2x2 wireless for the mesh network's 4x4 wireless backbone, which can
> > double speeds.
> 
> e) OK to Bad: Powerline Networking/Ethernet: PLC is a very fast and easy way
> to obtain slower, but reliable, speeds. See Powerline Networking [§E8].

2) Improve wireless PHY speeds

> a) Move as many devices as possible from wireless to wired: Avoid spectrum
> usage and contention whenever possible (free up TIME on the spectrum). Is
> there a smart tv that is heavily used for streaming? If so, try to move that
> to a wired connection, freeing up wireless time for those devices that are
> forced to be wireless only (like tablets). See this chapter immediately above.
> 
> b) Try different channels and TEST: Wi-Fi is a shared resource. If you have
> neighbors you may actually be sharing spectrum with them. Especially important
> at night when people come home from work and start streaming.
> 
> > In reality, selecting the 'right' channel is crazy hard to do (well) --
> > because a channel with many access points may actually be the 'most unused'
> > channel, if those access points rarely transfer data (vs a channel with one
> > other access point that is transferring data all the time). Often times you
> > just need to change the channel and test (a lot) [§D].
> > 
> > TIP: The non-DFS 5 GHz channels (at 80 MHz, there are only two) are allowed
> > to operate at a higher power level than the DFS channels. You should see
> > better (download) PHY speeds at distance from these non-DFS channels, but
> > everyone (you, neighbors, etc) want to use those channels. Whereas on a DFS
> > channel, you likely will have the channel to yourself.
> > 
> > So try to find an unused DFS channel, which will result in the channel being
> > all yours! A DFS channel is a must if you are in an apartment/condo building
> > (have a lot of close neighbors).
> > 
> > Inexplicably low PHY speeds?: If you see PHY speeds from your client device
> > (when standing right next to your router) that are strangely 'too low', that
> > is a huge tip off that you may be running into some 'interference' -- try
> > changing Wi-Fi channels on the AP/router.
> 
> 
> 
> c) Several low-PHY 'heavy' users can significantly slow down everyone else:
> Devices operating at a very low PHY speed, and using the channel a lot, can
> slow down an entire Wi-Fi channel. Because what is critical is TIME spent on a
> channel (which increases as PHY speed decreases). Adding an access point [§18]
> (and hopefully on a new unique channel), and greatly improving the PHY speed
> for those devices, can free up time needed by other Wi-Fi clients.
> 
> > Analogy: Imagine a highway (Wi-Fi channel) where a car (smartphone) is going
> > 5 mph (PHY 65 Mbps) when the speed limit is 55 mph (PHY 866 Mbps). That one
> > car will drastically slow down all other cars (Wi-Fi devices) wanting to use
> > the highway (Wi-Fi channel). Adding a new lane (via AP on new channel) to
> > the road not only puts the slow car (smartphone) onto a new lane (channel),
> > potentially causes the car to all of a sudden start driving 55 mph (PHY 866
> > Mbps).
> 
> d) Get a 4×4 MIMO network adapter: If on a PC with 2×2 MIMO, try using a 4×4
> MIMO network adapter (to a 4×4 router). The expectation is that PHY speeds
> will increase (but not quite double).
> 
> > But honestly, with a PC, always try to find a way to use Ethernet wired to
> > your main router.
> 
> e) Upgrade your client device: If your client device is 1×1 MIMO, get a brand
> new client device that supports at least 2×2 MIMO. The expectation is that PHY
> speeds should roughly double (moving from 1×1 MIMO to 2×2 MIMO).
> 
> > Brand new hardware might help: I have also seen 2×2 MIMO devices made in the
> > last year outperform (consistently stay on a higher PHY speed) than 2×2 MIMO
> > devices that are five years old. This can not be overstated. If you are
> > looking for top speeds, always use modern/new devices. Each generation of
> > newer hardware performs just a little bit better than prior generations.
> 
> f) Investigate 160 MHz channels: If your 2×2 client devices support 160 MHz
> channels (this was rare, but it is becoming a lot more common for Wi-Fi 6),
> look into a router that also supports 160 MHz channels. This is not always
> possible, but when possible and there are no DFS channel conflicts (or
> spectrum conflicts with neighbors), this has the 'potential' to double your
> PHY speeds (when compared to 80 MHz channels) -- but for only the few devices
> that actually support 160 MHz channels. But 160 MHz channels require a high
> SNR (you may need to be very close to the router). Also, remember that wider
> channels have slightly less range (than smaller channel widths) - details [§J]
> -- so this works best when very close to the router/AP.
> 
> g) Reserve 5 GHz for true 802.11ac devices: A requirement of any device being
> able to call itself 802.11ac capable is that it must support 80 MHz channels
> in 5 GHz. However, dual-band 802.11n devices can see your 5 GHz SSID and
> connect to it using 20 MHz (or 40 MHz) channels. If that 802.11n device is a
> heavy Internet user, this could slow down all of your 80 MHz channel devices.
> Move that problem device to the 2.4 GHz band SSID. This frees up time on the
> much faster 80 MHz 802.11ac channel for 80 MHz capable devices.
> 
> > Analogy: A 802.11n device operating at 20 MHz in 5 GHz is like a car using
> > one lane of a four lane Interstate and simultaneously preventing the three
> > other lanes beside it from being used (most devices operating in 5 GHz want
> > to use 80 MHz channels).
> > 
> > But if the dual-band 802.11n device is a lightweight when it comes to Wi-Fi
> > usage, then keep it on 5 GHz, as it is 'doing no harm'.
> > 
> > It is all a balancing act -- because if the 802.11n device does not work
> > properly in 2.4 GHz (too slow or unreliable due to congestion) then you may
> > need to keep the device in 5 GHz (eg: a Ring camera, where reliable video is
> > critical).
> 
> h) Update firmware: (1) Make sure that your router/AP is running the latest
> firmware. It is rare, but there have been times that a performance problem is
> found and corrected and new firmware is released that fixes the problem For
> example, this WAN to LAN performance bug (duckware.com). (2) And if on a
> Windows laptop, check for any updates to the Wi-Fi drivers.
> 
> i) Switch Wi-Fi bands: Often times (but not always), PHY speed increases
> dramatically when there is a switch from 2.4 GHz to 5 GHz. This is because the
> SINR for 5 GHz is often much better than the SINR in 2.4 GHz. This improved
> SINR together with wider channels (20 MHz to 80 MHz) often results is a big
> speed improvement. Confirm that your dual-band device is connecting to 5 GHz
> and using 80 MHz channels. If not, consider upgrading your client device
> and/or router.

Remember that everything in Wi-Fi is about TIME on the channel: It all comes
down to 'time' spent on the Wi-Fi channel -- So target the devices that spend
the most time on the Wi-Fi channel, and conversely, don't worry about (ignore)
low channel width, low PHY devices, that don't use Wi-Fi that much (eg:
thermostat). The worst 'time' offenders will be high usage devices with low PHY
rates -- so target those devices first.

> Analyze your client devices that download/upload the most data. They should be
> running at high PHY speeds [§4]; and if not, fix by moving devices around, or
> by adding an access point [§18].

First, did PHY speed increase?: Always check the PHY speed [§4] of your client
devices both before and after an upgrade to confirm that there was an actual
improvement in PHY speeds. Otherwise, there was no point in upgrading.

Next, did throughput increase?: Improving PHY speed is the first step. The
second step is a throughput test [§D] to verify overall Wi-Fi speeds increased.
Why? Because you could have the best PHY speed ever, but if you are sharing that
channel with others (a heavy usage neighbor), overall speeds could go down.





The easy way (restated again): One of the fastest, easiest, and BEST ways to
dramatically improve Wi-Fi speeds to 'the maximum speed possible' is to install
a brand new recommended [§21] router (configured as an access point [§18] and
wired/Ethernet back to the main router) in the same room where you want to
improve Wi-Fi speeds. Wi-Fi works best when the Wi-Fi signal does not have to go
through walls/floors/etc and does not have to travel too far. In Wi-Fi, distance
matters.

17. Wi-Fi Range Extenders


A typical Wi-Fi range extender

So you have a weak Wi-Fi signal on one side of your house. Will a "Wi-Fi range
extender / booster / repeater" solve the problem?

❌ NO, do NOT use a Wi-Fi range extender / booster / repeater. Only use Wi-Fi
extenders as a last resort. Why? Because Wi-Fi extenders are an old technology
and today, there are other much better (and modern) alternatives, like access
points [§18] and mesh networks [§19].

> ★ A GREAT way to see first hand 'what is wrong' with Wi-Fi extenders is
> install one where it is NOT needed like right next to your main router,
> connect to the extender, and observe Wi-Fi throughput speeds cut in half or
> much worse.
> 
> 
> Don't confuse great signal (at range) -- that a Wi-Fi extender does give you
> -- with great speed (at that range). Just because you used a Wi-Fi extender
> and get a great signal 'at range' does not mean that you now have great speed
> at that extended range.
> 
> Instead, you likely traded a more reliable signal for a slow speed (because
> there is still the 'extender to main router' path that packets must traverse)
> -- which in some edge cases, is a good tradeoff to make (sometimes a slow
> connection is better than no connection).

Limited throughput: Don't expect great throughput from a cheap Wi-Fi range
extender, even in the best case of connected to Ethernet and running in Access
Point mode. The typical $50 Wi-Fi extender hardware is simply not capable of
handling top-notch speeds (like a $200 router can).

How Wi-Fi extenders work (aka: the core problem with extenders): Wi-Fi extenders
work by retransmitting every single packet received -- so packets in a "Wi-Fi
extended" network are transmitted TWICE. This only puts even MORE pressure on an
already weak Wi-Fi network.

> Many extenders 'on paper' are a 'zero-sum game'. Because they are (typically)
> located in the middle between the main router and a client device, they
> improve the signal strength between the two paths (extender to router and
> extender to client device) only enough to make up for doubling Wi-Fi
> utilization (packet transmitted twice), and no more. Packets are (often)
> re-transmitted on the same channel, band, and channel width. So not much (if
> anything) is gained speed wise other than possibly a more reliable (but slow)
> Wi-Fi path.
> 
> ★ Where extenders may actually work and help a little bit is if the device
> connecting to the extender is using slow Wi-Fi 4, but the extender backhaul to
> the main router is using faster Wi-Fi 5. Especially when the extender
> implements a MIMO level (eg: 2×2) that is higher then the device you are
> trying to extend (eg: 1×1).
> 
> Also, sometimes extenders can help to 'bend' a Wi-Fi signal around a major
> physical obstruction (eg: elevator shaft).
> 
> But in all cases, if an extender helps a little, then a mesh network would
> work even better.

Wi-Fi extender Placement: Proper placement of an extender is very important in
order to properly maximize throughput. See the Router/AP Placement [§P].

PHY warning: Wi-Fi extenders do improve PHY speed, but the other half of that
story is throughput (because an extender must re-transmit packets and has its
own 'PHY' speed to the main router). Use a speed test [§D] program to confirm
the impact on throughput.

> Namely, if you are standing right next to the Wi-Fi extender, the PHY speed
> will be fantastic, but remember that there is still a hidden PHY speed between
> the extender and main router that you can't see that comes into play and
> limits throughput.

Alternative One -- Improve your main router: First, are you using a modern high
quality recommended [§21] router (4×4 MIMO Wi-Fi 6)? If not, try one. Next,
check router placement [§P]. By improving the router (or placement), you may
improve signal strength enough that you don't need an extender.





Alternative Two -- Add an Access Point: A GREAT way to improve the Wi-Fi signal
strength (and speed!) is to place an access point [§18] (wired/Ethernet back the
main router) exactly where it is needed most. And of course, with the major
caveat that you have the ability to run a new Ethernet cable from your main
router to the new AP.

> So the huge advantage that a wired/Ethernet Access Point has over Range
> Extenders (and mesh networks with a wireless backhaul) is that (1) packets are
> only transmitted ONCE over Wi-Fi AND that (2) the AP can be placed exactly
> where needed most so that client devices near that AP now get maximum Wi-Fi
> speeds (no more need to place extender devices 'mid-way')!

Alternative Three -- Mesh Wi-Fi network: Wi-Fi extenders typically work
(re-transmit) on the SAME Wi-Fi channel/band/width as the main router -- and
that is a major weakness of Wi-Fi extenders (channel utilization). Instead,
consider replacing your main router with a mesh network [§19], where the
'retransmissions' between the nodes in the network happen on a much faster
'backhaul' network (often) on a different Wi-Fi channel/band/width.

> Mesh networks DO also re-transmit packets, but they (virtually always) use a
> 'backhaul' channel/band/width that (1) is often different than the main
> channel and (2) much faster than the main channel (high end mesh networks use
> 4×4 MIMO for the backhaul). A huge bonus is if you can find a way to wire each
> mesh node back to the main mesh node/router via 2.5 Gigabit Ethernet (try to
> avoid using a wireless backhaul).

Price: You get what you pay for - but only to a point: Do you really think that
a $40 to $70 Wi-Fi extender is going to take the Wi-Fi signal from your $200
router and make it that much better at range? No. You get what you pay for. And
if you spend more, then the alternatives discussed above are a better option.

End note: Yes, Wi-Fi extenders may sometimes work for you. But they turn a low
speed unreliable weak Wi-Fi connection into a stronger, more reliable, higher
PHY -- but still low speed -- connection. And that 'low speed' issue is my core
problem with Wi-Fi Range Extenders. My expectation for Wi-Fi is to always have
both great signal strength AND great speed.

If you still really want a Wi-Fi extender: Review "The 3 Best Wi-Fi Extenders
for Your Home Network" (wsj.com), a BuySide article by the Wall Street Journal.



--------------------------------------------------------------------------------


◆ The math behind transmitting a packet twice

Say your Wi-Fi 4 client device is connected to your router using HT20 MCS1
(14.44 Mbps), but it is unreliable. So you add a Wi-Fi 4 extender perfectly
mid-way. That change in distance (in half) causes both the 'router to extender'
and 'extender to client' PHY links to each improve by around 6dB (see why [§I]),
which bumps the SINR levels on both links 'around' 6dB (approx two MCS levels),
to MCS3 (28.88 Mbps). But now all packets need to be transmitted twice, and over
the SAME channel (so contention). What effective speed is obtained depends upon
the channel usage over the two paths...

Same channel: The best-case effective PHY for an overall path composed of 'N'
individual PHY links is (intentionally ignoring the small latency introduced by
each node along the path), operating on the same channel, is:

> 

In other words, for each node in a path, convert PHY Mbps to 'time taken', sum
up the times (because access to the same channel is serialized), and convert
back to a PHY rate.

> For example: With two PHY links, each at 28.88, acting as one link, the
> effective PHY is 1/(1/28.88+1/28.88), or 14.44. The SAME as transmitting one
> packet once at 14.44 Mbps, but hopefully more reliably. Using a 'same channel'
> extender is often a zero-sum game, that just consumes Wi-Fi bandwidth.

Different channels: BUT, if we now switch to an extender with much better Wi-Fi
capabilities (Wi-Fi 5 and 2×2 MIMO) than the client device (at Wi-Fi 4),
operating on a different band and channel (eg: 5 GHz and PHY at 130 or 190),
then the effective PHY changes to:

> 

In other words, when each node in a path is using a unique channel, the overall
path speed is reduced to just the slowest node along the path.

> Example: Using the example above, the overall path speed increases to the
> minimum of 28.88 (no channel contention). And because Wi-Fi 5 PHY speeds are
> so much better than Wi-Fi 4, we can move the extender even closer to the Wi-Fi
> 4 client device (max PHY 72.22), and increase the PHY speeds even more.

With Wi-Fi extenders, the position of devices matters a lot, and best placement
depends upon knowing the PHY speeds of each link, AND which channels/bands will
most often be used!

More Information:
 * Consumer Reports: "Should you buy a Wi-Fi Range Extender"
   (consumerreports.org)
 * Techquickie: DON'T Buy A Wi-Fi Range Extender! (youtube.com)
 * CNET: "Mesh Wi-Fi vs. range extenders: The best option for your home" (CNET
   on youtube.com)


18. Wireless Access Points



Access Point (AP): A Wireless Access Point (WAP), or just Access Point (AP), is
a device that allows a wireless client device, like your
phone/tablet/laptop/etc, to connect to a wired network, like your Internet. And
as a bonus, many access points also have extra LAN ports for connecting any
Ethernet-only devices.

> TIP / Use a router: There are dedicated 'access points' that you can buy, but
> it is often much easier to just buy a normal recommended [§21] router (or
> re-use an old router) and configure it in 'Access Point' mode -- so that
> 'Router' functionality is disabled (as that is still handled by your main
> router), while keeping the 'wireless' AP functionality (and LAN Ethernet
> ports) intact.
> 
> 
> ♲
> 
> 
> Also, using a 'router as an Access Point' is a great way to recycle and re-use
> older routers when the main router is upgraded.

What about DHCP and NAT in an AP: The purpose of an 'Access Point' is to provide
access to all the services running in an existing network, so there is
(intentionally) no DHCP and no NAT in an access point itself. Instead, client
devices connecting to the AP are actually using DHCP and NAT from the 'main
router' on the network.

> A good way to think of an 'access point' is that it is a same as connecting an
> Ethernet cable to your client device -- you are just connecting to Ethernet
> wirelessly via the AP.

An AP is the BEST way to extend Wi-Fi: An Access Point is the BEST way to
'extend' Wi-Fi -- provided the AP can be wired/Ethernet back to your main router
-- because this guarantees (1) the fastest PHY speed possible for client
devices, and (2) the fastest possible consistent backhaul speed back to the main
router.


How to connect a new 'Access Point' to your Main Router -- green lines are
Ethernet cables
EASY: How to configure a Router in 'access point' mode: Here are some
instructions on how to set 'AP mode' for various router brands. Please note that
some vendors use the term 'Bridge' to describe their router running in 'Access
Point' mode.
 * TP-Link: How to configure the TP-Link wireless router as Access Point
   (tp-link.com)
 * Netgear: How do I change my NETGEAR router to AP mode? (netgear.com)
 * Asus: How to set up access point mode? (asus.com)
 * Linksys: How to use a Linksys router as an Access Point (linksys.com)
 * Wavlink: How to set up WAVLINK router to AP(Access point) mode? (wavlink.com)
 * Eero: How to change the Eero router to AP mode (eero.com)
 * Tenda: How to change a Tenda router to AP mode (bitdefender.com)

An example of how easy it is to change a 'router' into an 'access point':


Configuring a TP-Link "Router" as an "Access Point"
NOTE: AP Wi-Fi throughput will be limited to Ethernet LAN port speeds: If you do
go down this path, please be fully aware that Wi-Fi throughput to/from the
access point will be 'limited' to the 'weakest link' speed, which is likely to
be the Ethernet LAN port speed on the main router (likely 1 Gbps).

> But let's also be realistic -- Wi-Fi speeds maxing out at 1 Gbps is a very
> good problem to have. The solution is a main router with all LAN ports at 2.5
> Gbps (or higher).

SSID: All access points in a house are virtually always configured to use the
SAME Wi-Fi name, password, and encryption (eg; WPA2) as your main router. In
this way, you will only ever see a SINGLE Wi-Fi name in Wi-Fi lists on your
client devices no matter how many access points you have installed. Client
devices will automatically connect to the access point (main router or access
point) that provides the 'best' signal.

> You can name access points uniquely if you want to, but then all of those SSID
> show up in Wi-Fi lists on client devices (creates a lot of name pollution).
> 
> The only time I have used a unique SSID name for an access point is when I
> added an AP (on a unique channel) dedicated to service a single wireless Ring
> camera, and I wanted only the camera -- and no other clients in the home --
> using that access point.

Setup tip: Changing the SSID on a 'router as AP' is the LAST step. In other
words, first power on the new router, connect into your network and confirm it
works. Next, change the operation mode to "Access Point", and confirm that
works. Finally, as a last step, change the SSID/password to be the same as your
main router. Why in that order? Because if anything 'goes wrong' during the
setup process, you want to ability to easily connect directly to (only) the new
router/AP you are setting up. And having a unique SSID until that last minute
allows you to more easily do that.

Channels: When you have more than one AP in a house, manually configure the
channel used in each AP (don't use 'auto' channel selection). You want each AP
to have its own dedicated channel (not overlapping with other AP channels), so
that all AP can operate simultaneously.
 * For 2.4 GHz, only use non-overlapping channels 1, 6, or 11.
 * For 5 GHz, only use channels that fall into different 80 MHz channels (for
   example, 36, 40, 44, 48 are all in the same 80 MHz channel). Also, try using
   DFS channels first.
 * For 6 GHz, only use channels that fall into different 160 MHz channels.







Wired Ethernet backhaul: Access points must plug into your Ethernet [§E1]
network somewhere -- because they provide an interface from 'wireless' to
'wired' for your client devices. The big advantage of using an access point is
that they can then be placed precisely where they will do the most good -- as
Ethernet cables can be up to 100 meters (328 feet) long.

> For example, your 'main router' is in your living room, but you then also have
> an access point (wired/Ethernet back to your main router) in your home office.
> 
> In your office, you will get the fastest Wi-Fi speeds that are possible for
> your client devices, and because there is Ethernet back to the main router,
> packets are never re-transmitted twice (as they would be with extenders or
> mesh).
> 
> TIP: What Ethernet port should be used (on a router configured as an access
> point) to wire/Ethernet the 'access point' back to your wired network -- the
> WAN port or a LAN port? Technically, it should not matter at all -- as every
> port 'should' work equally well. However, be aware that some routers
> (especially older ones) do have some bugs in properly routing ALL packets
> across the WAN to LAN interface (when the router is in AP mode). So, AFTER
> configuring a router in "Access Point Mode", consider switching the Ethernet
> cable from the WAN port on the AP to a LAN port on the AP (and no longer use
> the WAN port).

Adds more Wi-Fi capacity: One of the huge advantages that access points have
over mesh (or a single main router) is that they allow you to add more 'overall'
Wi-Fi 'bandwidth' when Wi-Fi is over-utilized. Instead of everyone in the house
connected to Wi-Fi on a SINGLE channel, that Wi-Fi usage is instead spread over
multiple access points, as long as each access point is on its own unique
channel (and as long as people using Wi-Fi are spread throughout the house).
Each unique Wi-Fi channel can operate independently and concurrently.

> At one house, I added an access point into the 'kids playroom' -- so that any
> streaming or game play in (or near) that room will go through that access
> point (and not impact Wi-Fi in the rest of the house). Plus, everyone in the
> kids playroom obtains maximum PHY speeds, which results in the fastest Wi-Fi
> speeds possible.
> 
> The kids can max out their Wi-Fi channel (to 100%) without impacting Wi-Fi
> speeds for any other users in the rest of the house.

TIP: In your main router, configure a 'fixed' IP address for your new access
point: Your new access point will dynamically obtain an IP address from the main
router. But since that IP address can change over time, it is much easier to
manage the access point if the IP address never changes. So in the main router,
assign a fixed IP address for the AP. When a router is configured as an access
point, and you are connected to the AP's wireless SSID, most router companies
provide a URL that automatically connects to the AP. Here are some of the most
common ones:
 * TP-Link: tplinkwifi.net
 * Asus: www.asusrouter.com
 * Netgear: routerlogin.net
 * Tenda: tendawifi.com

> WARNING: But 'ping' these names from a Command Prompt window to 'find' the IP
> Address. If you type these names into Chrome (to access the router's web
> interface), that may FAIL. You may NOT get the correct name resolved if Chrome
> is using 'secure DNS', which bypasses normal DNS (where a router maps these
> special names to itself).

A real-world example: This house (below) already had Internet for the main house
(cable modem + 4×4 wireless main router; green "INET" box) and a single Ring
camera (green dot). The owner wanted to add four more Ring cameras around the
entire property (blue dots). Wi-Fi needed to be extended to the garage and
storage shed. The BEST solution was to add a 4×4 AP (access point; blue "AP"
box) in the garage (using a channel different than the main router),
wired/Ethernet back to the main router (dotted blue line).

> Access Point on non-overlapping 5 GHz channel 58 -- Main Router on 5 GHz
> channel 122

Running the Ethernet cable to the new AP admittedly was a pain through a partial
basement and a very tight crawl space, but the end result -- the 'best' Wi-Fi
possible for all cameras -- was well worth it.

> The significant advantage this 'Access Point' configuration has over many
> 'Mesh' systems is that there is NO overlap in Wi-Fi channel usage between the
> 'INET' and 'AP' nodes on any band, meaning that both nodes operate 100%
> independently -- both nodes can Tx/Rx at the exact same time (whereas for many
> mesh systems, only one node can be active at a time, because mesh nodes use
> the same channels).

Your choice: Once you have Ethernet wired into your house, you have a choice to
make. Use Access Points or use a Mesh Network (which also work great with a
wired/Ethernet backhaul). There is no right or wrong answer -- it all depends
upon your needs and requirements (and budget).

What about business access points?: If they work for you, great, go for it. But
if I need to add two AP's into a home (an office and a kids game room), that:
(a) has no existing PoE switches, (b) needs to plug in multiple wired devices at
each AP location, and (c) requires 1.5+ Gbps Wi-Fi throughput to the main router
-- then a 'router as AP' is very attractive, because this eliminates purchasing
(for a business AP solution):
 * a more expensive 'premium' product to fully support the throughput
   requirement
 * a 'controller' (or cloud service rental) to manage the business AP's -- and
   higher end business AP's require a controller
 * multiple PoE switches or adapters to power the business AP's and/or regular
   switches

Having a single independent device that quickly solves the problem -- and does
not need to be mounted on the ceiling -- is a very attractive option to many.

19. Mesh Wi-Fi Network Systems
Ultra Convenient: Mesh Wi-Fi Network Systems are ALL about incredible
convenience (super fast setup) -- but are NOT (necessarily) about top notch
performance.



Wi-Fi EasyMesh -- Wi-Fi Alliance



What is a Mesh Network? Just place mesh nodes around your home and power them
on. Your client devices connect green lines; right to the nearest mesh node and
all mesh nodes communicate with each other almost always wirelessly, but can be
wired/Ethernet; red lines; right to access the Internet One mesh node must be
physically connected wired/Ethernet to your modem/Internet.

The price of mesh: So, mesh networks do work, but that setup convenience comes
at a price -- (1) in actual cost, as they can be very expensive, (2) in reduced
MIMO level support to client devices, and (3) in moderate (not top notch)
performance when packets must traverse through multiple mesh nodes.

> Mesh networks often are only 2×2 MIMO for the backhaul and 2×2 MIMO to client
> devices. And if 4×4 MIMO is supported, most often that is only for the
> backhaul, with client device communication still only at 2×2 MIMO.
> 
> 
> 
> "An access point always delivers better performance than a wireless mesh
> satellite of the same Wi-Fi grade." -- from Dong Knows Tech (dongknows.com).
> 
> 
> A wired/Ethernet Access Point will ALWAYS have a significant guaranteed
> performance advantage over a wireless mesh network. Why? Because mesh
> satellite nodes: (1) transmit every packet twice (or more) wirelessly, and (2)
> all use the SAME backhaul channel, creating a bottleneck when multiple mesh
> nodes actively need to use Wi-Fi at the same time.

Mesh Networks: Some of the more popular and capable 3-pack mesh networks are (in
alphabetical order; price ballpark value as of September 2024; 2-packs also
available). These do not support the 6 GHz band, but in exchange they include a
dedicated backhaul channel (in 5 GHz):

> Asus ZenWiFi AX XT8 -- Wi-Fi 6 (2.5 Gbps WAN; 4×4 backhaul)
> $550
> AX6600 -- 2.4GHz 2×2 574 Mbps (40 MHz) -- 5GHz-1 2×2 1200 Mbps (80 MHz) --
> 5GHz-2 4×4 4800 Mbps (160 MHz) specs Netgear Orbi RBK853 -- Wi-Fi 6 (2.5 Gbps
> WAN; 4×4 backhaul)
> $800
> AX6000 -- 2.4GHz 4×4 1147 Mbps (40 MHz) -- 5GHz-1 4×4 2400 Mbps (80 MHz) --
> 5GHz-2 4×4 2400 Mbps (80 MHz) specs

And then you have mesh systems that DO support the 6 GHz band (and 2.4 GHz and 5
GHz), but in exchange, they have NO dedicated backhaul channel. This means these
mesh systems can behave like a mesh network, or a Wi-Fi extender, depending upon
which band your client device is using, because they are using a shared
backhaul:

> Amazon Eero Pro 6E -- Wi-Fi 6E (2.5 Gbps WAN; 2×2 backhaul)
> $549
> AXE5400 -- 2.4GHz 2×2 574 Mbps (40 MHz) -- 5GHz 2×2 2400 Mbps (80 MHz) -- 6GHz
> 2×2 2400 Mbps (80 MHz) specs Google Nest WiFi Pro -- Wi-Fi 6E (1 Gbps WAN; 2×2
> backhaul)
> $320
> AXE5400 -- 2.4GHz 2×2 574 Mbps (40 MHz) -- 5GHz 2×2 2400 Mbps (80 MHz) -- 6GHz
> 2×2 2400 Mbps (80 MHz) specs Linksys MX8503 Atlas -- Wi-Fi 6E (5 Gbps WAN; 4×4
> backhaul)
> $400
> AXE8400 -- 2.4GHz 4×4 1147 Mbps (40 MHz) -- 5GHz 4×4 2400 Mbps (80 MHz) --
> 6GHz 4×4 4800 Mbps (160 MHz) specs TP-Link Deco X75 Pro -- Wi-Fi 6E (2.5 Gbps
> WAN; 2×2 backhaul)
> $400
> AXE5400 -- 2.4GHz 2×2 574 Mbps (40 MHz) -- 5GHz 2×2 2400 Mbps (80 MHz) -- 6GHz
> 2×2 2400 Mbps (80 MHz) specs

Wi-Fi Usage: Mesh networks typically use a wireless backhaul which
doubles/triples Wi-Fi usage (every packet must be transmitted two/three times).
So increased Wi-Fi usage is one tradeoff.

> But unlike Wi-Fi extenders [§17] (which typically operate on the same channel
> as the main router), 'better' mesh networks often operate the wireless
> 'backhaul' on a different Wi-Fi channel (and often a different Wi-Fi band),
> which results in mesh networks being much faster then Wi-Fi extenders.
> 
> However, notice what happens when multiple mesh nodes all need to use the
> wireless backhaul network all at the same time -- they must do so serially.
> And very likely at speeds well below Ethernet speeds.
> 
> This is why a wired/Ethernet backhaul is greatly preferred (faster and
> operates concurrently) over a wireless backhaul. You just can't beat the speed
> of a dedicated copper wire.
> 
> 
> 
> WARNING: Watch out for inexpensive 'Mesh' systems with only two Wi-Fi bands
> and NO backhaul. They claim to be 'Mesh', but without a fast dedicated
> wireless backhaul in place, these mesh systems are little more than just
> expensive Wi-Fi extenders!


Mesh can be VERY expensive



Fast Roaming: A slight advantage that mesh networks have over access points is
something called 'Fast Roaming'. If you find yourself walking around the house a
lot and require the 'fastest' switch possible between wireless access points
(like in 50ms instead of 500ms), then mesh systems 'on paper' have a 'switching
nodes' speed advantage over access points, as mesh systems usually implement (1)
802.11k "Neighbor Reports", (2) 802.11v "BSS Transition Management Frames" and
(3) 802.11r "Fast BSS Transition". More Information (microsoft.com).

> Also, some mesh systems use "proactive 802.11k/v client steering" that cause
> client devices to switch to a different mesh node before a client would
> normally decide by itself to switch mesh nodes.

Mesh networks wireless 'same channel' problem: I am not impressed with the
channel usage that I see in some mesh network systems -- where the SAME Wi-Fi
channel is used for both front-haul and back-haul and once you start to label
all channel usage, the contention problem with this setup becomes obvious:

> Analogy: Using the same channel for everything is like building a six-lane
> highway but then forcing all cars to SHARE just a single lane -- is very
> inefficient! Instead, it makes a lot more sense in capacity and throughput to
> use access points, where each AP is operating 100% independently -- on a
> unique non-overlapping front-haul Wi-Fi channel AND using a wired backhaul.
> 
> And even if your mesh system does have a dedicated 'backhaul' channel
> different than the 'front-haul' channel (not all mesh do), there is still only
> a single backhaul channel that all mesh satellite nodes must still share. And
> on a busy network, that means contention.
> 
> TIP: Use wired/Ethernet backhaul with Mesh whenever possible: The better mesh
> networks add an Ethernet jack on satellite mesh nodes, which allows for a
> wired/Ethernet backhaul. So, whenever possible, plug mesh network devices into
> an Ethernet network (wired/Ethernet back to the primary mesh node).



My personal preference (a great alternative to mesh): When I setup a Wi-Fi
network, I want to (1) have 4×4 MIMO to client devices, (2) eliminate Wi-Fi
backhaul usage, (3) minimize cost, and (4) most importantly, maximum throughput.
Which is why my preferred network configuration is a 4×4 MIMO main router, and
as needed, 4×4 MIMO access points [§18] (on unique Wi-Fi channels)
wired/Ethernet back the main router.

> Every 'front-haul' AP node is on a unique channel, and the 'back-haul' is
> Ethernet. AP's never compete with each other for wireless spectrum.
> 
> Admittedly, running Ethernet cable is not always possible or cost effective
> (may need to pay someone else to install it). But, it is cost effective for me
> because I am willing to do all of the installation work myself. For me, this
> maximizes Wi-Fi speeds (performance) and minimizes costs.

The bottom line: The choice is yours. Maybe try a mid-range recommended [§21]
4×4 MIMO router properly placed [§P] first and if that does not provide the
Wi-Fi coverage you need, then consider either (1) adding an access point into
the network (IF you can install it wired/Ethernet back to the main router) or
(2) switch to a mesh Wi-Fi network. See 'the easy way' at How to improve Wi-Fi
speeds [§16].

If you still really want a mesh network: Review "The Best Mesh Wi-Fi Systems for
Your Home" (wsj.com), a BuySide article by the Wall Street Journal.

> Regarding this WSJ article and Eero: I also have measured poor Wi-Fi
> performance with Eero and also find it unacceptable that even basic Wi-Fi
> settings (like channel number or SSID name per band) can NOT be changed in an
> Eero.

Wi-Fi EasyMesh: EasyMesh is a new industry "mesh" standard being promoted the
Wi-Fi Alliance to standardize how mesh nodes communicate with each other. The
huge advantage will be that you can mix and match products from different
vendors (instead of being forced to buy all mesh nodes from the same vendor).
Only time will tell to see how that plays out (if most vendors really will
implement it).

> Look for mesh network systems that support EasyMesh. Avoid mesh systems that
> refuse to support EasyMesh.

More Information:
 * Compare Eero models (eero.com) -- "Radio Frequency" section lists MIMO level
   for each band
 * More details on Eero (eero.com)


20. A Reality Check
◆ Advertised router speeds are pure fiction: Consider this claim from a
manufacturer: "enjoy combined wireless speeds of up to 7.2Gbps". The speeds
advertised for routers are pure fiction because they are based upon various
maximum capabilities added together, and for hypothetical client wireless
devices that DO NOT exist. Can you name any laptop computer, smartphone, or
tablet that has 4×4 MIMO Wi-Fi?

> Router manufacturers' wireless speed claims are just like a used car salesman
> trying to convince you that a Formula 1 racecar will reduce the time of your
> morning commute to work. But what really matters more is the actual speed you
> can achieve for the roads you are driving on (your client device), NOT the
> 'maximum' potential vehicle speed (the router).

◆ Most wireless client devices are 2×2 MIMO: The capabilities of YOUR wireless
device (and not the router) almost always limits speeds, and today, that limit
is 2×2 MIMO [§7]. The reason for lack of 3×3 and 4×4 MIMO is due to the negative
impact increased MIMO has on battery life.

◆ 2×2 MIMO on client devices is actually good enough (for most people): You can
expect throughput of 455 Mbps (±45 Mbps) on a 2×2 MIMO client device at a medium
distance. Until there is some compelling app that actually requires throughput
greater than 455 Mbps, you can bet MIMO will remain at 2×2 on these mobile
devices (it conserves battery).

◆ Some people accessing 1 Gbps Internet -- using only Wi-Fi -- are wasting their
money: It is possible to get Wi-Fi throughput to 1 Gbps -- but only by being
very close to a Wi-Fi AP. The 'weakest' link is often Wi-Fi (distance) and your
client device (capabilities), and not the router or ISP -- meaning that with
Wi-Fi speeds below 1 Gbps, the 1 Gbps ISP speed is unattainable. Run a
throughput test [§D] to see if your Wi-Fi is actually reaching 1 Gbps.

> But as soon as you plug a PC into gigabit Ethernet and connect to your router,
> yes, you will easily obtain 1 Gbps speeds (technically, 949 Mbps throughput,
> max) -- and get what you are paying for.
> 
> I have been in several vacation rental properties where the owner brags that
> they have Gigabit Internet, only to arrive at the property and discover that
> Wi-Fi throughput is around 200 to 400 Mbps (because the main router, with
> Wi-Fi, is located several rooms away from the main living area, or is too far
> away from the provided work desk). The only way to get reliable Gigabit
> throughput wirelessly is to be very close to an AP (plus it helps to have a
> router and client device that both support 160 MHz channels). OR, the owner is
> paying for 1 Gbps Internet, but then only has a Comcast XB6 gateway with
> maximum Wi-Fi throughput right next to the router around 700 Mbps (again, the
> weakest link in the chain is the speed you get).

◆ Wi-Fi 5 is good enough for 400 Mbps Internet: For everyone with ISP Internet
speeds 'around' 400 Mbps (or less), Wi-Fi 5 is actually good enough. But if you
do have Wi-Fi 6 router/devices, great, even better.

> And frankly, it is actually pretty hard for most people to notice any speed
> difference between 100 Mbps and 1 Gbps for 'typical' Internet usage (email,
> web browsing, streaming, etc). Of course, if you are downloading a huge file
> you will notice, but for what most people use the Internet for, they simply
> won't notice the speed difference.

◆ Client PHY speed is the key: The speed at which your wireless devices connect
to a router is called the PHY speed and it is easily found (see chapter far
above [§4]). That PHY speed is what you should look at (in all your wireless
devices) to evaluate if a new router is helping you to achieve any faster speeds
(or not).

> And of course, PHY speed only indicates potential speed. You should then run
> throughput speed tests [§D] to confirm that the Wi-Fi channel performs well
> (not sharing bandwidth with others, or experiencing interference).

◆ Beamforming/diversity (extra antennas) really works: The one advanced feature
in Wi-Fi that really does work well is beamforming/diversity. A wireless device
connected to a 4×4 MIMO router with beamforming/diversity can expect better
speeds at a greater distance (than a non-beamforming router, or even a 2×2
router). But how can you tell that it is helping? As per above, by examining the
PHY speed [§4] at which 'at range' devices connect to your router, and by
running throughput speed tests [§D].

◆ MU-MIMO is mostly hype: You can get it to work in lab situations, but in the
real world -- no, it does not work very well today (will it in the future?).
There are just too many caveats and 'gotchas'. Do NOT go out of your way looking
for this feature, but if it just happens to come with a new router, fine.

◆ WAN port speed limit: Some new routers are now claiming 10 Gbit wireless
speeds (an aggregate speed you can never achieve). BUT the WAN port on the
router is only 1 Gbps. Hilarious. Because what do you think your maximum speed
to the Internet is? Your 1 Gbps link to the WAN. Always look for the weakest
link.

> Other crazy examples of this (from the past) are the Netgear R6120
> (netgear.com), Linksys E5400 (linksys.com), and TP-Link Archer A54
> (tp-link.com) -- all with advertised wireless speeds "up to 1200 Mbps".
> However, ALL Ethernet ports (both LAN and WAN ports!) are only Fast Ethernet
> -- meaning 100 Mbps maximum speed to/from the LAN and to/from the Internet.
> 
> Notice that many higher end Wi-Fi 6/6E/7 routers now offer 2.5 Gbps (or
> higher) WAN ports.

◆ Many 'enterprise' installations only use 20 MHz channels: You can almost
always get by with an 80 MHz (or even 160 MHz) channel at home, but most
'enterprise' installations still only use 20 MHz channels, and that
reduces/limits throughput (a PHY of 173 Mbps to a 2×2 MIMO Wi-Fi 5 client with
real throughput 'around' 100 Mbps is typical).

> The main reason for using only 20 MHz channels is that businesses often
> install a bunch of access points, and need each AP operating on a unique
> channel, that is not too close in frequency to that used by a neighboring
> access point. Otherwise, performance would tank with too much adjacent channel
> interference. And in 5 GHz there are simply not enough 80 MHz or 40 MHz
> channels to effectivity provide separation. But there are enough 20 MHz
> channels.




◆ Don't overlook Ethernet: Wired Ethernet (§E1) is still the gold standard of
speed and reliability. It is not always easy or realistic, but whenever
possible, always use Ethernet. Try to run Ethernet to every device with an
Ethernet jack (smart tv's, Blu-rays, game consoles, Chromecast, desktop
computers, etc). I did this in one house and Wi-Fi usage plummeted (and the only
devices left on Wi-Fi were low bandwidth wireless only devices -- like smart
thermostats and tablets). This left Wi-Fi in the house wide open for devices
that can only connect via Wi-Fi.

> Ethernet all of a sudden looks pretty cool when every smart TV in the house
> can RELIABLY stream 4K at the same time because NO Wi-Fi is being used!
> 
> Also note that most (all?) HDTV's have only a Fast Ethernet (10/100) port, and
> not Gigabit, as a cost savings measure -- since 100 Mbps is more than 'fast
> enough' for 4K streaming.

R7800 to Galaxy S6 long range

◆ YOUR client device often limits Wi-Fi range (not the router): Client devices
almost always transmit at power levels well below that of the maximum permitted
-- whereas an AP/router may transmit at much nearer to the maximum power level
permitted. The two key reasons why clients limit transmit power is: (1) to
improve battery life, and (2) most client Wi-Fi is download (AP/router transmit
power), not upload (client transmit power). Details [§H].

> This observation was confirmed by using the MCS Spy tool (duckware.com), which
> shows that clients are often transmitting (to the AP/router) at lower MCS
> levels (Rx MCS seen right) than the AP/router (Tx MCS seen right) is
> transmitting to the client.
> 
> Have you ever tried to connect to a weak Wi-Fi network, only to have your
> client device complain that it failed to connect? And then you wonder, 'but my
> device can clearly see the Wi-Fi network name, so why is a connect failing'?
> You move slightly closer to the AP/router and your device connects? This is
> almost certainly caused by the client transmitting at lower power levels than
> the AP/router (is transmitting).



◆ Increased range is actually NOT a good thing: I was reading a post by someone
in a forum exclaiming the merits of some new router being installed (at an
airport) because range was twice that of the prior Ruckus AP's. That increased
range might be true, but counterintuitively, increased range in dense (lots of
clients) environments is absolutely NOT a good thing. And once analyzed, the
reason makes sense. Everyone on an AP shares that AP's Wi-Fi bandwidth. Period.
Which is exactly why you want shorter Wi-Fi range AND more AP's in dense
environments -- so fewer people per AP (on different channels) means INCREASED
Wi-Fi bandwidth per person.

> TIP: The same principle applies to large homes, where you want everyone evenly
> connected to several different AP's, not everyone connected to one AP. When
> each AP is on its own unique (and non-overlapping) Wi-Fi channel, this
> 'creates' more bandwidth (as unique Wi-Fi channels all operate independently
> and concurrently).


Unrealistic test setup

◆ Beware reviews testing unrealistic 'ideal' situations: I have seen too many
online reviews test to a new router that is only inches away (see example right)
, or 'line-of-sight' in the same room as the router. Of course the router should
always get maximum speeds in those situations! But what really matters is the
performance of the router in YOUR real world environment, which almost always
means the signal must go through walls, floors, deal with local interference,
etc. -- which can significantly impact throughput. Plus, YOUR client devices
will likely limit maximum speed capabilities, not the router.

> Online reviews often do demonstrate 'maximum' router capabilities. But you may
> experience something different with your client devices. If you want those
> fast speeds, now you know how to obtain them -- with the router, or an access
> point [§18] installed right next to you.

◆ Most vendors implement Wi-Fi using the SAME chipsets: There are NOT a ton of
companies making Wi-Fi chipsets. Instead, there is only a very small handful of
companies. The two giants in the consumer router Wi-Fi chipset game are Broadcom
and Qualcomm (with over 50% market share). Other players are Intel, MediaTek,
etc. All chipsets are widely used. So baring some major bug, all AP/routers
within a 'class / generation / wave' are comparable. So other factors, like
vendor firmware/software, quality of support, antenna design, etc, are the
differentiator.

> And it even goes further. For example, Qualcomm makes 'reference designs'
> (actual working products), allowing other companies (like Netgear) to then use
> the reference designs to make their own routers. An example of this is the
> Netgear Nighthawk R7800, which is just a "Qualcomm Atheros AP161 reference
> board".
> 
> But also, each new wave/generation of hardware/chips does seem to perform just
> a little bit better than prior generations. So sometimes, just being 'newer'
> can be better.

64-QAM 3/4, 20 MHz Wi-Fi versionSpeed Wi-Fi 3 (802.11a/g)54.0 Mbps Wi-Fi 4
(802.11n)65.0 Mbps Wi-Fi 5 (802.11ac)65.0 Mbps Wi-Fi 6 (802.11ax)77.4 Mbps Wi-Fi
7 (802.11be)77.4 Mbps

◆ Fully understand where Wi-Fi speed increases are coming from: Look at the PHY
speed tables [§F] for the same 64-QAM 3/4 'modulation and encoding scheme' (MCS)
for the same 20 MHz wide channel between Wi-Fi versions, and you will find only
small modest speed increases (table right). The core 'encoding' techniques in
Wi-Fi have actually NOT changed that much. Instead the dramatic increases in
speeds in Wi-Fi are coming from:
 * MIMO: 2x (2×2 MIMO [§7]) or 4x (4×4 MIMO)
 * Increased channel width: over 2x (HT40), over 4x (VHT80), over 8x (HE160), or
   over 16x (EHT320)
 * Higher MCS levels: 2x (Wi-Fi 4 to Wi-Fi 6, due to higher QAM encoding)

The result can easily be an 18x to 36x to 72x improvement in speed from just
these other factors. And if you are operating your device 'at distance', then
these factors reduce down to just (a) MIMO level and (b) channel width (as
higher MCS levels can only be used when close to the router/AP)!





◆ Distance from AP really does matter, A LOT (you WANT full bars): Using the
TxRate throughput measurement tool [§D], when my laptop connects to my
centralized main house Wi-Fi 5 router (one room away; 5 GHz; 80 MHz channel;
866.6 PHY max), I get 'around' 390 Mbps of real throughput (left graph below).
And if I move farther away, throughput drops to around 200 Mbps due to more
floor+wall obstacles. But when connecting to the AP in the same room, I get
'around' 730 Mbps of real throughput (right graph below). Throughput is MUCH
faster when using the closer AP. In Wi-Fi, distance matters!

> Throughput to Main Router one room away - GOOD
> Throughput to AP in the same room- GREAT

Admittedly, very few people will actually notice this speed difference, but if
your goal is maximum real Wi-Fi throughput (ie: minimum latency), then you now
know how to accomplish that.

When I repeat this speedtest exercise with Wi-Fi 6 client devices and Wi-Fi 6 AP
(using 160 MHz channels), I get 'around' 1800 Mbps of real throughput (in
TxRate) to the 'same room' AP.

◆ Wi-Fi works best for everyone else when YOU are physically close to an Access
Point: It is all about time spent on the channel. Wi-Fi does work through walls,
floors, etc., and at distance, but often at reduced speeds and reduced channel
widths (more time on the channel). However, Wi-Fi works BEST -- for everyone --
when YOU are physically close to an access point.





If you want the best Wi-Fi possible (for you and others), find a way to add an
access point [§18] (wired/Ethernet back to your main router) in the same room
where you spend most of your time -- like in a home office. Why is this best?
Because:
 * You get the fastest Wi-Fi PHY speeds possible (real throughput) - directly
   benefiting you
 * This frees up TIME on the channel for all other Wi-Fi devices - indirectly
   benefiting everyone else

And the need to be close to an AP is critical if you have Gigabit Internet. To
actually achieve that speed wirelessly, you want a Wi-Fi 6/6E access point in
the room with you so that your client device can obtain an excellent PHY speed
[§4] with a real throughput of just over 1 Gbps. You must be close to an AP if
you want to take advantage of Wi-Fi features that require a high SNR:
 * higher QAM encodings like 1024-QAM and 4096-QAM
 * wider channel widths [§J]

◆ You want the fastest Wi-Fi speeds possible, even with 'slow' Internet: By
minimizing the amount of time your devices are on Wi-Fi (by being very fast),
that frees up TIME for other much slower client devices at range operating on
the same channel to use Wi-Fi (like a Wi-Fi camera).

◆ Your physical location impacts your Wi-Fi experience: Test a Wi-Fi router and
client device at your location and obtain great/good/fair/horrible results and
everyone else will experience the same results from that router, right? No, not
necessarily. Some people will see different results, but how is that possible?
The root cause is often a very different 'noise floor' (spectrum usage) between
the two physical locations. See the Noise Floor [§K] appendix. Stated another
way, channel usage all around you will impact how well (or not) your Wi-Fi
works.

> For example, you may easily obtain QAM-1024 data rates (because you experience
> a low noise floor), but your friend with the same setup (but at a different
> location) may never get QAM-1024 data rates (because they experience a high
> noise floor) due to adjacent channel interference [§P], or some other factor.
> 
> The spectrum usage all around you (from ALL of your neighbors) from Wi-Fi, and
> other sources, will impact how you perceive and experience Wi-Fi at your
> location. Don't assume that everyone else experiences Wi-Fi the same way that
> you experience it.

◆ Wi-Fi will never fully compete with wired Ethernet: Ethernet uses a dedicated
full-duplex channel for every device, so each device individually obtains the
maximum speed. Whereas Wi-Fi uses a shared half-duplex channel, where all
devices SHARE a single common maximum speed.

◆ You get what you pay for: Most dirt-cheap inexpensive routers and access
points will have performance (throughput) limitations. Don't expect top level
sustained performance like 80% of PHY from an entry level router/AP. So keep
this in mind when looking for new equipment. But if you don't need top level
performance right next to the router you are OK with only a couple hundred Mbps
max, then a cheap router could be a great value, for you.

21. Recommendations
Be very critical: There is no point in replacing your router/AP if PHY speeds
to/from your wireless devices do NOT improve by at least some reasonable amount.
So, be very critical. Take note of client PHY speeds [§4] and throughput speed
tests [§D] before and after an upgrade. If you see an improvement in speeds you
wanted, great, job accomplished! However, if not, then you have to ask the
serious question: Did you just spend a bunch of money and not get the
benefit/improvement you needed/wanted?

> When updating a router, verify that client device PHY speeds (and throughput
> speed tests) actually increase, especially for devices 'at range'.

When buying a new router, what features do you need to best support 1 Gbps
throughput wirelessly?

The attainable goal: The goal is to find a the best Wi-Fi speeds for the best
price (namely, the best value). And today that means Wi-Fi throughput 'around' 1
Gbps for around $200. To achieve this, you will need modern Wi-Fi 6 (or later)
devices and be close to the router/AP...

> For example, if you have high-end Wi-Fi 6 client devices supporting HE160 and
> upgrade your VHT80 Wi-Fi 5 router to an HE160 Wi-Fi 6 router (like those
> recommended below), I would expect to see maximum PHY speeds increase from 866
> Mbps to 2400 Mbps and over 1 Gbps throughput (standing right next to the
> router), and then only modest speed increases over Wi-Fi 5 'at distance'.
> Warning: Apple Mac/iPhone/iPad client devices don't support HE160 in Wi-Fi 6
> in the 5 GHz band!

The features you WANT in a new Router/AP: Virtually all Wi-Fi devices today
(laptops / tablets / smartphones / smart tv's / etc) are STILL only 2×2 MIMO (at
best; some are even still at 1×1). And THAT limits the PHY speed at which those
devices will connect to any router/AP (not the max speed of the router). To
maximize range and throughput from these devices, get a 'whole house' router/AP
with a minimum of:
 * 4×4 MIMO - increases signal quality, reliability, and range for all 2×2 MIMO
   client devices. Do not buy a 2×2 or 3×3 MIMO router. You really do want a 4×4
   router.
 * DFS channels - you need a router that supports ALL 5 GHz 80 MHz DFS channels,
   to increase the likelihood of NOT sharing a channel (and therefore bandwidth)
   with a neighbor, and needed for 160 MHz channel support.
 * HE160 (160 MHz channels) - you need a router that supports HE160, because
   support for 160 MHz channels in Wi-Fi 6 client devices is now becoming very
   common (so you want to take advantage of that).
 * Multi-Gigabit Ethernet - you want a router with a multi-gigabit WAN port (eg:
   2.5 Gbps, or higher). If your Internet speeds are not above 1 Gbps, then a 1
   Gbps WAN port is just fine (and expect max throughput around 949 Mbps).
 * A mid-range Wi-Fi 6 (802.11ax) router - this is the best VALUE today
   (September 2024) -- expect to spend 'around' $200 plus/minus $50. Avoid
   routers below $150 and above $300.
 * beamforming - improves signal strength, which increases the range at which
   devices stay at fast speed -- should be a standard feature of any new
   mid-range Wi-Fi 6 router.
 * Access Point Mode - look for a router that supports "access point mode" (most
   routers will). Because when you do upgrade to a newer version of Wi-Fi years
   from now, you may want to reuse the old router as an 'AP' in your new network
   (and not have the old router sit on a shelf unused).
 * Wi-Fi Certified - not absolutely required, but if the router is certified,
   this guarantees "interoperability, security, and reliability." Product
   Search. Also, watch out for routers certified to a lower specification than
   expected (eg: a 802.11ax router certified for only 802.11ac is a red flag).

Critical features: The most important criteria to look for in a new router are
(1) 4×4 MIMO [§7], (2) DFS channel [§14], and (3) HE160 support. This eliminates
so many entry-level routers from consideration, as there are a LOT of cheap
non-DFS + 2×2 MIMO + HE80 routers out there.

> Also, it is easy to find these criteria in high-end (rather expensive)
> routers, but the challenge is to find that combination in an affordable
> mid-range router!

Purchase TIP: Always buy electronics from a vendor with a great return policy.
For purchases on Amazon, look for "Sold by Amazon.com" under the "Add to cart"
button, because Amazon itself has a great return policy. Third party sellers on
Amazon may have a very different (and much more restrictive) return policy.

> And don't forget to keep an eye out for clickable 'coupons' (example seen
> right) on Amazon product pages. You do NOT get the discount unless you click
> on the coupon.

★ Wi-Fi 6 Router Recommendations: Find a mid-range router that supports 4×4 MIMO
for both the 2.4 GHz and 5 GHz bands -- that is 'best' (otherwise, if only for
the 5 GHz band, that is 'good'):

> Best ($150-$230) Best ($220-$300) Good ($150-$250)
> TP-Link
> Archer AX80
> 
> 
> ✓ WAN: 2.5 Gbps Ethernet
> ✓ Max Throughput: 1.8+ Gbps
> ✓ 2.4 GHz: 4×4 MIMO
> ✓ 5 GHz: 4×4 MIMO
> ✓ DFS channels
> ✓ HE160
> Now $150 -- 35% off
> 
> 
> Asus
> RT-AX88U Pro
> 
> 
> ✓ WAN: 2.5 Gbps Ethernet
> ✓ Max Throughput: 1.6+ Gbps
> ✓ 2.4 GHz: 4×4 MIMO
> ✓ 5 GHz: 4×4 MIMO
> ✓ DFS channels
> ✓ HE160
> 
> Netgear
> Nighthawk RAX50
> 
> 
> WAN: 1.0 Gbps Ethernet
> Max Throughput: 949 Mbps
> 2.4 GHz: 2×2 MIMO
> ✓ 5 GHz: 4×4 MIMO
> ✓ DFS channels
> ✓ HE160

★ A Wi-Fi 6E honorable mention: There is currently not a Wi-Fi 6E recommendation
due to the high cost of tri-band (2.4 GHz + 5 GHz + 6 GHz) 4×4 MIMO routers
today (for example, the Netgear RAXE500 (amazon.com) at $600), but I do expect
prices will drop quickly over time.

> TP-Link Archer AXE75
> 2×2 MIMO
> 
> However, there is an 'honorable mention' -- the TP-Link Archer AXE75
> (amazon.com) (seen right) deserves to be pointed out. It is an entry-level
> Wi-Fi 6E router supporting ALL three Wi-Fi bands (2.4 GHz, 5 GHz, and 6 GHz),
> 160 MHz channels (HE160), and DFS channels. And of note is that it only has 1
> Gbps WAN+LAN ports (not 2.5 Gbps). This router is not a recommendation ONLY
> because the router is missing 4×4 MIMO (it has 2×2 MIMO).
> 
> However, let's also be very realistic. If you have less than 1 Gbps Internet
> and don't need extended range (or if you do, are willing to install as a main
> router and as an access point elsewhere where needed), then this router, at
> around $180, is actually a FANTASTIC value, due to support for all three Wi-Fi
> bands (2.4 GHz, 5 GHz, 6 GHz).
> 
> Another 'honorable mention' is the TP Link Archer AXE95, but availability is
> limited. It is very similar to the AXE75 above, but adds (1) 4×4 to only the 5
> GHz band (other bands remain 2×2), and (2) a 2.5 Gbps WAN port.


TP-Link BE550
2×2 MIMO

★ A Wi-Fi 7 honorable mention: There is currently not a Wi-Fi 7 recommendation
due to the very high cost of 4×4 MIMO Wi-Fi 7 routers (like the TP-Link BE800
(amazon.com) at $400). Besides, Wi-Fi 7 is still not even an official ratified
standard.

However, there is a standout 2×2 MIMO Wi-Fi 7 router (so missing 4×4 MIMO), but
supporting all other wanted features that deserves consideration -- the TP-Link
BE550 (amazon.com) -- mainly because the price (at around $220-$250) is around
half the price of any other Wi-Fi 7 router AND because all WAN+LAN Ethernet
ports are 2.5 Gbps.

> The 2.5 Gbps ports are a huge added bonus -- because for most people, there
> will now be no additional expense in buying an external 2.5 Gbps switch. The
> four 2.5 Gbps LAN are enough for connecting a desktop computer and an access
> point (or two).

But the flip side of '2.5 Gbps Ethernet' instead of 5 or 10 Gbps Ethernet seen
in very high end routers is the 2.5 Gbps limit. Technically, maximum Wi-Fi 7 MLO
throughput should be well above 3 Mbps, so throughput will be limited by 2.5
Gbps Ethernet. But honestly, for most people, that limit is more than acceptable
because their ISP speeds are not above that.

TP-Link is clearly targeting 'value' customers with this router (with ISP speeds
less than 2 Gbps) -- and it looks like they hit a home run.

❌ Why no 'high-end' (expensive) router recommendation? Because a single
mid-range router around $200 is actually good enough for most people. And when
not, consider this: I find it impossibly hard to justify a very expensive single
Wi-Fi router for $600 that you then have to place perfectly in a house, when for
the same price, you can obtain three mid-range routers (each around $200) and
then place around your house precisely where they are needed the most (with one
as the main router and the other two in Access Point mode [§18] and
wired/Ethernet back to the main router).

> Plus, you actually don't want everyone in a larger home connected to a single
> router/AP channel. Instead, you want multiple access points, with each AP
> operating on a different Wi-Fi channel. The result is that traffic to each AP
> can happen concurrently (instead of serially as it would be to a single
> router). In effect, you have increased Wi-Fi capacity.
> 
> 
> 
> Most people will obtain FASTER network throughput via a 'recommended' router
> in the same room as they are in (configured as an access point and
> wired/Ethernet back to a main router) than they will from an expensive $600
> main router one room away!
> 
> 
> Analogy: An expensive router is like a Ferrari. When your family needs
> transportation, do you buy one Ferrari for the whole family to use (share), or
> do you buy several other vehicles for everyone to use (independently) -- for
> the same price?



❌ Why no Mesh Wi-Fi Network recommendation? Because of shared Wi-Fi channel
contention between satellite mesh nodes. Also, because virtually all mesh
systems don't provide 4×4 MIMO to client devices. Most mesh networks, providing
only 2×2 MIMO to client devices, are actually then quite 'expensive' as compared
to a router + AP combination both providing 4×4 MIMO.

> I prefer wired/Ethernet backhauls as they can quite simply be much faster than
> a Mesh systems wireless backhaul. When multiple users are all using Wi-Fi at
> the same time (each to their own AP), wired/Ethernet backhauls from multiple
> access points can all operate concurrently (in parallel). But a mesh wireless
> backhaul is a shared medium (between mesh nodes), so concurrent traffic must
> be serialized (and is slower).
> 
> 
> 
> Most people will obtain FASTER network throughput via a 'recommended' router
> in the same room as they are in (configured as an access point and
> wired/Ethernet back to a main router) than they will from an expensive $900
> mesh Wi-Fi node one room away!
> 
> 
> However, mesh networks can be incredibly convenient (just plug in and go), and
> if you want to go down that path, see Mesh Wi-Fi Network Systems [§19].
> 
> Analogy: Going back to the transportation analogy, mesh networks are like
> sharing a taxi with someone else -- you will definitely get to your
> destination, but there may be an intermediate stop (mesh node). Not as fast as
> having your own vehicle and taking a direct route (using an AP).

❌ Why no entry-level router recommendation? Mostly due to no DFS channels [§14],
no MIMO 4×4 [§7], and no 160 MHz channel support. My advice is to generally
avoid any AP/router that does not have all of these features. Also, I learned a
long time ago that 4×4 MIMO offers better signal strength and speeds 'at range'
than a 2×2 MIMO router. So those features are mandatory for me when looking for
a single centrally placed [§P] 'whole house' router.

> TP-Link Archer AX21
> 2×2 MIMO
> 
> Also, don't expect top notch performance (throughput) from some cheap routers.
> While Wi-Fi specifications may be seem similar, run a speed test [§D], and
> throughput results can be disappointing.
> 
> Honorable mention: It is interesting to point out that the "#1 Best Seller"
> router on Amazon.com is only an entry level TP-Link AX21 (amazon.com) for
> 'around' $75, but with only 80-MHz channels, 2×2 MIMO, and no DFS channel
> support and yet does seem to get decent speed test results up to 800 Mbps
> (highspeedinternet.com) from client devices.
> 
> But let's also be practical. If you lower throughput expectations (or can't
> spend hundreds), or you don't need a 'whole house' router, then even a $28
> entry level router configured as an access point and wired/Ethernet back to
> your main router, installed in a single room, to provide Wi-Fi only in that
> one room (like for a kids play room), should be able to provide at least 230
> Mbps (and possibly much higher; depends upon router) Wi-Fi connectivity to
> client devices. Not great, but also not bad for $28. Source (duckware.com).
> 
> Analogy: Going back to the transportation analogy, buying an very low end
> router is like buying a moped. It works, providing you with slower basic
> transportation, but don't expect it to have all the speed and features of a
> car.



TIP: Adding a new Access Point where it is needed most is often the best
solution: For most people, one great router centrally located in the home is all
that is really needed. However, if you have a wireless device (or two) that
absolutely must always have the fastest wireless possibly (no contention with
other wireless), or have a large home, simply add an access point [§18]
(wired/Ethernet to your main router) for those devices.

> For example, place the main router some place centrally located in the house
> -- and then add an access point (wired/Ethernet to the main router) in your
> home office.
> 
> TIP: Don't forget to look into 'Enterprise' grade Access Points. Ubiquiti
> sells a line of 4×4 access point products that DO support ALL 5 GHz channels,
> and are very reasonably priced. For example, the UniFi 6 Pro which is a 4×4
> Wi-Fi 6 AP.



A huge caveat - COST: The cost of the latest and greatest consumer-grade (not
even enterprise-grade) Wi-Fi routers approaching $600 is insane. And $1500 for a
mesh system (seen right) -- talk about pouring money down the drain!

You would be FAR better off spending that money on several high-grade 4×4 APs,
provided the APs can be wired/Ethernet to your existing gigabit router and
provided each AP can be assigned a unique channel.

Then distribute the AP's around so that everyone in the house gets the maximum
PHY speed possible!

> TIP: The hidden YEARLY cost of electricity [§E4] for 24x7 'always on' devices
> can really add up.

Beware Combo (all-in-one) Modems + Routers: These 'combo' devices are a great
convenience and do work, but the problem with these units is that firmware
updates are under the control of your ISP (you are NOT able to update/change
firmware yourself). Or if you can update the firmware, the version often lags
the non-combo hardware (by a lot). Besides, you often need to update just the
router or just the modem, but are now (with a combo unit) forced to upgrade both
at once.

> One example: Compare Netgear's R7800 (router) to the C7800 (router + cable
> modem). With the R7800, you have full control over firmware updates. But with
> the C7800, you have NO control and Netgear states "Firmware upgrades are
> pushed down by your ISP". But if your ISP is a small regional player, you
> might get NO firmware updates at all. Or what if the ISP pushes new FW to your
> combo device and you run into a problem -- It is then impossible to revert FW.
> 
> Worse yet is that some ISP's refuse to push any new firmware to customer owned
> cable modems! I know this from first hand experience with Spectrum outside of
> Orlando FL. I bought a Netgear CM1000V2 and noticed that it was still on V1.0
> firmware. I contacted Spectrum and they refused to do anything about it since
> the modem is 'customer owned' (and they then emphasized that I should be using
> their modem instead).

Remember to properly setup your new router: After installing a new router,
manually configure the 5 GHz band to use a DFS channel. I have never seen the
'auto' channel setting in a consumer router auto select a DFS channel. But you
definitely want to use a DFS channel, because those will be the most unused
channels!

Newer hardware is often better 'at range' then older hardware: I find that new,
next generation routers often perform better -- at distance -- than the prior
generation of routers (like Wi-Fi 6 over Wi-Fi 5 over Wi-Fi 4). Clearly improved
digital signal processing and RF architectures in newer chipsets plays a major
role. Keep this in mind when making purchasing decisions. Examples:

> Newer hardware is demonstratable better than older hardware (youtube.com).
> 
> "Broadcom's RangeBoost technology provides up to 9 dB of receiver sensitivity
> improvement by utilizing advanced signal processing techniques coupled with
> unique receiver architecture .... will benefit from an expanded coverage area"
> -- Source (broadcom.com).

Client device Wi-Fi capabilities matters, a LOT -- so buy the 'right' client
devices! Since client device capabilities often limit Wi-Fi speeds and not the
router, pay attention to the Wi-Fi options when purchasing brand new devices.
Today (September 2024), you want a client device (phone/tablet/laptop/etc)
supporting a minimum of: (a) Wi-Fi 6E, (b) HE160 and (c) 2×2 MIMO.

22. Wrapping it all up
Wi-Fi is messy -- it can be hard to understand, is full of gotchas, caveats,
contradictions, tradeoffs, a ton of hype, and is constantly changing and
improving. And yet, and spite of all of that, Wi-Fi can work remarkably well,
provided that you follow just a couple of very simple rules:



You want
'full bars'

 * use a Wi-Fi 6 or later router
 * use Wi-Fi 6 or later client devices
 * be as close to the router/AP as possible -- you want 'full bars'
 * have the fewest obstacles walls, floors, etc possible between client+router

You 'should' get near 1 Gbps of Wi-Fi throughput when right next to the router.
But you may not. It all depends. If you pay attention to the details, you can
dramatically improve your chances of successfully obtaining 1 Gbps throughput...

What is needed for 1 Gbps Wi-Fi throughput (most of the time): In general, keep
the following points in mind, and you can achieve 1 Gbps throughput over Wi-Fi
to the 'typical' 2×2 MIMO client device:
 * Use modern devices -- older devices may not have the 'horsepower' to sustain
   high throughput
 * In Wi-Fi 5, the only way to obtain 1 Gbps throughput is via 160 Mhz channels,
   which most client devices do not support and if supported, you need to be
   very close to the router.
 * In Wi-Fi 6, you CAN obtain 1 Gbps throughput via an 80 MHz channel in 5 GHz,
   but you need great signal strength topmost 1201 PHY speed, by being very
   close to the router.
 * But you WANT to use 160 MHz channels instead of 80 MHz channels in 5 GHz
   because requires DFS channel support that will extend the range at which you
   can successfully obtain 1 Gbps throughput. But Apple devices do not support
   HE160 in 5 GHz.
 * But using 160 MHz channels in 5 GHz (there are only two) could either (1)
   work just fine, or (2) be impossible -- due to potential DFS channel
   conflicts.
 * Due to this DFS channel uncertainty, using Wi-Fi 6E (6 GHz band) is much
   better (can always use 160 MHz channels), but only if you are close to your
   router, and both client+router must fully supports Wi-Fi 6E. Also, Wi-Fi 6E
   routers (with all the features you really want) can be very expensive AND
   range of 6 GHz channels is reduced as compared to 5 GHz channels.
 * And in all cases above, you WANT a 4×4 MIMO router (instead of 2×2 MIMO), as
   that helps to to extend the range at which you can successfully obtain 1 Gbps
   throughput.

NEED (not just want) 1 Gbps+ Wi-Fi throughput? If you NEED (and not just want) 1
Gbps Wi-Fi throughput, then you need a Wi-Fi 6E/7 router+client so that you can
100% guarantee that a 160 MHz 6 GHz band channel can always be used. The
challenge today is (1) finding a Wi-Fi 6E router, with 'recommended' 4×4 specs,
at a great price there is a 2×2 MIMO honorable mention in the chapter above and
(2) do you have client devices that support Wi-Fi 6E?

With Wi-Fi 6E and a 160 MHz 6 GHz channel, you should be able to easily get 1
Gbps throughput in the same room as the router/AP, and 'likely' get 1 Gbps
throughput even one room away through a single wall and if not, then something
close to 1 Gbps speeds one room away. Also, remember also that 6 GHz band
channels have less range than 5 GHz channels!

> TP-Link BE550 (Wi-Fi 7)
> 
> And then of course, with increasing distance and obstacles, Wi-Fi speeds will
> decrease quickly. Add Wi-Fi 6E access points [§18], wired/Ethernet back to the
> main router, where needed to again guarantee 1 Gbps speeds.
> 
> Wi-Fi 6: You can use a Wi-Fi 6 router and 160 MHz 5 GHz channels to obtain 1
> Gbps speeds, but since DFS channels are then used, there is a risk that the
> router will 'kick you off' the DFS channel, resulting in the router reverting
> to an non-DFS 80 MHz channel. For Wi-Fi 6 80 MHz channels, you CAN still get 1
> Gbps throughput, but only when very close to the router/AP, and most likely
> not through any obstacles (expect only 700+ Mbps one room away).
> 
> Stick your toes into Wi-Fi 7 waters?: The TP-Link BE550 (amazon.com), seen
> right, is a Wi-Fi 7 router, and stands out from all its competitors due to the
> value price of $220 to $250. Virtually all other Wi-Fi 7 routers start out at
> around double that. The BE550 is only 2×2 MIMO, has Wi-Fi 6E built-in, and
> comes with Wi-Fi 7 MLO (5 GHz+6Ghz) for future Wi-Fi 7 clients.

WANT great throughput (often 1 Gbps; sometimes less) + best value: If you want
the best value and fastest Wi-Fi speeds that comes with, then it is hard to
argue against a mid-range recommended [§21] Wi-Fi 6 4×4 MIMO router. I expect
that most readers of this paper will fall into this category. A single well
placed Wi-Fi 6 router in a home using 80 Mhz channels can deliver 1 Gbps speeds
right next to the router, 700+ Mbps speeds in the next room, maybe 300+ Mbps two
rooms away, and so on. For what most people use the Internet for, these Wi-Fi
speeds are very respectable. And as a bonus, if you are fortunate enough that
160 MHz channels in 5 GHz can be used (potential DFS channel issues), then
nearly double these speeds is possible (only for 160 MHz capable client
devices).

> And adding a Wi-Fi 6 access point [§18] wired/Ethernet back to the main
> router, can guarantee 1 Gbps Wi-Fi throughput somewhere else in the house.

4×4 MIMO: You want a router with 4×4 MIMO because that gives a small dB boost to
both transmitted and received Wi-Fi signals. Often times, that means a slightly
better MCS level can be used for 'at range' devices, which then directly means
slightly better throughput. If you can live without that small boost in
range/speeds, then there is nothing wrong with a modern 2×2 MIMO router,
especially if cost is a factor for you, and especially if most Wi-Fi use will be
'near' to the router.

> Why I keep pushing 4×4 MIMO: My personal experience over many years of testing
> is that 4×4 MIMO routers 'at range' regularly outperform 2×2 MIMO routers. For
> example, I have a test setup going 40 feet through three walls, a closet full
> of stuff, and a major obstacle (HVAC air handler), where Wi-Fi throughput in 5
> GHz (160 MHz channel) can either be 300 Mbps (a 2×2 MIMO AP), or 600 Mbps (a
> 4×4 MIMO AP).


Extra 1: Ethernet / PoE (Power over Ethernet)


Cat 5e - notice twists?

Ethernet: Ethernet (wikipedia.org) in 1973 started out at 10 Mbps running over a
'thick' coax cable, with devices tapping into a single common coax cable run. It
worked fine, but was often problematic as 'cable faults' would 'take down'
everyone connected to that single coax cable run. Source (flukenetworks.com).

In the late 1980's, Ethernet quickly changed over to 'twisted pair' wiring, a
cable with four twisted pairs, or 8 wires -- connected to hubs. Today, everyone
uses much smarter 'switches'. It was MUCH more robust since with 'Category
cable' (Cat5e, etc), everyone got a dedicated link to a centralize network
switch, so a cable fault would often just take down a single user, not everyone.

Fast Ethernet: Around 1995, speeds quickly increased to 100 Mbps and was called
Fast Ethernet (wikipedia.org). For both 10 Mbps and 100 Mbps Ethernet, only two
pairs (of four) in the cable (so four wires out of eight, pins 1/2, 3/6) are
actually used. One twisted pair is for Tx (transmit) and the other pair is for
Rx (receive). The other two twisted pairs (pins 4/5, 7/8) are unused.

> Why is it called 'twisted pair': Because each pair of wires is actually
> 'twisted' to help to reduce noise / interference, and each pair in a cable
> twisted at a different rate (twists per inch) to help reduce crosstalk between
> the pairs of wires in a cable. Very noticeable when you compare the orange to
> brown twisted pairs in the photo (upper right).
> 
> And the number of 'twists per inch' for Cat 6 is noticeably even way higher
> than in Cat 5e.
> 
> See also more information (wikipedia.org) and interesting demonstration video
> (youtube.com) on why twisting helps.
> 
> Interesting Fact: What does "TP" in TP-Link mean? Answer (wikipedia.org)





Gigabit Ethernet (GbE): Today, Gigabit Ethernet speeds are very common, with
speeds of 1 Gbps, 2.5 Gbps, 5 Gbps, and 10 Gbps (wikipedia.org). All four pairs
(eight wires) of wires in a cable are used for Gigabit Ethernet. Each pair is
bi-directional (can Tx and Rx at same time).

Cat5e/Cat6: Most everyone is familiar with Ethernet cables with a RJ45 jack on
each end. This 'category' cable comes in several versions/quality, such as Cat5,
Cat5e, Cat6, Cat6a, etc. Supporting higher speeds over a long distance requires
higher quality cable (high speeds over a short distance does not require the
highest quality cable).

> Cat5e cable can be used up to 100 meters (328 feet) for all Ethernet speeds up
> to (and including) 5 Gbps. It is only when you get to 10 Gbps that you need to
> think about higher quality Cat6a cable. But Cat5e can still work for 10 Gbps
> distance up to around 100 feet. Anything past that and you definitely need to
> switch to Cat6a.
> 
> Also, the quality of the cable, and the quality of 'termination' matters a
> LOT. There is a right way to terminate cables and a wrong way, and a LOT of
> what I find on youtube is the wrong way! The bottom of this section has links
> to several trueCable videos on the right way to terminate cables.

PoE (Power over Ethernet): PoE is a technology uses an Ethernet cable to send
both Ethernet/Internet and electrical power over a single Ethernet cable to a
client device (Wi-Fi Access Point, VoIP phone, camera, etc). This is often
implemented by using a 'PoE switch'. The main methods of implementing PoE today
are:
 1. 802.3af / 802.3at standards: Very common. These are industry standards
    (fully plug/play). See Wikipedia. 802.3af allows for a maximum of 15.4 watts
    and 802.3at (aka PoE+) allows for a maximum of 25.5 watts. And there is a
    new recent PoE++ that can deliver even more power. The advantage of these
    standards is that the devices 'negotiate' the amount of power needed and
    delivered. No harm is done if plugged into a device not expecting power (as
    the power is negotiated).
 2. Passive PoE 1 Gbps: This is 'relatively' new. Kind of 'in the middle'
    between standards above and passive below. All pins send data, to achieve 1
    Gbps, and some pins also send power (not negotiated). Expect damage if
    plugged into a device not expecting power.
 3. Passive PoE 100 Mbps: This was very common years ago, but not so much
    anymore given the 802.3af/at standards now exist. Ethernet was on pairs 1/2
    + 3/6 and power was delivered over pair 4/5 + 7/8. Expect damage if plugged
    into a device not expecting power.
 4. Proprietary: In some very old PoE equipment you will find proprietary
    methods of sending Ethernet and power over the 8 pins. Expect damage if
    plugged into a device not expecting power.



Long Reach PoE (long distance cameras): There are Ethernet over Coax adapters
(amazon.com) that can supply power+Ethernet to a PoE device over long distances
(like 3000ft), and work over either a dedicated (a) RG6 coax cable or (b) a
single twisted pair. All devices (the adapters and the PoE device at the end)
are all powered by a PoE switch at the source. These devices work great, but if
the cable is run outdoors, you then need to worry about lightning issues,
especially with long cable runs.

> Example: A 802.3at PoE switch, to adapter (seen right), to 1000' coax, to
> adapter (seen right), to 'PoE powered switch', to two Ring PoE cameras -- all
> powered by a single PoE switch port at the source (no electrical power
> available along the path nor at the destination).
> 
> TIP: Do not confuse EOC with MoCA -- these are two completely different
> technologies.

Ethernet 'crossover' cables: These are an artifact of history. Ethernet cables
were made to connect computers to switches. But in the 'old days', if you wanted
to connect a computer to computer, or a switch to a switch, you needed to use a
special 'crossover' cable. These are no longer needed since all modern Ethernet
ports automatically (electronically) perform the crossover, if it is needed
(automatic MDI-X). See Wikipedia.




Comparing CCA to solid wire

TIP: Avoid CCA (copper clad aluminum): There is a lot of wire coming from
overseas that is really aluminum wire clad in copper that is pure JUNK. It
actually violates U.S. building codes and must never be used for PoE. Make sure
the 'category' wire (Cat5e, Cat6, etc) you install in walls is quality SOLID
(not stranded) 'pure copper' or 'bare copper' wire -- anything else not labeled
as such is likely CCA. More Info (truecable.com).

> It is all about money. Aluminum is 'cheap' compared to copper (copper costs 4×
> more than aluminum).
> 
> So a price too good to be true for cable is a big tip off. All 'CCA' wire is
> relatively inexpensive. All 100% copper wire is noticeably more expensive --
> and labeled as such to explain the increased price.
> 
> The Cat 6 specification actually requires conductors to be pure copper
> (wikipedia.org).

Learn More:
 * Ethernet cables explained (eaton.com) - TONS is great information on cable
   types
 * Understanding Ethernet Wiring (practicalnetworking.net)
 * CMP (plenum) vs. CMR (riser) CAT cable (lanshack.com)
 * Looking at 2.5 and 5 GbE (nextgigsystems.com) - where increasing GbE speeds
   are coming from


--------------------------------------------------------------------------------



UbiGear Ethernet Network kit

Getting started making your own cables / terminations: A quick way to learn and
gain experience is to start playing around with a 'kit' that includes everything
you need to start making cables. For example, this Cat5e kit (amazon.com) for
$16 includes:
 * a crimper for RJ45 ends, blue handle
 * RJ45 ends works for both solid and stranded cable
 * a punch down tool for keystone jacks, yellow
 * a cable testing tool tests pin to pin connectivity, green

Video on how to properly terminate Cat5e RJ45 (youtube.com). Then cut the ends
off a spare cable and start practicing. If you start making a lot of cables, you
may want better tools, but if you only have a couple of terminations to make, a
kit is a very cost effective way to get started.

Pairs are color coded: Each twisted pair in an Ethernet cable (there are four)
are color coded: Blue, Orange, Green, and Brown. With one wire in a twisted pair
'solid' and the other wire 'striped' (or white).











TIA/EIA 568B: Always follow TIA/EIA 568B or 568A, less common color coding order
when terminating RJ45 (see Wikipedia). It ultimately does not matter which
standard you use, but use the same standard on both ends of the same cable
TIA/EIA 568B is the de-facto standard today.

Never 'split' twisted pairs: Ethernet expects not just 8 wires connected, but
rather, four twisted pairs connected on pins 1/2, 3/6, 4/5, and 7/8 (see right).
If you mess up and change that, you are actually 'splitting the pairs', and are
eliminating the benefits from a tightly twisted pair of wires (reduced
noise/interference) -- the split pairs will have a high amount of 'crosstalk'.

> For example, if you wire a cable (incorrectly) so that the twisted pairs are
> on 1/2, 3/4, 5/6, and 7/8 -- then when Ethernet sends a signal down pins 3/6
> (which it expects is a single twisted pair), that signal is now
> inappropriately 'split' and travels down individual wires in two twisted pairs
> (instead of two wires of a single twisted pair).

RJ45 Connectors -- Cat5 vs Cat6: Cat5 and Cat5e RJ45 connectors have the wires
lined up in a single row. Cat6 and Cat6a RJ45 connectors have the wires line up
in a staggered row configuration.

> Cat 5/5e RJ45 - wires in a single row
> Cat 6/6a RJ45 - wires in staggered rows

Making your own cables TIP: If making your own cables, make sure you are buying
the correct RJ45 Cat5 / Cat6 connectors -- that matches the cable type. This is
more about making sure that the 'slots for the wires' are properly sized for the
wire gauge (that changes from Cat5 to Cat6). Also, verify after crimping that
the pins are 'pushed down' all the way -- that means below the RJ45 plastic
ridges (see photos above).

> If not, verify that the 'teeth' on your crimp tool are properly interleaving
> with the RJ45 connector (wiisfi.com).

'How To' Guides: How to correctly terminate...
 * an unshielded punch down keystone jack (truecable.com)
 * a Cat6-6A unshielded RJ45 connector -- with load bar (truecable.com)
 * a shielded Cat6-6A RJ45 connector (truecable.com)
 * How untwisting CAT cable too far can reduce Ethernet speeds (youtube.com)
 * Residential Bonding and Grounding of Shielded Ethernet Cable (youtube.com)

Informative "Low-Voltage Lowdown" Episodes (by trueCable):
 * Episode 1 - Low-voltage best practices Q&A
 * Episode 2 - A keystone jack Q&A
 * Episode 3 - A breakdown of all ethernet cable jacket types
 * Episode 4 - Everything to know about RG6 coaxial cable
 * Episode 5 - Network installation horror stories
 * Episode 6 - Are you terminating your Ethernet cable wrong?
 * Episode 7 - How to get into the low-voltage industry
 * Episode 8 - Is cheap Ethernet cable safe to use?


Extra 2: Bufferbloat
BufferBloat: "Bufferbloat is a cause of high latency and jitter in
packet-switched networks caused by excess buffering of packets." - Source:
wikipedia.org

Bufferbloat takes place at the 'choke point' in networks (where packets buffer)
and this is where mitigations must be implemented (it is presumed that your
modem is often the 'choke point'):
 * Download (at the ISP): When your internet connection starts multiple
   downloads (like, multiple people using the Internet), buffering takes place
   within the ISP network (which holds packets to enforce your Internet download
   speed). The ISP must implement Bufferbloat mitigations.
 * Upload (at your Modem): When you transmit/upload to the Internet faster than
   your Internet connection speed, packets buffer. In cable modems, DOCSIS 3.1
   implements mitigations, that regardless of router used, will solve
   Bufferbloat for most people. So if you only have a DOCSIS 3.0 (or earlier)
   cable modem, strongly consider upgrading to a DOCSIS 3.1 modem.

However, if you feel that you are still impacted by Bufferbloat, look for
routers that advertise that they manage Bufferbloat. These routers work by
limiting the speed of packets transmitted to the Internet modem (intentionally
buffering packets in the router so that the router can implement mitigations) to
a speed below the 'known' Internet upload speed (hopefully fully eliminating any
buffering in the modem). So this will only work (well) if your Internet upload
speed is a rock solid single speed AND if the BufferBloat mitigation algorithm
works in your use case.

Seeing the bufferbloat problem: Without bufferbloat mitigations, all received
packets simply get placed into a single FIFO queue for transmission. For
example, consider a large file downloading (blue) and a VoIP call (red)
happening at the same time -- the delay introduced by the buffered 'download'
packets would wreak havoc on the VoIP call (the VoIP packet must wait for all
packets ahead):


How most bufferbloat mitigations work: Most mitigations solve this problem by
giving priority to those packets that need priority (by transmitting packets in
an order different than received). For example, if the 'download' and 'VoIP
call' packets were instead placed into separate FIFO queues, and the queues were
serviced round-robin, then the VoIP packets (red) would have virtually no delay,
even with an ongoing 'large download' hogging buffer resources at the ISP:


The core difference in various bufferbloat mitigations is what algorithm they
use to re-order packets. Some algorithms even intentionally drop packets (create
packet loss) to hopefully cause the offending program (using too much buffering)
to 'slow down'.

More Information (duckare.com) on Bufferbloat.

Extra 3: New home construction TIPS


LAN/Phone Structured Wiring Cabinet
Gray Cat5e = 4-line Phone / room
Gray Cat5e = spare Cat5e / room
Blue Cat5e = LAN1+Internet / room
Yellow Cat5e = LAN2+Internet / room
Red Cat5e = Thermostats (not used)
Orange = Modem/Router Internet



CATV Structured Wiring Cabinet
Black RG6 = CATV / room
White RG6 = spare RG6 / room
Red Cat5e = Internet to each TV



The bottom line: Add Cat6a Ethernet cable (plus spares) everywhere you can in a
new home during construction, like to all potential Wi-Fi 'Access Point'
locations and all streaming (TV) device locations, etc.

> KEY Wi-Fi concept: Install and use wired/Ethernet connections whenever
> possible. Only use Wi-Fi when that is the only connectivity option.

But isn't the future all wireless? If so, then so is slow Wi-Fi and packet loss!
Yes, wireless continues to improve (a lot), but the simple fact is that wireless
can't come close to the incredible speed and reliability of wired connections
(which are also improving a lot).

> When I built a new home in 2005-2006, my builder tried to convince me that
> installing Cat5e everywhere was not needed because everything was going
> 'wireless'.
> 
> I am sure glad now that I did not take his advice!

Install 'higher quality than needed today' cable: Try to future proof your home
somewhat by installing a higher 'category' of cable than is needed 'today'. Can
you get by with only Cat5e or Cat6 cable today? Yes, of course. But if possible
(and within your budget), install Cat6a cable everywhere instead.

> Back in 2005, I installed a special Belden 'bonded-pair' Cat5e cable and am
> very glad I did. At the time I was only looking to support Fast Ethernet (100
> Mbps), but since then, I have converted to 1 Gbps Ethernet and suspect this
> high quality cable will easily support even higher speeds (especially
> considering that the 'run length' is way shorter than the maximum 100 meter /
> 328 feet).

Wire everything you can: So, for speed and reliability, you should use Ethernet
for everything in a house that you possibly can, and that requires planning (and
then use Wi-Fi for only those devices that can't be wired).

> Ethernet is full duplex, and Ethernet 'switches' allow for full speeds between
> different ports on the switch. Namely, a 16-port 1 Gbps switch allows for 32
> Gbps of non-blocking bandwidth, and that is something that Wi-Fi simply can
> not do.

Structured Wiring: Take some time to research 'structured wiring'. The bottom
line is that all 'low voltage' wiring (Internet / CATV / phone / etc) in a home
must be direct one-to-one (a 'home run') to a centralized location (the
structured wiring cabinets).

NO wiring may be 'daisy-chained', looped, or split. This provides for the
highest quality connections (and allows for reuse for maximum flexibility 'down
the road').

> One room in a new home had two CATV jacks, but the structured wiring cabinet
> only had ONE run to that room. The installer accomplished this via a CATV
> splitter somewhere in the walls! He clearly did not understand 'structured
> wiring' (or was trying to hide a wiring mistake).
> 
> Every 'run' in 'structured wiring' home MUST be a 'home run'.

Add spare cable everywhere! To every location you can think of, run the wires
needed for the service at that location, but then also add spare CAT6a as well.
Consider these locations:
 * telephone jack
 * CATV
 * wired LAN
 * thermostat
 * doorbell
 * security camera
 * Wi-Fi access point
 * desks
 * game console
 * any other location you can think of

The cost of adding spare CAT6a during construction is nothing compared to the
inconvenience of trying to add wires after-the-fact (which frankly, is often
impossible).

> An alternative?: Consider adding strategic 'smurf cable' ENT (electrical
> non-metallic tubing conduit) runs to many locations.

TIP: Avoid CCA (copper clad aluminum): Don't use cable that uses CCA wire --
this is aluminum wire. See Ethernet [§E1]

Wired LAN/Internet in every room: Plan on having one or two wired LAN/Internet
jacks in every room. Even if not connected to RJ45 ends immediately, you want
the wires 'in the walls' when the home is built, so that you can turn them into
wired connections immediately, or as they are needed.

'Category' cable is incredibly versatile: You can run almost anything over CAT6a
cable, with the right adapter (USB / VGA / HDMI / audio / etc).

> So when you have spare CAT6a in every room, that allows you to make changes
> 'down the road' that you can't foresee today.

An example where planning ahead really paid off: In a new home, there was a
spare CAT5e to every TV RG6 location. I am sure glad that I had that 'spare',
which just recently was turned into wired Internet for every Smart TV in the
house (so no Wi-Fi is used for streaming). Also, since every room had two
additional wired Ethernet connections, several runs were repurposed for wireless
access points.



--------------------------------------------------------------------------------


My personal experience: I built a new house in 2005-2006 and here was my
structured wiring checklist (at that time):
 * Every room/desk 'phone location' got one Cat5e (four phone lines) and one
   spare Cat5e.
 * Every room/desk 'Internet/LAN location' got two Cat5e (each connected to a
   different switch in the structured wiring cabinet - for redundancy).
 * Every 'CATV location' got two RG6 and one spare Cat5e.
 * Every 'thermostat location' got a spare Cat5e.
 * The 'CATV demarc location' got one RG6 and one spare RG6.
 * The 'phone demarc location' got one Cat5e and one spare Cat5e.
 * The (potential) 'satellite dish demarc location' got two RG6.

However, in hindsight, the changes I should have made:
 1. I should have run spare Cat5e to: (a) doorbell locations, (b) anticipated
    security camera locations, (c) security alarm base station location, and (d)
    all potential Wi-Fi AP locations.
 2. Run more spare CAT5e to ALL the utility demarc locations (where
    phone/CATV/etc enter the home) at the side of the house, because I have
    already repurposed all of the CAT5e there for other purposes (PoE security
    cams) and need more.
 3. Plan on more electrical outlets in the structured wiring cabinet than you
    think you need, and place onto a dedicated electrical circuit. You don't
    want it on a shared bedroom circuit, overloading (hair dryer + something
    else), tripping, and taking down Internet for the entire house.
 4. Plan on having 'expansion room' in the structured wiring cabinets for future
    changes. Over the years, I added (a) Internet to all TV locations (b) VoIP
    telephone, (c) six PoE security cameras, (d) etc.

A final thought/observation: Plan for more structured wiring space/cabinets than
you think you need. One thing that I did not properly plan for is all the
additional powered devices added into the structured wiring panels after the
fact (many years later) -- and taking into account the SPACE that the additional
power cords/bricks uses up.

> I had initially planned on having only four powered devices (modem, router,
> and two 16-port switches), but over fifteen years, I have since added seven
> more powered devices into the cabinets. Managing the space for all of those
> power cords/bricks was a huge challenge. I actually needed to change the
> layout of both panels (years later) to make room.


Extra 4: The hidden cost of 'always on' devices -- it adds up!
The Q+D 'minimum cost' estimate: Very few people think about how much a device
-- like a modem, router, or switch -- costs in electricity to operate each year.
But there is an incredibly quick and dirty way to estimate the costs for any
device that is on year round:

> Just put a dollar sign in front of the 'watts consumed' to obtain a 'minimum'
> estimate of the yearly cost in dollars.

¢ / kWh: Electricity cost is often cited in terms of 'cents per kilowatthour' --
or how many $0.01 (a cent) to use 1000 watts for one full hour.

Power Supply Watts: Many devices only disclose power supply Volts and Amps.
Calculating (maximum!) Watts given just those two pieces of information is easy:

> Watts = Amps × Volts   (W=A×V)

The trick for easily remembering this formula is the "WAV" acronym (first letter
of each word).




Kill A Watt P3

Power consumption: Look in the manual for your device, which may sometimes will
disclose the maximum power consumption for your device, in Watts. Or better yet,
just directly measure watts used...

Measure Watts consumed: But, a device often times does NOT consume the maximum
watts (often times, something much lower). There are very well known "Kill A
Watt" devices that can very accurately measure actual Watts consumed (example
seen right).

> Namely, the power adapter for my TP-Link AX80 says it can output 12V 3.3A
> (12×3.3=39.6 watts). But using a Kill-A-Watt measuring device, the router uses
> around 7 watts idle, and around 13 watts when Wi-Fi is actively being used.

Why/How this 'put a dollar sign in front of watts' estimate works: Because if we
assume electricity costs 11.408¢/killowatthour the formula for cost/year for an
'always on' device is:

> Estimated $/Year = Watts/1000 × 11.408¢/100 × 24 × 365.25

Which simplifies greatly to just putting a dollar sign in front of watts:

> Estimated $/Year = Watts × $1.00

Corrective factor by State
(as of June 2024) State¢ / kWhfactor Arizona 15.24×1.34 California 34.26×3.00
Florida 14.65×1.28 Georgia 14.10×1.24 Idaho 11.45×1.00 Illinois 16.68×1.46
Michigan 19.23×1.69 New Jersey 18.45×1.62 New York 22.97×2.01 North Carolina
14.98×1.31 Ohio 16.75×1.47 Pennsylvania 17.95×1.57 Texas 15.02×1.32 Virginia
14.99×1.31 Washington 11.84×1.04

Correcting this estimate using your State's actual electricity cost: To find a
'corrective factor' (multiplier) for your location, divide your State's average
Residental 'cents per killowatthour' (eia.gov) by 11.408.

> Actual $/Year = Watts × $1 × corrective_factor

Examples: How much will an 'always on' device consuming 10 watts cost? The
'estimate' is $10. But in Alaska, the actual cost of electricity is around
24.89¢/kWh, resulting in a multiplier of 2.2 (24.89/11.408) -- so Alaska's cost
estimate is $22 (2.2×$10). In California, the multiplier is 3 (34.26/11.408),
for a cost estimate of $30 (3×$10).

> See the table right for more 'corrective factors' for various states.

Conclusion: I find that most people are surprised to learn that a 100 Watt
(incandescent) light bulb kept on for a full year (365×7) would cost around $100
in electricity alone (or $300 in California) -- and have to be replaced nine
times (using a 'typical' bulb lifespan of 1000 hours)! This alone is a great
argument for switching to LED light bulbs! Much longer life span and MUCH lower
electrical costs. And you are incurring this cost each and every year.

Namely, when making purchasing decisions, don't just look at up-front purchase
cost, but also look at 'total cost of ownership', which adds in the 'cost of
electricity' over the expected product life.

Extra 5: MoCA - Ethernet over Coax
MoCA: MoCA means "Multimedia over Coax Alliance". MoCA is a way to add wired
Ethernet to remote locations (up to 16) in a house that has cable TV (RG6)
wiring, but has no way to add/retrofit CAT5+.




MoCA adapter setup

Cat5+ alternative: If you can't use or run new Cat5+ cable, but have access to
RG6 (CATV) in every room, MoCA adapters may be for you. These adapters integrate
into the Cable TV wiring (RG6) that most rooms in a home have, to distribute
Ethernet around your house (example right).

MoCA 2.5: Expect MoCA 2.5 speeds (rated at 2500 Mbps) to easily max out 1 Gbps
Ethernet (949 Mbps throughput). But, test in your environment to confirm. More
information: Wiki info on MoCA (wikipedia.org) and this interesting MoCA 2.0
adapter review (youtube.com).

TIP: Avoid older MoCA versions (that are rated at lower speeds) and instead only
consider using "MoCA 2.5" adapters, many of which now also offer 2.5 Gbps
ethernet ports as well (as opposed to just a Gigabit Ethernet port).




Hitron HETM5 Adapters

MoCA Adapter: For example, on Amazon, look at the Hitron HTEM5 MoCA 2.5 Adapter
2-Pack (amazon.com) for around $120 (has 2.5 Gbps Ethernet ports).

> WARNING: I have also seen disclosures stating that MoCA adapters should NOT be
> used in homes where the coax cable is already being used by satellite TV, or
> for AT&T services (as they already use MoCA internally for their own boxes).
> This need more research and verification (because different MoCA devices in
> the home should be able to just use different channels and easily coexist).

MoCA in ISP provided gateway: If you have an ISP provided gateway device
providing internet service, investigate if there is already one MoCA adapter
'built-in' to the ISP provided gateway There often IS for Verizon FiOS and
Comcast gateways -- but likely only MoCA 2.0.




MOCA 2.5 Splitter

Requirements for MoCA to work: If you decide to use MoCA, make sure you pay
attention to:

> Splitters: All cable splitters everywhere in your network (both inside and
> outside the house) must support MoCA frequencies -- up to 1675 MHz. Example
> splitters (amazon.com) -- notice that the splitters are expressly labeled as
> supporting "MoCA 2.5". If your existing splitters do not say this, they should
> be replaced.
> 
> 
> 
> 
> MOCA PoE Filter
> 
> PoE filter: There must be a MoCA PoE (point of entry) filter on the cable line
> feeding your entire house. Example filter (amazon.com). It is common to find
> an all-in-on 'grounding block' and 'MoCA filter' inside the cable demarc box.
> The filter prevents MoCA signals generated inside your house from feeding back
> into the cable network, and reaching your neighbors.
> 
> If your house has a powered amp and it has a "MoCA" logo on it, it almost
> certainly already has a PoE filter built-in. Just Google the model number of
> the amp to find a manual and confirm.
> 
> An example DOCSIS 3.0 MoCA PoE Filter (amazon.com) is seen upper right, which
> allows 5-1002 MHz to pass, but blocks MoCA frequencies 1125 MHz and above.

DOCSIS 3.1 concerns: DOCSIS 3.1 and MoCA frequencies actually overlap! MoCA uses
1125 MHz to 1675 MHz and DOCSIS 3.1 uses 5 Mhz up to 1218 MHz (or higher?). If
your ISP uses these high frequencies very rare today, but in the future as your
ISP upgrades their network, it will become very common, you will need configure
your MoCA adapters to avoid the overlapping frequencies (and speeds over MoCA
will reduce).

> For example, see this blog post (gocoax.com) on preventing overlap that is
> happening today!

DOCSIS 4.0 risk: This is possibly one of the big downsides of using MoCA today
on cables also containing ISP DOCSIS signals. We know that DOCSIS based ISP's
will eventually 'take over' the entire frequency range that MoCA uses when the
ISP adopts DOCSIS 4.0. It is not a matter of 'if' they will, but 'when' they
will. And when that happens, Ethernet over Coax using MoCA will then no longer
work.

> The bottom line: If you have a DOCSIS ISP and use MoCA, please do so
> understanding that MoCA is NOT a long-term solution to not having Cat5+ cable
> -- it is only a short-term workaround. With MoCA, you are sharing the RG6
> cables with the ISP, and the ISP WILL eventually start using MoCA frequencies.
> An exception is if the RG6 cable is dedicated to MoCA (not shared with any
> DOCSIS ISP signals).

Learning More:
 * The details of MoCA from the source (mocalliance.org)
 * MoCA basics (drive.google.com) - SCTE presentation by PCT
 * ma2500d (gocoax.com) - Network setup diagrams


Extra 6: Wi-is-Fi Windows Tools

tools.zip (wiisfi.com)

Download: Click on the download link (right), open the ZIP file, extract all
files to a folder, and use the tools.

> NOTE: When you try to run an Internet downloaded BAT file, Windows will --
> rightly so -- display a "Windows Protected your PC" message. To run the BAT
> file, click on the "More Info..." link, and then click on the newly exposed
> "Run anyway" button may need to scroll to find the button. You only need to do
> this once (Windows remembers the choice).
> 
> TIP: Always inspect BAT file source code before running any downloaded BAT
> file -- Right click on the filename and select "Edit in Notepad".

★ WI.BAT -- Display "Wi-Fi Information": The WI.BAT file (source code) is
designed to repeatedly display updated Wi-Fi Tx and Rx PHY speed information,
and a bunch of other helpful Wi-Fi information, refreshed every couple of
seconds. It is interesting to see that reported Wi-Fi PHY speeds can be inflated
during times of low Wi-Fi activity, and during Internet speed tests, the PHY
speed decreases to its 'true' value. When finished using the tool, just close
('X') the window.

> 

The metrics of primary interest are:
 * Receive rate (Mbps): Wi-Fi Rx PHY.
 * Transmit rate (Mbps): Wi-Fi Tx PHY.
 * BSSID: MAC address of the AP connected to. Very helpful when you have
   multiple AP.
 * Radio Type: displays 802.11n / 802.11ac / 802.11ax / etc.
 * Band: 2.4 GHz, 5 GHz, 6 GHz, etc.
 * Channel: Wi-Fi channel number (important to know because some perform better
   than others)


--------------------------------------------------------------------------------


Manually running Windows tools: From a Windows 'command prompt', type the
following command

> netsh wlan show interfaces

and

> netsh interface ip show config

to obtain output similar to what WI.BAT displays above. Or if you want a TON of
information about the wireless configuration on your computer, use:

> netsh wlan show all

which includes information about all SSID within range of the computer issuing
this command -- so a very simple and basic 'site survey'. However, please note
that you may first need to click on the Wi-Fi icon in the system taskbar and
wait 10 seconds (to give the computer time to refresh its internal list of
networks that it can see/find).

Extra 7: Recommended Cable Modem
What 'cable modem' should you buy to avoid having to pay monthly modem rental
fees to your ISP?

What to buy: Only look for DOCSIS 3.1 modems that are on your ISP's 'approved
modems' list (but a warning: those lists may not include newly released
hardware). The latest DOCSIS 3.1 standard includes BufferBloat [§E2]
mitigations. Just look for the modem that provides the most value to you.

Voice/Phone: If you get phone service through the cable company, that may
require you to use an ISP provided modem.




Hitron CODA56 - DOCSIS 3.1 - 2.5 Gbps

Recommendation: The Hitron CODA56 (amazon.com), is approved by all major ISP's,
supports Internet speeds up to 2.5 Gbps, comes with a 2.5 Gbps Ethernet port,
and it is bargain compared to its peers ($110 to $140). It does NOT include a
phone port.

> "Certified to work with Xfinity, Charter Spectrum, Cox Gigablast, Cable One
> Sparklight, and Zito Media. Also works with Astound, Grande, RCN, and Wave" --
> source.
> 
> When I called up my ISP and asked them about the Hitron CODA56 not being on
> their approved modems list the reply was 'Oh, that list is old -- if the
> CODA56 supports DOCSIS 3.1, it will work' -- and it did.

Bottom line: In the end, it really does not matter what modem you get -- but do
make sure it is DOCSIS 3.1.

What not to buy: Do not buy DOCSIS 3.0 modems, because that is an old standard.
Also, do not buy Motorola Surfboard Modems, because they (1) have insufficient
upload buffering in them (source), and (2) they 'hang' and need to be powered
off/on (my own experience).

> I had to replace all of my Surfboard modems because they had a buffering
> problem with Ring cameras, resulting in poor video quality. No other brand of
> modem I tested had this problem with Ring cameras. I learned my lesson.


Extra 8: Powerline Networking / PLC


Electrical plug

What is powerline networking / PLC? Powerline networking uses PLC (Power-Line
Communication) to establish an (approx) 20 Mbps to 300 Mbps Ethernet 'switch'
connection between multiple locations (2 to 6+) by using the electrical plugs
already everywhere in your home (seen right).

> Standard 120/240V in the U.S. is AC (alternating current) at 60 Hz. Powerline
> networking works by injecting MHz (megahertz) frequencies into the same wiring
> (aka: intentional 'noise').

Why use PLC? If you need wired Ethernet, but find it impossible to run a Cat5+
cable, give powerline ethernet a try. Plug in one next to your router. Plug the
others exactly where they are needed.




Powerline networking -- TP-Link AV2000

Recommendation: Look for AV2000 Mbps (or higher) adapters. The TP-Link AV2000
(amazon.com), for around $80 for a pair (seen right) is always 'at the top' of
reviews. Example review (nytimes.com)

> Why this adapter in particular? Because you might as well test the 'latest and
> greatest version' of powerline networking that has the best chance of success
> Just like you don't want to be testing Wi-Fi 4 today, you don't want to be
> testing a much older and slower version of powerline networking.
> 
> Over the years, powerline networking speeds have improved dramatically (like
> ×10) -- from AV200, AV500, AV600, AV1000, AV1200, and AV1300 to now AV2000 --
> for the HomePlug AV2 (wikipedia.org) standard. Note that there is also a
> competing G.hn (wikipedia.org) PLC standard.

Setup / Pairing: Setting up the TP-Link AV2000 is very easy. Plug both in and
after a few seconds, hold the 'pair' button down on each unit (for one second)
and then the two devices find and pair to each other. And that is it! The
Ethernet jacks on the two adapters are now paired with each other and act as a
single network switch.

Throughput: Typical throughput ranges from 20 Mbps to around 300 Mbps for PLC
devices -- and you won't know until you install and test. And depending on what
you need/want, that may be acceptable, or horrible.

> So why mention powerline if it is potentially 'slow': (1) because what is 'too
> slow' or acceptable depends upon your situation and (2) powerline is
> inexpensive and easy to test/try. So if you get the speeds you need remotely,
> even if slow -- job done.

TIPS to improve powerline networking speeds:
 * Do NOT plug PLC adapters into power strips -- instead, plug powerline
   adapters directly into a wall outlet. Why? Because many power strips also
   filter out 'noise'. But the PLC signal IS actually intentional high frequency
   noise, which is then blocked!
 * Minimize the total distance / 'stops' over internal house electrical wiring.
   From (a) one powerline adapter, (b) through outlets, (c) to the circuit
   breaker panel, (d) traversing to a new circuit breaker, (e) through outlets,
   and finally to (f) the next powerline adapter. Reducing the number of 'stops'
   on the path will increase powerline networking speeds.
 * Avoid outlets on AFCI circuit breakers -- some models can kill speeds (are
   are harmful to MHz frequencies), while other models are OK. Source
   (smallnetbuilder.com).

What about adapters on different electrical legs/phases: Phases may have
mattered a lot in the past, but that matters much less to modern powerline
networking today. Source: Powerline Communications Cross-phase Coupling
(smallnetbuilder.com) -- a white paper by InTellon stating putting modern
powerline adapters across phases of electricity is just fine. I tested and that
is largely true, BUT, testing shows that different phases can still have a
little impact.

Outbuilding TIPS: Powerline networking speed is all about minimizing: (a) the
distance that PLC signals need to travel, and (b) the number of electrical
devices that the PLC signals need to traverse. So, to best support an
outbuilding, find a way to minimize these! For best supporting an outbuilding,
this means installing electrical outlets for the PLC devices right next to (a)
the circuit breaker panel in the main house and (b) next to the sub-panel in the
outbuilding. And if possible, put both outlets on the same electrical phase.

My speed test results for the TP-Link AV2000: In a house using Square D circuit
breakers:
 * 475 Mbps - 8 feet to 45 feet apart on the same extension cord
 * 320 Mbps - same room different outlets (same circuit breaker)
 * 120 Mbps - different rooms, across two circuit breakers (both same and
   different phases)
 * 100 Mbps - to an outbuilding (70' electrical run), to an outlet right next to
   the sub-panel
 * 30 Mbps - to an outbuilding (70' electrical run), to an outlet in the
   outbuilding

HomePlug AV vs HomePlug AV2: HomePlug AV2 adapters can have a significant
throughput advantage over older HomePlug AV devices:

> "Whereas HomePlug AV always transmits on the Line-Neutral pair, HomePlug AV2
> can transmit on any two pairs formed by the Line, Neutral or Ground wires
> (Line-Neutral, Line-Ground or Neutral-Ground). This allows for significantly
> improved peak data rates and performance" - Source.

The bottom line: Powerline networking speeds are not going to win any awards.
But compared to the alternative of 'no Internet', a slowish -- but reliable --
link may work for you (especially given how easy PLC is to install).

Other resources:
 * How to troubleshoot your Powerline network (smallnetbuilder.com)
 * Powerline networking FAQ (smallnetbuilder.com)


Extra 9: POTS (Telephone) to Ethernet


Back side of RJ14 telco jack



Question: Can you convert POTS (Plain Old Telephone Service) RJ14 jacks to
Internet/Ethernet?

> Look inside a wall jack. If you see 8 pins, that is likely Ethernet. But if
> you see only 2, 4, or 6 pins, that is very likely a phone jack. Look at the
> back side to confirm.

What if your house has NO Ethernet jacks, but it does have telephone jacks. Can
the telephone jacks be removed because everyone uses cellphones today and the
telephone cables rewired and used instead for Internet/Ethernet?

If Cat5 (or better) cable was installed in the walls for phone lines, YES! And
Cat5+ for phone is very common in homes built since 1995 (see blue arrow, photo
right).

★ Step 1: Verify Cat 5 (or better) cable: The first thing you need to do is look
at the back side of all the telephone jacks. How many twisted pairs do you see
in each cable entering the phone box? Hopefully you will see something similar
to what is seen in the image, right -- namely, a cable with four twisted pairs
(blue, orange, green, and brown) -- so eight wires, with one (or two) pair
connected to the phone jack.

> 

Next, can you read the lettering on the cable jacket and see "CAT 5", "CAT 5E",
"CAT 6", or "CAT 6A" (or better)? If you can, then 'YES', the telephone cable
can 100% be reused for Ethernet!

> If you see four twisted pairs, but do not see lettering on the cable, or see
> "CAT 3", then you can still try to convert over to Ethernet. BUT, you may get
> 1 Gbps, but you could also get 100 Mbps, or even 10 Mbps. You just won't know
> until you try -- the shorter the cable run, the better the chances of 'out of
> spec' cable working.


'Home-run' cable installation

★ Step 2: Verify 'daisy chained' or 'home run' cable installation: Do you only
see one cable (and only four twisted pairs) in every single telephone box? If
so, then this is very likely a "home room" installation:

> Home run: Every single phone location in the house and demarc location -- the
> phone signal from the telephone company has its own cable or 'run' to a single
> common utility location (see example right), hopefully inside the house.
> 
> How to convert 'Home Run' to Ethernet: Each CatX cable can be re-terminated
> for Ethernet, and directly used. You can pick and choose which phone jacks you
> want to convert, possibly leaving some phone service remaining.
> 
> What if all home runs terminate outside at the demarc box (instead of inside
> the house): Options: (A) Inside the house opposite the demarc box, open up the
> wall and pull the runs inside the house. (B) Terminate the demarc runs with
> RJ45 and plug runs into a PoE powered mini switch (and send PoE to power the
> switch over one of the runs) example device (ui.com). If there is not room in
> the demarc box for the switch, then push the phone demarc box the to side, and
> add your preferred enclosure over all the runs/switch. (C) same as above, but
> use an outdoor rated PoE powered switch outside the demarc box Example device
> (amazon.com)

However, if you see two cables (each cable with four twisted pairs) in most
telephone boxes but one telephone box with one cable then you likely have a
"daisy-chained" installation:

> Daisy-chained phone: There is ONE long CatX cable run (a) from the phone
> company wiring (often a 'demarc' box on the outside of the house) (b) 'in' to
> the first telephone location, (c) 'out' to the next location, (d) and so on
> until all locations have a cable. Like this:
> 
> 
> 
> When a phone jack is installed, the cable run is cut in half (so now you see
> two cables), and then several 'same colored' wires (like the blue pair for
> line 'one') are attached to the phone jack to: (a) activate the phone jack,
> and (b) continue the phone signal 'out' to the next phone location.
> 
> Select a 'break point' anywhere along the daisy chain (where one of the 'out's
> is disconnected). Everything before the break (one or more jacks) remain
> unmodified and connected to the phone system. But everything after the break
> is converted to Ethernet.
> 
> It is rare, but in some installations the cable at the last telephone location
> loops back to the demarc location -- and both ends of the cable loop are
> connected in the demarc location. This adds fault-tolerance into the wiring
> (the loop can be cut in half anywhere and all jacks remain active/working).

How to convert 'Daisy-Chained' to Ethernet: First, terminate the ends of all
phone cables after the 'break point' (red line below) with an Ethernet keystone
jack (and get two-port keystone wall plates). Next, find and test each run --
and meticulously label the jacks, similar to below left-side label #, and top
port, means earlier in the chain; right-side label #, and bottom port, means
later in the chain; number means cable run number:

> 

Remember that one cable likely runs outside to the phone company demarc
location, and can remain a phone jack. Finally, you have a choice to make --
implement either:
 1. Add ONE Ethernet run: You can directly use any single run (jacks with the
    same numbers). If you need a longer run, add a small Ethernet patch cord
    between in/out jacks at the in-between locations. For example to create an
    Ethernet run all the way from the first location ('2' on first plate) to the
    last location ('4' on last plate), add a patch cord connecting ports at the
    two in-between locations.

 2. 
    Sodola PoE Extender + 1 Gbps Switch
    
    Add Ethernet to all / multiple locations: At one of the phone jack
    locations, add a 802.3at PoE switch and connect to the Ethernet wall jacks.
    Then, at phone jack locations 'down the line', add a $20 PoE extender
    (amazon.com) -- which: (a) is powered by PoE itself, (b) continues PoE
    downstream, (c) leave two Ethernet ports open for use at the phone location,
    (d) provides 1 Gbps Ethernet. Repeat as needed as this PoE extender can be
    daisy-chained.
    
    NOTES: (1) If you use PoE extenders to add Ethernet to each daisy-chained
    phone location, don't actually plug any PoE device in. The intention is that
    the entire PoE budget is there to power only the daisy-chained PoE
    extenders. (2) Of course, you can also use powered switches, and then a PoE
    switch and PoE extenders are not needed (for example, if you wanted 2.5 Gbps
    Ethernet, instead of the 1 Gbps PoE solution).

Or, mix and match the two techniques above. Either way, with the right cable
installed in the walls, you CAN repurpose telephone locations for
Internet/Ethernet.

TIP: Alarm systems: If you are trying to make sense of what you see in the phone
demarc box, the 'typical' way the phone signal travels when there is an alarm
system present is: (1) phone company » (2) demarc box » (3) alarm panel » (4)
demarc box » (5) phone jacks in the house. This allows an alarm system to
'seize' control of the phone line during alarms because it is 'first in line'.

TIP: Re-terminating phone CatX for Ethernet: When the phone lines were
installed, some twisted pairs were likely untwisted way more than they should
have been. But Ethernet needs those twists for reducing noise and EMI. So before
re-terminating, take the time to ensure that each wire pair is properly twisted
all the way to the termination point. You will very likely need to carefully
retwist some of the pairs (but note that it is also bad to overtwist).

Other interesting items:
 * Ubiquiti Flex Mini - "A compact, 5-port, Layer 2 switch powered by PoE or a
   5V USB-C adapter."


Appendix A: Troubleshooting 'slow' Wi-Fi
Goal: With a modern (2×2) Wi-Fi 6 client device (phone, tablet, etc) connecting
to a modern (4×4) Wi-Fi 6 router, you should be able to easily see and verify
[§4] a 1201 (80 MHz) Mbps or 2402 Mbps (160 MHz) PHY connection between the two
devices, when standing right next to the router. Then as you move away from the
router (adding distance/walls), PHY speeds will decrease.




> If you get good Wi-Fi throughput standing right next to your router, but poor
> throughput far away from the router, then you simply have a 'distance'
> problem. Consider adding an access point [§18] into your network at that
> distant location, wired/Ethernet back to your main router.

Disconnect/Reconnect Wi-Fi: You might be surprised how often simply
disconnecting from Wi-Fi on the client and reconnecting to Wi-Fi resolves some
(unknown) speed problem. Technically this should never happen, but it does due
to bugs.

Did you reboot everything? It can't hurt to power cycle your modem, router,
client device, etc, and see if the problem goes away. I was once unable to track
down the cause of slow Wi-Fi, and rebooting all devices solved the problem.
Crazy, but it happens. Again, technically, this should never happen, but due to
(unknown) bugs, it does sometimes happen.

> In another case, I was getting great PHY speeds, but very slow Internet
> speeds. A bunch of speed tests eliminated Wi-Fi and my router as the cause.
> This suggested the problem was with the only remaining hardware -- the (cable
> company provided) modem. So I rebooted (only) the modem and instantly had my
> fast speeds back. Frustrating.
> 
> At one location, ISP download speeds were sometimes OK, and sometimes bad.
> Rebooting the cable modem caused it to acquire a 'random' set of bonded
> download channels from the CMTS. As one particular channel was having issues,
> this reboot caused the modem to randomly use/avoid the problematic channel.

Verify PHY speed: Go to your wireless device and check the PHY speed [§4] at
which the device is connecting to your router. Then take 70% of the PHY speed as
a fair estimate for the maximum realistic throughput speed that one device can
achieve.

> Also, walk to your router and stand about five feet (line of sight) away from
> the router, cause some Internet activity, and then re-check PHY speed. On a
> 2×2 Wi-Fi 6 client, I would expect to see a very strong signal with a PHY
> speed of 1201 Mbps (or better). And for a Wi-Fi 5 client, expect a 866 Mbps
> PHY.

Verify channel/band: Verify that you are actually connecting to the 5 GHz SSID
on your router, as accidentally connecting to the 2.4 GHz SSID could be the
problem.

> Newer client devices (in the Wi-Fi settings) may tell you the 'Frequency' (2.4
> or 5 GHz) your device is connected to. If not, the PHY speed you see on your
> device should be a huge tip off as to which band (and channel width) you are
> connecting to. Look up the speed in the PHY tables [§F], which then reveals a
> ton of information about how you are connecting.
> 
> Split the Wi-Fi bands: Try turning off the 'Smart Connect' feature of the
> router (that tries to push the client device to the 'best' band). Or, for
> testing, go to the router configuration and confirm that the 2.4 GHz SSID and
> the 5 GHz SSID are uniquely named. If they are the same, append a "-5G" to the
> 5 GHz SSID name temporarily. When you see the SSID for each band, connect to
> the 5 GHz SSID. If PHY speed increases dramatically, you likely have a problem
> with your device connected to the (slower) 2.4 GHz SSID, instead of (much
> faster) 5 GHz SSID.


Wi-Fi Analyzer Access Points

Verify router capabilities: Install the "WiFiAnalyzer (open-source)"
(play.google.com) Android app on your smartphone and verify that the 5 GHz SSID
from your router is using the channel number and channel width that you expect
to see. If wrong, correct in the router configuration. Or, try switching
channels.

> Verify channel width: It is important to confirm that your 5 GHz SSID is
> operating at an '80 MHz' channel width. If you see '40 MHz' or '20 MHz' for
> your 5 GHz SSID (notice this for 'Goofy-5G' in example right?), go into your
> router configuration and fix the problem. The other possibility is that the
> router is only a dual-band Wi-Fi 4 (802.11n; not capable of 80 MHz channels)
> router and not actually a Wi-Fi 5 (802.11ac) router.

Site Survey: Use the "WiFiAnalyzer (open-source)" (play.google.com) Android app
and find out exactly what channels are being used (who you are sharing spectrum
with), and then set your router to use the most unused channel. Channels 1, 6,
11 are 2.4 GHz channels. Channels 36 - 165 are 5 GHz channels. And then verify a
good channel by running a speed test [§D].

Q: I upgraded my ISP Internet speed from 60 Mbps to 120 Mbps and my wired
Internet speeds dropped!

> A: The most likely cause is that NOT all devices in your network are 1 Gbps
> capable. If there are any Fast Ethernet (100Mbps) devices between you and your
> ISP, that device becomes a 'choke point' that will very likely cause dropped
> packets and speed problems.
> 
> TIP: It is very common for ISP's to provision internet modems to 110% of the
> advertised speed. So if you sign up for 100 Mbps internet, your 100 Mbps
> network will not work, since the internet speed is more than likely actually
> 110 Mbps -- and any Fast Ethernet devices between you and the internet are a
> problem.
> 
> Another possibility is a bad Ethernet cable between two Gbps devices, causing
> that link to operate at the problematic 100 Mbps speed instead of the desired
> 1 Gbps speed. TIP: Many RJ45 jacks have tiny LED lights indicating speed.

Q: I am connected to my Wi-Fi router at 866 Mbps, but a speedtest shows only 500
Mbps?

> A: Due to Wi-Fi protocol overhead, the expected throughput at the application
> level is around 60% to 80% of the physical (PHY) Wi-Fi speed. This is normal
> and sadly, the router industry has done a horrible job explaining this to the
> general public.

Q: I bought a '1733' AC Wi-Fi router, but I can only connect at 650 Mbps (PHY)
from my smartphone?

> A: Blame router companies -- they love to advertise maximum speeds. Looking at
> the 5 GHz speed table far above, we can see that '1733' implies a 4×4 access
> point supporting 256-QAM. 650 is the PHY speed for a 2×2 device at 64-QAM. The
> conclusion is that your smartphone is a 2×2 MIMO device and that you are maybe
> 20 to 30 feet from your router. Expected application throughput will be around
> 70% of that, or 455 Mbps (±45 Mbps).

Q: My speedtest proves I only get a slow XXX Mbps?

> A: Maybe, but maybe not. Always try several different Internet speed test
> programs, like speedtest.xfinity.com, fast.com, or cfspeed.com. The very
> nature of the Internet is that everyone will not always be fast. I find that
> the Xfinity speed test gives the most reliable results, almost all of the time
> (and it had better, since Comcast is the largest broadband provider in the
> U.S.).

Q: I have 250 Mbps Internet, but I max out at 95 Mbps both wireless and wired to
my router?

> A: Fast Ethernet (100 Mbps) maxes out at a throughput of around 94.92 Mbps
> within applications. So the most likely cause is that the router is only 'Fast
> Ethernet' and you will need to upgrade to a Gigabit capable router (modem is
> likely 1Gbps ethernet, but router is only 100 Mbps, causing the bottleneck).
> OR, if your router is Gigabit, is there a (slow) Fast Ethernet switch
> somewhere in the network between you and your router? OR, double check the
> color of LED lights for RJ45 connections (LED color should indicate 1 Gbps and
> not 100Mbps). You may need to replace a bad ethernet cable (as Gigabit
> requires all eight wires to be good; 100Mbps only uses four of the eight
> wires).

Q: No matter what, my PHY speed maxes out at 54 Mbps. What is wrong?

> A: This can happen when 'WMM' (Wi-Fi Multimedia) is turned off in the router
> configuration. To fix, turn 'WMM' back on. WMM is actually required for any
> speeds past 54 Mbps.

Q: Why do I not see 802.11ac link speeds connecting to my router's 5 GHz band?

> A: The most likely cause (if your router is configured for 80 MHz channels) is
> that your client device does NOT support 802.11ac. Instead, your client device
> is likely only 802.11n (no 256-QAM support), and supports 'dual-band' and is
> connecting to the 5 GHz band using the maximum 40 MHz channel width of
> 802.11n.
> 
> Visit devicespecifications.com or phonescoop.com to research versions of Wi-Fi
> supported by your smartphone (802.11n vs 802.11ac, etc).

Q: Why don't I get fast Wi-Fi speeds from phone upstairs to my router
downstairs?

> A: Wi-Fi speeds decrease with distance from the router, and especially
> decrease through obstacles (walls, floors, etc). Your maximum speed will be
> when you are just feet from the router (and line-of-sight), and speeds will
> slowly decrease the farther away you move from the router. Sorry, but that is
> just how Wi-Fi signals work.

Q: Why is my client PHY speed stuck at 86.6 Mbps (or 173.3 Mbps)?

> A: This can happen when a router is set to use channel 165. When channel 165
> is selected, there are actually NO 40/80/160 channels available (so the router
> can only operate using 20 MHz channels). So even if the router is set to use
> 80 MHz channels, every Wi-Fi client connecting to channel 165 will only use 20
> MHz channels. To fix, select a different channel.

Q: Why is my PHY speed 866 Mbps, but a throughput test shows only 100 Mbps?

> A: Try a different channel on the router, or try updating Wi-Fi driver
> software on the client. The only time I have seen this is with an Intel
> AC-7260 laptop (channel 144 not supported) connecting to a Netgear R7800
> router on DFS channel 140. The router transmitted to the laptop at 20 MHz
> speeds and the laptop transmitted to the router at 80 MHz speeds. Upgrading
> Wi-Fi drivers on the laptop resolved the problem.

Q: My wireless Internet is horrible at video calls, what can I do?

> A: First, can you connect wired/Ethernet to the Internet and confirm the
> problems go away (this confirms a wireless problem)? Try moving much closer to
> your wireless AP to maximize signal strength during the call. Try connecting
> to your router's other Wi-Fi band (2.4 GHz to 5 GHz, or vise-versa). Try
> changing channels on the router. As a last resort, try setting your router to
> use 20 MHz channels in 5 GHz and try different channels (this will greatly
> reduce your maximum Internet speed, but hopefully in return, you will gain an
> ultra-reliable connection for your video calls).

Q: Why are Wi-Fi Internet speed tests abnormally slow on my brand new Dell
laptop?

> A: If you see the SmartByte application installed (to check, go into 'Control
> Panel / Uninstall a program', search on 'SmartByte', and if you get a hit, it
> is installed), disable the SmartByte service, as that Dell service is the
> likely cause. Google "beware of SmartByte" (google.com) for a discussion and
> instructions on how to disable/uninstall. This Dell service, designed to give
> priority to video streaming, actually causes slow Internet speeds. Anything
> designed to 'speed things up' that actually 'slows things down' is pure
> garbage. I have personally experienced this problem on multiple brand new Dell
> laptops. Absolutely crazy.

Q: Internet speed tests are sporadic and the modem diagnostic page shows that
one "Channel ID" has a lot of 'uncorrectable' errors. What is wrong?

> A: The most likely cause of the modem diagnostic page showing that all
> "Channel ID" have very low 'uncorrectable' errors, expect for one Channel ID
> with a sky high 'uncorrectable' count -- is is a bad RG6 cable or connector.
> Try using a different RG6 cable. See this paper (duckware.com) on cable modem
> connectivity problems.

Q: Is it true that placing a Wi-Fi router right next to a cable modem can
sometimes cause Internet up/down issues?

> A: Maybe. I experienced a similar issue first-hand with a Netgear R7800
> sitting right next to a Netgear CM1000V2 cable modem. The Internet would
> sporadically drop. The cable modem was seeing a fair amount of "Uncorrectable
> Codewords", and T3 errors in the events logs. Moving the R7800 over three feet
> away from the cable modem seemed to reduce the problem. Your experience may be
> different, but in general, it is a good idea to keep RF devices away from
> other devices as much as possible.

Q: I have an older device that can't connect to my Wi-Fi (and I know
SSID/password are correct) -- why?

> A: Check to see if your router/mesh device has something called 802.11r "Fast
> Roaming" turned on. If so, turn off and see if that helps. Some older Wi-Fi
> client devices can't connect at all to networks with 802.11r enabled.

Also, the chapter above on how to improve Wi-Fi speeds [§16] might help.

TIP: Digging deeper on Windows: Use the Wi-is-Fi tools [§E6].

Appendix B: Investigating Router Specifications
Finding Router Specifications: How do you look at the specifications for a
router and make sense of them? One of the best ways to 'figure things out' is to
take all cited information that is disclosed, and find the 'Mbps' speed in the
PHY tables spreadsheet (docs.google.com), which then tells helps to figure out
the non-disclosed specifications. To investigate, go to a vendor's web page for
a router model and look for the 'technical specifications' section:
 * TP-Link Routers (tp-link.com) - click on "Specifications" on top of product
   pages
 * Netgear Routers (netgear.com) - on product page, click on "Specs", then
   "Technical Details"
 * ASUS Routers (asus.com) - click on "Tech Specs" on top of product pages
 * Linksys Routers (linksys.com) - find "Technical Specification" on product
   pages
 * Eero Routers (eero.com) - expand 'Radio Frequency' section for MIMO specs
 * Tenda Routers (tendaacn.com) - click on "Specification" on top of product
   pages
 * WavLink Routers (wavlink.com) - click on "Specification" part of product
   pages
 * Synology Routers (synology.com) - click on "Specs" section of product pages
 * Google Routers (google.com) - click on "Tech Specs", expand 'Performance'
   section
 * D-Link Routers (dlink.com) - click on "Specifications" section on product
   pages
 * Ubiquiti Routers (ui.com) - click on "Technical Specifications" on product
   pages

MCS/QAM: Router companies always cite (and add) Mbps values for the highest
supported MCS level, in each band. This tells you what 'row' to lookup values in
the PHY tables spreadsheet (docs.google.com). Highest supported MCS levels will
often be:
 * "256-QAM 5/6" for Wi-Fi 4/5
 * "1024-QAM 5/6" for Wi-Fi 6/6E
 * "4096-QAM 5/6" for Wi-Fi 7

However, be careful, as there are sometimes exceptions -- like a Wi-Fi 6 router
that is only Wi-Fi 6 in the 5 GHz band, but is an older Wi-Fi version in the 2.4
GHz band!

Channel widths: Router companies always cite Mbps values for the highest
supported channel widths, in each band. This tells you which PHY tables
spreadsheet (docs.google.com) to use.
 * For the 2.4 GHz band, router companies almost always cite speeds for 40 MHz
   channel widths, even though most people will only be able to use 20 MHz
   channel widths.
 * For the 5 GHz band and Wi-Fi 5, channel width is often 80 MHz or 160 MHz. If
   not disclosed, it must be deduced from all the other information provided.
 * For the 5 GHz band and Wi-Fi 6, channel width for higher end routers will be
   160 MHz, but will be 80 MHz for low end routers.
 * For Wi-Fi 7, high end routers will cite speeds for 320 MHz (but lower end
   routers could cite for 80/160).

Warning: Your client device may only support a smaller channel width (eg: HE80)
than the cited Mbps / channel width (eg: HE160).





MIMO level: For Wi-Fi 5, router companies just cite the highest MIMO level used
in any Wi-Fi band (eg: two or four). But for Wi-Fi 6/6E/7, router companies now
add the MIMO levels for each band together and report that sum as the number of
"Streams" that router supports. Details [§Q]. That 'sum' can be helpful to
decipher incomplete router specifications.

Example One: For example, take the TP-Link AX80 router (seen right). Immediately
take note of the 'stream' sum (8) and the two Mbps values (4804 and 1148).

> Note that two Mbps numbers added together means 'dual band'. If you saw three
> Mbps numbers added together, that is tri-band; and four Mbps numbers is
> quad-band.

So, for "8 Stream", and dual-band, two numbers added together (the MIMO level
for each band) must equal 8. A good first logical guess is just 4+4=8, and sure
enough, when we look at the PHY tables for Wi-Fi 6 (802.11ax), 4×4 MIMO, and
look up the two values 1148 and 4804 in the PHY tables spreadsheet
(docs.google.com), we find:
 * in the 802.11ax 40 MHz channel table, 1024-QAM 5/6 row, 1148 is value for 4×4
   MIMO
 * in the 802.11ax 160 MHz channel table, 1024-QAM 5/6 row, 4804 is the value
   4×4 MIMO

So this 'confirms' that both the 2.4 GHz and 5 GHz bands are 4×4 MIMO for the
TP-Link AX80.





Example Two: For the Netgear RAX50 (seen right), go to the Netgear product page
and scroll down and expand the "Technical Specs" section and immediately find
that the 2.4 GHz band is 2×2 MIMO (max 545 Mbps) and that the 5 GHz band is 4×4
MIMO (max 4.8 Gbps).

> In this case, the specification told us the MIMO levels for each band, but if
> not, we could have deduced it from the fact that Netgear told us this was a
> six-stream router. Looking up the disclosed Mbps values in the PHY tables
> spreadsheet (docs.google.com), the only MIMO combination that works and makes
> sense is that 5 GHz is 4×4 and that 2.4 GHz is 2×2.

Example Three: Take the TP-Link AX21 Router, dual-band, four-stream router with
Mbps speeds of 1201+574. It does not take long to figure out from the PHY tables
spreadsheet (docs.google.com), that the only 'values' that 'work' are:
 * in the 802.11ax 80 MHz table, 2×2 MIMO is 1201 Mbps -- for 5 GHz band
 * in the 802.11ax 40 MHz table, 2×2 MIMO is 574 Mbps -- for 2.4 GHz band

Also helpful is that TP-Link, under "Specifications" on product web pages
(tp-link.com), enumerates the "WiFi Reception Sensitivity" for each Wi-Fi band,
which also reveals maximum supported channel widths (HE40 is a 40 MHz channel;
HE80 is an 80 MHz channel; etc).

Example Four: Take the TP-Link AX75, a tri-band, six stream, 5400 Mbps class
Wi-Fi 6E router. The manufacturer's specifications web page (tp-link.com) tell
us the 6 GHz band supports HE160, the 5 GHz band supports HE160, and the 2.4 GHz
band supports HE40. Our first guess is that there is 2×2 MIMO for each band
(because 2+2+2 = 6), and adding up the Mbps for all bands (using 2×2 MIMO and
1024-QAM) gets us to the 5400 Mbps specification, confirming our MIMO guess:
 * in the 802.11ax 160 MHz table, 2×2 MIMO is 2402 Mbps -- for 6 GHz band
 * in the 802.11ax 160 MHz table, 2×2 MIMO is 2402 Mbps -- for 5 GHz band
 * in the 802.11ax 40 MHz table, 2×2 MIMO is 574 Mbps -- for 2.4 GHz band

An alternative: If you know you want a dual-band Wi-Fi 6 router supporting
HE160, 4×4 MIMO, and the topmost QAM-1024, go to the PHY tables spreadsheet
(docs.google.com) and lookup those values in the appropriate tables. Use the
802.11ax 40 MHz table for 2.4 GHz to find 1147 Mbps. Use the 802.11ax 160 MHz
table for 5 GHz to find 4804 Mbps. You now know you want a router with a minimum
specification of "4804 Mbps + 1147 Mbps". That quickly narrows down the list of
routers. Then from that much shorter list, compare prices and confirm actual
specifications.

Appendix C: Netgear 'Mode' means Channel Width
What does Netgear 'Mode' mean? So you just got a new router and are setting it
up and you see 'Mode' and various Mbps under the Wi-Fi settings, but what does
that mean? It means 160/80/40/20 MHz channel width! It does NOT change or adjust
MIMO level support (which can not be directly changed).

> TIP: Netgear 'Mode' means "channel width".



Netgear R7800 (4×4) 5 GHz example:
 * 1733: Lookup 1733 in the PHY tables far above and you find it under the 80
   MHz PHY table with 256-QAM 5/6 modulation and 4×4 MIMO.
   
   
 * 800: Lookup 800 in the PHY tables far above and you find it under the 40 MHz
   PHY table with 256-QAM 5/6 modulation and 4×4 MIMO.
   
   
 * 347: Lookup 347 in the PHY tables far above and you find it (346.6) under the
   20 MHz PHY table with 256-QAM 3/4 modulation with 4×4 MIMO. Note that this is
   because 256-QAM 5/6 is not available in 20 MHz mode (for most commonly used
   MIMO configurations).




Netgear R6250 (3×3) 5 GHz example:
 * 1300: Lookup 1300 in the PHY tables far above and you find it under the 80
   MHz PHY table with 256-QAM 5/6 modulation and 3×3 MIMO.
   
   
 * 600: Lookup 600 in the PHY tables far above and you find it under the 40 MHz
   PHY table with 256-QAM 5/6 modulation and 3×3 MIMO.
   
   
 * 289: Lookup 289 in the PHY tables far above and you find it (288.8) under the
   20 MHz PHY table with 256-QAM 5/6 modulation with 3×3 MIMO.




Netgear JNR3210 (2×2) 2.4 GHz example:
 * 300: Lookup 300 in the PHY tables far above and you find it under the 40 MHz
   PHY table with 64-QAM 5/6 modulation and 2×2 MIMO.
   
   
 * 145: Lookup 145 in the PHY tables far above and you find it under the 20 MHz
   PHY table with 64-QAM 5/6 modulation and 2×2 MIMO.
   
   
 * 54: This is the exception. In Netgear routers on the 2.4 GHz band, this sets
   the router to 802.11g (54 Mbps) operation only.


Appendix D: Throughput Testing Tools

ISP speed tests AT&T
Google Fiber
Spectrum
Verizon
Xfinity


The first step in setting up an AP/router is selecting a Wi-Fi channel that you
think is the 'most unused channel' and confirm that you obtain a good PHY speed
[§4] from your client devices.

The second step is to test and verify that you are obtaining the expected
throughput, when next to that AP -- 'around' 70% (±10%) of the PHY speed. If you
see substantially less than this, try a different channel.

> TIP: Reboot all of your test equipment before running throughput tests. I have
> spent WAY too much time trying to track down the cause of a slow performance
> test (and NOT finding the problem), only to at the end, rebooting, and having
> the problem go away. Frustrating.


Xfinity speed test

ISP speed test (very easy to use; result quality unknown): Run your ISP's
website speed test tool (see table upper right). Very helpful if all you need to
confirm is that speeds are past some value, but if results are not what you
expected, you don't know where the 'choke point' is (Wi-Fi, the ISP, something
else?).

> ISP speed tests will only be 'reliable' at testing Wi-Fi throughput if the ISP
> speed is known to be faster than your expected Wi-Fi throughput.
> 
> TIP: But if ISP speed tests over Wi-Fi improve significantly as you move
> closer to the router, that is a classic sign that Wi-Fi is indeed the 'choke
> point', and limiting throughput.

◆ TxRate (easy to use; often great results): A very helpful speedtest tool that
I use all the time to 'see' and visualize Wi-Fi first hop Tx throughput in
real-time on my laptop/PC is TxRate (duckware.com). The huge advantage of this
tool is that it only needs to be installed on the laptop/PC running the speed
test (no other laptop/PC is needed), but this comes with the slight disadvantage
that it only tests Wi-Fi transmit speed (not receive speed) of just the first
hop so not for mesh or extenders, that have multiple hops to the main router.

> But these disadvantages are a very small price to pay for the huge advantage
> and convenience of (1) not having to install any software elsewhere in the
> network being tested and (2) working independently of LAN Ethernet speed (not
> limited when LAN is only 1 Gbps). Besides, there is often just a single Wi-Fi
> hop to test as client devices often connect directly via Wi-Fi to the main
> router or AP, and client Tx speed is often (but not always) slower than Rx
> speed, so this often ends up testing the desired 'weakest' Tx link.

This graph (below) is TxRate capturing the maximum transmit speed from my laptop
to a Wi-Fi 5 main router (in the next room), and halfway through the test, I
started a speed test on a second computer, intentionally creating 'contention'
for the Wi-Fi channel. My laptop is then forced to share the Wi-Fi channel with
that other computer, and as expected, we then both get around 50% of the channel
capacity:

> 

Downloading from 192.168.0.211 port 33333...
  242,548,736 bytes in 1000 ms = 1,940,389,888 bps
  240,844,800 bytes in 1000 ms = 1,926,758,400 bps
  240,975,872 bytes in 1000 ms = 1,927,806,976 bps
  242,417,664 bytes in 1000 ms = 1,939,341,312 bps
  239,992,832 bytes in 1000 ms = 1,919,942,656 bps
Uploading to 192.168.0.211 port 33333...
  228,851,712 bytes in 1000 ms = 1,830,813,696 bps
  229,048,320 bytes in 1000 ms = 1,832,386,560 bps
  218,038,272 bytes in 1005 ms = 1,735,628,035 bps
  234,684,416 bytes in 1000 ms = 1,877,475,328 bps
  225,902,592 bytes in 1000 ms = 1,807,220,736 bps


Wi-Fi 6 160 MHz channel throughput with 2402 Mbps PHY

◆ SpeedTest (BEST; super accurate; harder to use): The best speed test tool
(great results) is also the hardest to setup and use because it requires using
another computer on the network as a speedtest server, and a fast LAN, which is
not always possible.

Use this SpeedTest program (duckware.com) to easily test both download / upload
Mbps speeds between any two computers on your network. One computer is connected
to Ethernet and runs the 'server'. Another computer is connected to Wi-Fi and
runs the 'client'.

This speed test program works on your LAN -- so it avoids Internet usage. Plus,
ISP speeds may not max out your Wi-Fi connection. Ideally, your Ethernet LAN
speed is 1 Gbps (or higher), which should allow very accurate Wi-Fi
download/upload speed measurements (as Tx PHY and Rx PHY for Wi-Fi can sometimes
be very different).

> TIP: Validate your LAN speeds first. Run this speed test program between two
> computers on your LAN first, to confirm that both client+server computers, and
> your LAN, can handle full LAN speeds. Expect around 5.1% overhead for TCP over
> Ethernet. On a 1 Gbps LAN, expect measured throughput speeds to max out at
> around 949 Mbps. On a 2.5 Gbps LAN, expect the same 5.1% overhead and a
> maximum measured throughput speed around 2.37 Gbps.
> 
> 
> 
> 
> USB-C 2.5 Gbps Ethernet
> 
> Inexpensive 2.5 Gbps Ethernet: You will need a 2.5 Gbps (or faster) Ethernet
> network to measure any Wi-Fi speeds past 1 Gbps (and up to 2.5 Gbps). The uni
> USB-C to Ethernet Adapter 2.5 Gbps (amazon.com) (seen right) is an inexpensive
> (around $27) way to add 2.5 Gbps Ethernet to computers with USB-C.
> 
> > Do NOT buy the Anker brand of this adapter. I did and found that it is
> > unable to connect when a laptop of plugged into AC power (and on battery, it
> > only works to 1 Gbps). Crazy bug. The 'uni' works just fine for me. Both the
> > Uni and Anker use a Realtek chipset. So make sure that your Realtek drivers
> > are up-to-date. Compare to the latest version at Realtek USB FE / GBE / 2.5G
> > / 5G Ethernet Family Controller Software (realtek.com). When I connected to
> > a Windows 10 laptop, speed was bad and the driver was crazy old (from 2015).
> > Problems were fixed after updating the driver (but not for Anker).




Another option: Use OS provided tools: A relatively 'easy' way to test Wi-Fi
throughput is by transferring a large file from one PC (on wired Ethernet) to
another PC (on Wi-Fi) and looking at OS provided network performance graphs.

> Windows: Run the "Task Manager" app, select 'Performance', select 'Wi-Fi'. The
> result is a graph of Wi-Fi performance. Optionally (and a little harder to
> use), right click on the graph, select "View network details", and the result
> is real-time (once/sec) information (example seen right).

FastPing: Another very helpful tool is FastPing (duckware.com), which sends
large pings to your router, measures the round-trip response time, and
calculates Mbps throughput. Very helpful because this tests the reliability of
the full path to/from your main router. Note that no other server is needed
(only your router). FastPing is great at testing modest (say 200 Mbps) speeds
and reliability (when pinging a modern router), but it is NOT for testing
maximum peak speeds.

> TIP: The FastPing manual page shows how to use the built-in Windows+Linux
> 'ping' tool to manually perform very basic throughput testing, using large
> pings. Great for testing throughput to/from IoT devices with limited (72 Mbps)
> Wi-Fi speeds.
> 
> There is a very small 'turn-around' time in the router that is also
> (unavoidably) measured by FastPing. For slow connection speeds, this delay is
> inconsequential. For fast connection speeds, this delay dominates and makes
> the speed less accurate. For this reason, FastPing is great at comparing two
> Wi-Fi connections (fastest speed wins).



Advanced: How to test Router WAN to LAN speeds: What if you have 1 Gbps
internet, AND are able to get true Gigabit wireless throughput -- you don't want
to then find out that you can't access the Internet at gigabit speeds due to a
problem with your router (eg: the Netgear R7800 router has a bug in older
firmware that limits WAN to LAN throughput to 340Mbps over port 80). How to test
WAN to LAN router speed (duckware.com).

WARNING: What to watch out for when running speed tests: Here are some tips on
things to watch out for when running speed tests (and not getting the results
you expected):
 * Multiple computers on Wi-Fi: When running speed tests between two computers,
   one computer must be on Ethernet (with Wi-Fi off) and the second computer
   must be on Wi-Fi. If both computers are on Wi-Fi, you will not get accurate
   speed test results.
 * Multiple-use Ethernet ports: Avoid using dual-functionality Ethernet Router
   ports for speeds tests. For example, a port labeled WAN/LAN. I have seen bad
   speed test results from a WAN/LAN port used in LAN mode. I assume because
   packets are being routed between ports in the OS, instead of being routed
   between ports in hardware (the switch)?
 * Flow Controller: Some routers have a 'Flow Controller' which allows for a
   'pause' frame to be sent out (with the goal of reducing/eliminating lost
   packets). The interesting impact of this is that it can restrict Wi-Fi speeds
   above 1 Gbps to the Router's WAN port speed of 1 Gbps. One tip-off is Wi-Fi
   speed results (using TxRate above) capped at 949 Mbps.
 * Kicked off DFS channel: Your router is configured to use a 160 MHz DFS
   channel, but is currently using a non-DFS 80 MHz channel. Verify channel
   number and width before and after running throughput tests.
 * Buffering: The amount of TCP/IP socket buffering in the speed test program
   (or not) will have a major impact on speed test results. Programs not written
   for 'high performance' in mind may have trouble accurately obtaining and
   measuring speeds above 1 Gbps.
 * Power Plans: If using a laptop, test with both the laptop plugged into the
   power brick and not (on battery). You want to understand if 'on battery' has
   any 'slow down' issues.
 * Slow client devices: Always test Wi-Fi speeds against multiple client
   devices, including workstation class computers. You always want to measure
   true Wi-Fi capabilities, and not end up measuring a client device limitation.
 * Older Routers: Many older routers (even 'high-end' Wi-Fi 5) don't have the
   horsepower to fully max out Wi-Fi capabilities and match throughput results
   from Wi-Fi 6 routers in Wi-Fi 5 mode. Newer often is better.
 * WAN port speed limit: Speeds tests can only run as fast as the slowest 'choke
   point', which can be the Ethernet ports. For example, Wi-Fi on the Netgear
   RAX50 can run much faster than 1 Gbps, but throughput to the WAN or LAN is
   limited to around 949 Mbps by the 1 Gbps Ethernet ports.
 * Give beamforming time to work: After first connecting to Wi-Fi, don't
   instantly run a speedtest. Instead, give 'new' connections a little time to
   'synchronize'.


Appendix E: PHY speed is (often) asymmetric




Background: There are actually TWO PHY speeds for any Wi-Fi device: (1) the Tx
(transmit) PHY speed and (2) the Rx (receive) PHY speed. In many cases, these
two PHY speeds are 'close' to each other, but in some cases they can be very
different.

> KEY Wi-Fi concept: A device's Tx and Rx PHY speeds are independent (and are
> often different).

> To the right is an unusual example of actual measured PHY speeds in real life
> between a router and a laptop computer. Notice that the laptop might report a
> 'good' 270 Mbps for the Link Speed (out of max possible of 300 Mbps), but that
> downloads from the Internet will only use a PHY speed of 216 Mbps! Admittedly,
> this example is unusual, as most often the (higher powered) router can
> transmit at a faster PHY to the PC than the PC can transmit to the router.

Range: If a client device is close to an AP, PHY speeds may not be asymmetric
(by much). But at range, as client devices moves away from an AP, the more the
asymmetry will be seen. Details [§H].

So what is 'Link Speed'?: So is the 'link speed' displayed by Wi-Fi client
devices showing Tx PHY, or Rx PHY? At least on Windows 7/8, it appears to be the
maximum of Tx PHY and Rx PHY. But on Android, it seems to match Tx PHY. This is
complicated and needs a more research.

Router PHY speed for Wi-Fi clients: Some routers display a single 'link speed'
for every client associated with the router. This is (most likely) the Tx PHY
speed from the router to the Wi-Fi client. Or, from the client's point of view,
this is the critical Rx PHY speed we want to know.

Seeing asymmetric speeds in throughput tests: If you don't have a router that
displays Wi-Fi client link speed, the best way to see this asymmetry is in
throughput speed tests [§D].

> The best way to notice and see asymmetric PHY speeds is to place a client
> device 'at range' away from the AP/router being used. The closer a client
> device is to the AP/router, the less you will notice the asymmetry.

The bottom line: The PHY number that clients report is not yet very clear. Is it
Tx PHY, Rx PHY, or a combination of the two? The next best thing is actual
performance throughput benchmark tests, which are a real pain, especially on
smartphones and tablets. So instead everyone just uses and reports Tx PHY speed
as an indicator of device speeds. As an indicator, it works pretty well.




Asymmetric PHY (as seen at AP)

Another example: In another test between a laptop and a router, I found what I
suspect is very typical asymmetric PHY. Running a throughput speed test,
download measured 438 Mbps, but upload measured 200 Mbps. And using the method
described in the Router deep dive appendix [§L] below, I found that MCS 7 (650
Mbps) was being used for download and MCS 4 (390 Mbps) was being used for
upload. Considering that the laptop transmit power (25 mW) is way below that of
the router (200 mw), this outcome is expected. BUT, the 'link speed' reported by
the Windows laptop was 650 Mbps.

Appendix F: PHY speed tables

wiisfi.com/phy

PHY speed tables: PHY tables for Wi-Fi can be found in this online Google docs
spreadsheet (docs.google.com). This spreadsheet is the full raw (but read-only)
spreadsheet on purpose so that you can inspect the formulas that go into
creating every number in the tables!


Useful: I often lookup the PHY speed on a client device, and then find that
speed in the PHY tables, which reveals a ton of information about Wi-Fi on that
client (802.11 mode, MIMO level, modulation, encoding, guard interval, channel
width). When the same value appears in multiple places, usually a little common
sense and deduction about client device capabilities (use
devicespecifications.com or phonescoop.com) can resolve the conflict.

> Most modern clients are capable of 80 MHz channel widths and 2×2 MIMO. If you
> are not seeing PHY speeds indicating that, then you have an issue to
> investigate.



Understanding the math -- a dive into 802.11g: Take a look the 802.11g MCS table
(seen right) from the spreadsheet above.

Symbols per second per subcarrier: Each 'symbol' time is 3.2 microseconds with
an 0.8 microsecond 'guard interval' (GI), or space between symbols. In effect,
each symbol then takes 3.2+0.8, or 4 microseconds of time to send. That means
that there are 1 sec / 4 microseconds, or 250,000 of these symbols per second
(per subcarrier).

SubCarrier: But 250,000 symbols/sec can be sent for each subcarrier, and there
are 48 subcarriers out of 64 that are used for sending data (for 802.11g).
Multiplying (48 × 250,000), we get 12,000,000 -- the total number of symbols
transmitted per second, for a 20 MHz channel, for all sub-carriers.

> A key Wi-Fi concept: Notice that ALL 802.11 a/g Wi-Fi devices operate at the
> SAME 12,000,000 symbols/sec (at the same 'metronome' frequency). It is only
> the 'Modulation' and 'Encoding' that changes and determines true Wi-Fi
> throughput. So an 802.11g device operating at a minimum PHY of 6 Mbps AND an
> 802.11g device operating at a maximum PHY of 54 Mbps are BOTH transmitting
> symbols at a rate of 12,000,000 per second. Using terminology from decades
> ago, this is like saying that the 'baud' rate of 802.11 a/g Wi-Fi is a fixed
> 12,000,000 symbols/sec.
> 
> The same 'concept' applies to all other versions of Wi-Fi as well, but with
> multiple guard interval options. 802.11 n/ac has either 13,000,000 symbols/sec
> (800ns long GI) or 14,444,444 symbols/sec (400ns short GI) for a 20 MHz
> channel.

Modulation bits/symbol ModulationBits BPSK (2)1 QPSK (4)2 16-QAM4 64-QAM6
256-QAM8 1024-QAM10 4096-QAM12

Modulation: BPSK can send 1 bit/symbol. QPSK sends 2 bits/symbol. 16-QAM sends 4
bits/symbol. 64-QAM sends 6 bits/symbol, and so on (see table right). So
multiply 12,000,000 by either 1, 2, 4, or 6 to obtain the total number of bits
transmitted. For example, for 16-QAM, the multiplier is 4 and so 12,000,000
symbols/sec becomes 48,000,000 bits/sec.

> The modulation level (QAM 16, 256, 1024, etc) is just the number of unique
> values that can be 'understood' per Wi-Fi symbol, and bits is the 'power of
> two' that represents that value. For example, 256-QAM can represent any one of
> 256 values, or 28, values -- meaning that 256 is able to transmit 8 bits of
> information.

Encoding: However, some transmitted bits are used for error detection and
correction. That is often represented as a fraction, like 3/4, which means that
four bits are transmitted to encode three data bits. And 1/2 means that two bits
are used to transmit one data bit, etc. So this becomes a fraction that we
multiply the value in the prior step. So if using 1/2 encoding, take 48,000,000
bits/sec from the prior step, multiply by 1/2 and get 24,000,000 bits/sec --
which matches the value (24.00 Mbps) seen in the table for 802.11g, at 16-QAM
1/2.





Wi-Fi bitrate math: To investigate Wi-Fi bitrate math more, refer to the "Data
Rate" formula at the math behind it (semfionetworks.com).

Other sources of PHY information:
 * mcsindex.com - an old resource, very recently redone and updated to include
   802.11ax info
 * Cisco's 802.11ac MCS rates (cisco.com) - 802.11ac MCS rates
 * SemFio Networks (semfionetworks.com) - includes a discussion of the math
   behind the numbers
 * wlanpros.com (wlanpros.com) - interesting as table also includes minimum SNR
   and RSSI values
 * 802.11ac Missing MCS's (youtube.com) - why some encoding combinations are not
   valid
 * Wi-Fi (802.11) PHY Data Rates (blogspot.com) - How data rates are calculated


Appendix G: mW, dBm, dB

Signal Strength mWdBm  1000.0-030  0100.0-020  0010.0-010  0001.0-000 
0000.10-10  0000.010-20  0000.0010-30✔ 0000.00010-40✔ 0000.000010-50✔
0000.0000010-60✔ 0000.00000010-70✔ 0000.000000010-80  0000.0000000010-90 
0000.0000000001-100 

In short: dBm directly represents mW (see table right), but on a logarithmic
scale.

mW: Signal strength in Wi-Fi is all about the mW (milliwatt, or 1/1000 watt).
Most Wi-Fi devices (routers, clients, etc) have a power output somewhere between
25 mW and 1000 mw.

> ✔ Most devices receiving a Wi-Fi signal only 'see' a signal strength of
> 'around' 0.001 mW (close to AP) to 0.0000001 mW (far away from AP).

Received signal strength (RSSI) decreases VERY quickly with distance: Let's say
the power output of a router is 975 mW. Inches from the router, you may have a
signal strength of 0.04 mW. At five feet maybe 0.0016 mW. In the next room,
maybe 0.00001 mW. And across the house, maybe 0.00000001 mW. This is because
signal strength decreases very quickly with distance and obstacles (more details
[§I]). Look at the table (right) for actual values possible.

How should signal strength be represented?: Working with all of these very small
mW numbers, like 0.00000001 mW, is very awkward and error prone -- because are
you using (or reading) the correct number of zeros? Can we come up with a new
unit and numbering scheme to represent mW that is much easier to use?

> First cut: Use scientific notation. 100 mW becomes 1E2, 0.001 mW becomes 1E-3
> and 0.00001 mW becomes 1E-5. Well, we are on the right track because we don't
> have to count zeros, but 'E' notation is still awkward to use.
> 
> Second cut: Use only the exponent. Instead of 1E2, just say 2. Instead of
> 1E-5, just say -5. But we still need to account for the non-exponent
> (mantissa) part, so use logarithms. For example, log(100) is 2, log(0.00001)
> is -5,and log(0.00002) is -4.699. This is workable, but can we eliminate the
> decimal digit...
> 
> dBm solution: Multiply the log(mW) result (from step above) by 10 and round to
> a whole number (no remaining digits). The unit of the resulting number is dBm
> (decibel milliwatts; deci=tenth). The dBm scale is expressly based/referenced
> upon 1 mW (the 'zero' point). The (easy to use) result is the table seen upper
> right.
> 
> 
> 
> "dBm" (decibel milliwatts) still represents milliwatt values (in decibel
> units), but dBm is a MUCH easier way to represent "mW" (milliwatt) values that
> have 'too many zeros'.

mW to dBm: To convert from mW to dBm:

> dBm = 10×log(mW)

dBm to mW: To convert from dBm back to mW:

> mW = 10(dBm/10)

10 dB: mW powers of ten: Just by looking at the table (above right) and the
discussion above, it becomes really obvious that increasing (or decreasing) mW
power by a factor of ten equals changing the dBm value by adding/subtracting ten
dB. Very convenient.

> · Multiplying mW by 10 equals adding 10 dB to dBm
> · Dividing mW by 10 equals subtracting 10 dB from dBm

3 dB: mW powers of two: What do we have to add or subtract from dBm to adjust mW
by a factor of two? The answer is incredibly close to 3. So adjusting by 3 dB is
halving/doubling mW power (but note, not a halving/doubling of distance [§I]).

> We can actually 'deduce' this from the table above. How many 'times 2' steps
> are there in going from 1 to 1000? 1, 2, 4, 8, 16, 32, 64, 128, 256, 512,
> 1024? It turns out that 210=1024, so there are (very close to) ten steps. And
> ten steps from 0 dBm to 30 dBm means the step size is 30/10 = 3 dB.
> 
> FYI: The precise value is 10×log(2) = 3.0102999566...

dB vs dBm: Whereas dBm refers an to absolute power level (translates to a
specific mW value), dB expresses a 'magnitude' difference between two power
levels (the difference between two dBm values; the ratio between mW).

Summary: dBm is just the mW power level in logarithmic scale, but multiplied by
ten. When you see a dBm of -37, you should instantly think that is just a mW of
10-3.7. Think in terms of how many digits the decimal point is moved left/right
for the number "1.0" and it 'should' all make sense. For example, -40 is moving
the decimal point -4.0 places (so left 4 digits), which results in 0.0001.

NOTE: all 'log(#)' in this chapter are log base 10, or 'log10(#)'

More information:
 * dBm (wikipedia.org) - Wikipedia information on dBm


Appendix H: Maximizing Wi-Fi Range



The short answer: The BEST way to improve Wi-Fi range (and far more importantly,
to get great speeds at extended range) is to install an access point [§18]
exactly where it is needed (and wired/Ethernet to your main router).

> Do NOT extend Wi-Fi signals wirelessly. Instead, use a wired Access Point.

> Seriously, because if you do extend Wi-Fi wirelessly, all that accomplishes is
> a 'very slow speed' at distance (that then also uses up a lot of TIME on the
> channel). Namely, after successfully extending Wi-Fi wirelessly, you are then
> left with the problem of slow Wi-Fi speeds at that extended range -- and then
> you are back to square one -- how are you then going to improve Wi-Fi speeds?
> 
> TIP: First confirm that your main router is placed properly [§P] (centrally)
> in your home.

> KEY Wi-Fi concept: Do NOT extend Wi-Fi signals wirelessly. Instead, extend
> Wi-Fi by adding a wireless access point that is wired/Ethernet back to your
> main router.

60 AP's providing Wi-Fi for Cisco in San Francisco

Reduced Wi-Fi range, with a bunch of access points, is how businesses implement
Wi-Fi: Great Wi-Fi performance in "high density" environments is NOT about
increasing Wi-Fi range. In fact, it is just the opposite. It is all about
reducing range and adding a lot of access points.

Seen right is an AP floorplan from a "High density (Wi-Fi installation) with 60
access points, Cisco San Francisco, 4th Floor" as documented in the High Density
Wi-Fi Deployments (meraki.com) paper.

Keep this principle in mind when designing your home Wi-Fi network.

TIP: Test both bands: If your primary goal is long range (and not fastest
speed), split your Wi-Fi band SSID names and first try connecting to your
router's 2.4 GHz band SSID and test. This will only work as long as the 2.4 GHz
band in your area is not too congested (there are not many neighbor AP's
nearby). Next connect to the 5 GHz band and test. Technically, the 2.4 GHz band
'has more range'. on paper than the 5 GHz band, but sometimes in the real world,
the 5 GHz band performs much better due to lower noise floor, less interference,
etc (the key factor is the SINR [§K] that can be obtained, at your location).

> By splitting the Wi-Fi band names, you are taking an automated decision (of
> which band to connect to) away from Wi-Fi devices and manually deciding which
> band to use. The assumption is that you have tested and know which band is
> 'best' (for you, at your location).

Range to an R7800 AP Device2.4 GHz CH 15 GHz CH 36 mWRangemWRange 1000 mW
maximum for AP's Netgear R7800 975 100% 995 100% 250 mW maximum for 5 GHz DFS
channels Google Pixel 4 XL 386 63% 189 44% iPhone 11 279 53% 186 43% Samsung
Galaxy S20 Ultra 122 35% 111 33% Samsung Galaxy Tab S6 504 72% 104 32% iPad Air
583 77% 89 30% Ring Video Doorbell Pro 82 29% 53 23% iPhone 6 Plus 326 58% 49
22% Samsung Galaxy S6 56 24% 41 20% Motorola e5 Play 114 34% 35 19% Ring
Floodlight Cam V1 216 47% n/a n/a Apple Watch 5 209 46% n/a n/a

Client Device Range sorted by 5 GHz Range

Client devices almost always determine maximum range (not the AP/router): What
determines the maximum range at which a client device can successfully
communicate via Wi-Fi with an AP? Assuming that you already have a high quality
AP, almost always, the answer is that the capabilities of the client device (and
not the AP/router) determines maximum range. But why? The short answer is
because Wi-Fi requires two-way communication. From the 'AP to client', and from
'client to AP'. And the weakest of those two directions determines maximum
range.

> Think about it, if you could increase the transmit power in an AP by a
> thousand times, that certainly would improve 'AP to client' communication, but
> that power increase actually would do nothing to help 'client to AP'
> communication, right?
> 
> With an AP using a higher transmit power, the PHY speed (and range) from the
> 'AP to client' would be greatly improved. But the PHY speed (and range) from
> the 'client to AP' would remain unchanged (because the client Tx power level
> is unchanged). So 'at range', clients will see asymmetric PHY speeds [§E].
> 
> TIP: To lookup the FCC ID for your phone, use phonescoop.com to find the web
> page for your phone, and the FCC ID is listed near the bottom of the web page.
> Then Google search the FCC ID and the search result at fccid.io should be what
> you want (displays power levels at various frequencies).



Power Level Model: When an AP transmits to a client, what happens? The AP
outputs a certain mW power, which is then sent to the AP antennas, the signal
travels through the air (lots of 'path loss'), hits the client antennas, and is
received. Likewise, when a client transmits to an AP, the client outputs at a
certain mW power, which is then sent to the client antennas, the signal travels
through the air (lots of 'path loss'), hits the AP antennas, and is received. So
transmit power of both the AP and the client plays a big role in range traveled
(to the client, and to the AP). There is significant signal loss (in the air)
between the antennas, but (usually) a small amount of 'gain' with each antenna.
The wildcard here is that each client device will have a different dBi for its
antennas (as compared to another client device), sometimes very small (or even a
small loss), sometimes larger (5.2 dBi).

> The key for understanding this idealistic model is noticing that what impacts
> received signal strength (in both directions) is: (1) transmit power, (2)
> transmit antenna, (3) path loss, and (4) the receive antenna.
> 
> 2.4 GHz example: The Netgear R7800 router has a transmit power in 2.4 GHz of
> 975 mW and an antenna gain of 0.21 dBi. Source (fcc.gov). The Ring Video
> Doorbell Pro has a transmit power in 2.4 GHz of 82mW and an antenna gain of
> 1.08 dBi. Source (fcc.gov). Since both antennas are used in both directions,
> that is a essentially a 'wash' for comparison purposes. What becomes important
> is that the client has a rather large dB disadvantage of 10.8 dB
> (10×log(975)-10×log(82)). And that means that the Ring cam only has 29%
> (1/sqrt(975/82)) of the range that the R7800 router has.
> 
> > This is why a Ring cam displaying the strong 'RSSI' ('AP to cam' signal) as
> > an indicator of the connection quality is NOT very helpful, as the critical
> > path for the Ring cam is the much weaker 'cam to AP' signal strength
> > (uploading videos).
> 
> R7800 to Galaxy S6 long range
> 
> 5 GHz example: The Netgear R7800 router has a transmit power in the upper 5
> GHz of 969 mW. Likewise, for the Ring Stick Up Cam wired, the transmit power
> is 48.5 mW. The cam therefore has a 13 dB disadvantage -- or 22% of the range
> of the R7800 router!
> 
> 5 GHz smartphone example: A Samsung Galaxy S6 is very similar, and results
> using the MCS Spy tool (duckware.com) can be seen right, and confirm the very
> asymmetric PHY speeds (as 'router to client' power is much higher than 'client
> to router' power).

2.4 GHz vs 5 GHz: The actual frequency (MHz) of the channel used affects range.
Because range goes down as frequency goes up. Using this RF loss calculator
(archive.org) we can see that the difference (ratio) in range is exactly the
same as the difference (ratio) in frequency. For example, in general and at
identical power levels, 5 GHz channels have around 50% of the range of 2.4 GHz
channels, or a 6 dB hit.

> So on paper, the 2.4 GHz band has 'more range' than the 5 GHz band, and in
> real-world testing, that happens often, but not all of the time. It is not
> uncommon to see the 5 GHz band provide just as much range (and MUCH better
> throughput). You just need to test. It is all about the SINR that can actually
> be achieved at a particular location.


12 dB hit (6 dB DFS + 6 dB 5 GHz)

DFS Channels: Using DFS channels causes an AP to use lower power levels
(regulatory constraints). For example, the Netgear R7800 uses 995 mW for non-DFS
5 GHz channels, but uses 243 mW for DFS channels, a change of 6 dB. However,
while this does affect PHY speed used, it rarely affects maximum range as most
clients are already using a mW power level below 250 mW for ALL 5 GHz channels
(so again, the client determines maximum range, not the AP).

> Example: The Netgear R7800 for DFS channels transmits with 243 mW. The Ring
> Stick Up Cam Wired for DFS channels uses 54 mW, a 6.5 dB disadvantage for the
> camera. So the camera ultimately limits range (not the AP).

Channel width: Channel width (20/40/80 MHz) used can also actually impact range.
You can 'technically' double range (but greatly reduce throughput) by switching
from 80 Mhz channels to 20 Mhz channels. Details [§J].

Diversity / Beamforming: Using a high MIMO [§7] router has significant benefits
in improving range and signal strength.





"Can't I increase range by using high-gain antennas on my router?": Yes, very
likely. But increased range in Wi-Fi is actually a huge double-edged sword. Yes,
you might get the extra range you need (but at a slow PHY speed), but that also
means that the router is also seeing all of the Wi-Fi devices in that extended
range! Meaning the likelihood of seeing neighbors' routers and sharing a Wi-Fi
channel with neighbors goes up significantly. And sharing a channel ultimately
means sharing that channel's bandwidth (reduced bandwidth). So unless you
literally have no neighbors, don't use high-gain antennas.

> Instead of thinking "How can I extend the range of my one router", the much
> better question is "How can I improve Wi-Fi signal strength and still maintain
> good bandwidth". The answer is: by installing another access point [§18]
> exactly where it is needed, wired/Ethernet to your main router.
> 
> 
> 
> 
> 
> Also, high-gain antennas are not without consequence. They work by altering
> the 'shape' of the signal. Namely, instead of sending the Wi-Fi signal out in
> all directions (think 'sphere'), a high-gain antenna 'flattens' the signal
> pattern, sending more of the signal out in one direction (eg: horizontally)
> and much less in the other direction (eg: vertically) -- think 'doughnut'.
> This change is likely OK for a single story house, but not for a multi-story
> house. There can be a 'dead zone' directly above/below the high-gain antenna.
> 
> It is not uncommon to find 9 dB or even 12 dB high-gain antennas on Amazon.
> However, I have no idea if they are quality antennas, or not.
> 
> NOTE: With Wi-Fi 6E [§11] routers and later routers, expect only 'built-in'
> antennas that can NOT be changed (done to prevent abuse in the 6 GHz band).

"But I need to improve range of my router!" Are you sure? Re-read above about
the likelihood of sharing a channel/bandwidth with neighbors. Plus, you greatly
increase the likelihood of experiencing the 'hidden node problem'. Instead, an
AP wired/Ethernet to your main router is far superior.

> Assume you do install high-gain antennas. What happens? All you have
> accomplished is adding several client devices at range at very slow PHY
> speeds, causing them to use the bunch of TIME on the channel, potentially
> slowing everyone else (on the same channel) down.

"How about Wi-Fi range extenders?": In general, don't use Wi-Fi extenders [§17]
at all. By definition they consume Wi-Fi bandwidth to perform their job (every
packet is sent over Wi-Fi twice). Instead, an AP wired/Ethernet to your main
router is far superior.

Less range in an AP -- and more AP's -- can actually mean higher throughput:
Counterintuitively, less range in an AP can actually translate to higher
throughput (in areas with a lot of other AP's). With less range, the AP will be
a lot less likely to see a neighboring AP operating on the same channel --
meaning the channel used is 100% yours (instead of the channel and bandwidth
being shared with neighboring AP). And then just use as many well-placed AP as
needed.

Summary: In general, the power levels of client devices are anywhere from 6 dB
to 12 dB below that of your AP -- which means that your client devices (and
which band they use, 2.4 GHz or 5 GHz) ultimately determine maximum range
possible, not the transmit power levels of the AP.

> Also, I can not stress enough the need to test actual throughput [§D] in both
> Wi-Fi bands. The conventional wisdom and 'on paper' calculations show that 2.4
> GHz has 'better range' than 5 GHz, but in the real world, this is NOT always
> the case! It may be true, but it often may not be true. Actual SINR [§K] is
> far more important. I frequently see a big jump in performance switching to
> the 5 GHz band. The ultimate cause is 'noise floor' and interference
> differences. It does not matter if 2.4 GHz has a stronger signal strength if
> that also means a lower SINR!
> 
> At one test location the 2.4 GHz band had a noise floor of -87 dBm, whereas
> the 5 GHz band had a noise floor of -108 dBm -- a huge difference of 21 dB --
> explaining why 5 GHz at this location performed WAY better than 2.4 GHz. You
> simply won't know until you test both bands.

A final note that explains some strange observations: The asymmetric Tx power
levels between a client and AP can cause strange observations -- like Wi-Fi
seems good and then you extend distance a little bit and Wi-Fi drops off a cliff
and no longer works. For example, at near maximum range for a client, download
PHY speeds may still actually be very reasonable and good because it is upload
PHY speed that is 'maxing out' and about to drop to zero.

Appendix I: Wi-Fi signal strength vs distance



Wi-Fi signal strength decreases VERY quickly with distance, even in free space
(with no obstacles). But why?

> Every house around you (likely) has a Wi-Fi router, and there are a very
> limited number of unique non-overlapping Wi-Fi channels. So you are definitely
> sharing Wi-Fi frequencies with some neighbors. But most neighbor Wi-Fi
> activity won't impact you due to the fact that Wi-Fi signals attenuate rather
> quickly. Your router often only sees neighbor Wi-Fi activity as a slightly
> elevated 'noise floor'.

> KEY Wi-Fi concept: The 'quick' attenuation of Wi-Fi signals is what actually
> allows 'everyone everywhere' to successfully use Wi-Fi 'all at once'.

How does mW signal power (seen in router specifications) relate to distance?:
The discussion in the chapter above [§G] was all about raw mW power levels, but
how do power levels relate to distance traveled by a Wi-Fi signal? Naively,
twice the mW power means twice the distance, right? NO. To understand why not,
we must first understand the Inverse-square law (wikipedia.org), which states
that changes in Wi-Fi signal strength are inversely proportional to the square
of the change in distance. For example, three times the signal distance means
1/(3×3) times (or 1/9) the signal power (at that 3x distance).

> Analogy: A great way to visually 'see' and understand this is to consider ever
> increasing spheres. Imagine an antenna at the center of the sphere and the
> surface area of the sphere is the radio signal as it travels outwards (in all
> directions). The (fixed) power output of the antenna must be distributed
> across the entire surface area of the sphere. Seen right are three spheres of
> radius 1, 2, and 3 (so a doubling and tripling of distance/radius). The
> formula for sphere surface area is 4×PI×r2. The critical term to focus on is
> r2. And you can confirm with your eyes that the sphere of radius 3 has a
> surface area that is not just three times larger, but 32 (nine) times larger
> (than the sphere of radius 1).
> 
> TIP: You don't have to memorize the Inverse-square law formula. Instead, just
> remember this 'sphere' analogy and you can easily mentally derive the
> relationship between distance and signal strength.
> 
> 
> Illustration of Inverse Square Law from Wikipedia.org

Formulas: The Inverse-square law (wikipedia.com) states that a change in
distance squared is inversely proportional to the change in power (in mW).
Written literally, we get:

> (new_dist/old_dist)2 = (new_pow/old_pow)-1

And simplifying slightly (re-writing the right hand side inverse term), we get:

> (new_dist/old_dist)2 = old_pow/new_pow

And then cross multiplying, you get the following formula that makes a lot of
'intuitive' sense:

> old_pow × old_dist2 = new_pow × new_dist2

Namely, the 'old' (left) side of the equation is a 'fixed' value. Then on the
'new' (right) side, you can change either term (power or distance), but there
must be a corresponding (and inverse) change in the other term (distance or
power) to keep the equation balanced:

> TIP: The best way to use this final formula is to fill in the three terms that
> you do know, and then solve for the one remaining unknown term.

> KEY Wi-Fi concept: The change in Wi-Fi signal strength (in mW) is inversely
> proportional to the square of the change in distance.

6 dB: Doubling/halving distance: To double/halve distance means we need to
multiply/divide mW power by a factor of four (22), which is in dB units (prior
appendix [§G]) is 3 dB twice, or 6 dB.

> Unless you are working in a pure 'line of sight' environment, walls and other
> obstacles will have a far greater (negative) impact on signal strength than
> distance will.
> 
> BUT, if you have a Wi-Fi analyzer app on your smartphone, it is sure
> interesting to see this actually work in practice. Stand five feet from your
> router, cause internet activity, and then run the analyzer app and check the
> dBm value. Then exit all apps and double the distance from the router (try to
> remain 'line-of-sight') and repeat the process. At least in my testing of
> this, I do see a 6 dB drop in dBm every time I double my distance from the
> router.
> 
> 
> 
> 
> A model of Wi-Fi Distance vs RSSI behavior (line-of-sight; NO obstacles)
> 
> 
> Notice how the power level is adjusted by 6 dB many times before we even get
> to 40 feet away from the router. And then the next 6 dB adjustment doubles
> that distance. The 'sweet spot' for most Wi-Fi connections is between -40 dBm
> (pretty close to a router) and -65 dBm (any farther away and lower throughput
> may be very noticeable).

Example 1: One router has a power output of 975 mW. A second router has a power
output of 216 mw. Everything else being equal, how much farther (distance) can
the higher power router communicate with Wi-Fi clients? Trick question, because
virtually always, the answer is no change at all, because the Wi-Fi clients'
must still transmit back to the router, and client's power level has not changed
at all (and is often slightly less than the router power level) -- so client
power levels often limit distance (not the router).

Example 2: But, in Example 1 above, how much should dBm improve on Wi-Fi
clients, hopefully resulting in slightly better PHY speeds (download) from the
router to Wi-Fi clients, but not a better PHY speed from Wi-Fi clients to the
router (upload). Answer: Up to 10×log(975)-10×log(216) dB, or 6.5 dB.

Example 3: You are 90 feet away from your router and see a -65 dBm. At what
distance should you be able to see a -55 dBm? The power ratio is
10(-55/10)/10(-65/10) = 10. So the distance ratio is sqrt(10) = 3.16, and
solving for distance we get 90/3.16 = 28 feet.

Example 4: At 7 feet from a router you observe a -35 dBm. Estimate at what 'line
of sight' distance you will observe -65 dBm. The answer is that with a
difference of 30 dB, that is 6 dB five times, meaning a doubling of distance 5
times, so 7×25 is approximately 200 feet. Of course, walls and other obstacles
will likely get in the way first and have a greater impact than 'line of sight'
distance.

An observation: The 'distance' you have to move to halve signal strength starts
out very small (very close to the router), but then grows exponentially larger
as you move farther away from the router. Let's say you are 1 foot from a
router. At what distance will signal strength be 1/2 as powerful? Well, distance
must be adjusted by sqrt(2) to keep the equation above 'balanced', so 1.41 feet
(a change of 0.41 feet). Now step 10 feet away and repeat -- the adjustment is
now 4.1 feet. Now step to 100 feet away and repeat -- the adjustment is now 41
feet.

Appendix J: Wi-Fi Channel Width vs Range
Please note that this appendix applies only to Wi-Fi in the 2.4 GHz and 5 GHz
bands, but that it does NOT apply to Wi-Fi in the 6 GHz band -- since Wi-Fi in 6
GHz changes how power levels work (juniper.net).



Channel WidthdBRangeSpeed 20 MHz   100% ×1 40 MHz -3 dB 71% ×2 80 MHz -6 dB 50%
×4 160 MHz -9 dB 35% ×8

Channel width RSSI hit: It is not immediately obvious, but the decision as to
which channel width to use in a Wi-Fi access point (20/40/80/160) actually
alters and affects the 'signal strength' of the received signal -- by 3 dB every
time the channel width doubles. This in turn impacts range and possibly speeds.
Almost all home access points use 80 MHz channels because speed ends up being
far more important than range (and don't even notice, or just live with, the
slightly reduced range and MCS levels used).

> KEY Wi-Fi concept: Wi-Fi in the 2.4 GHz and 5 GHz bands (but not the 6 GHz
> band) requires you to be 'just a little closer' to your router in order to
> successfully connect with increased channel widths using maximum QAM
> modulation.

What a spectrum analyzer displays: Each time Wi-Fi channel width is doubled
(20➤40, 40➤80, 80➤160) in the 2.4 GHz and 5 GHz bands, the SAME amount of
'signal power' is still transmitted (EIRP, or Equivalent Isotropically Radiated
Power), but over double the channel width. A spectrum analyzer (with constant
RBW), shows that doubling channel width results in a 3 dB reduction in received
signal strength: Impact on Range: And from the math above [§G], a reduction in
signal of 3 dB translates to a 30% reduction in range.

How the rest of the Internet sees this behavior: All over the rest of the
Internet you will see this behavior described as the 'noise floor' increasing 3
dB whenever the channel width doubles, and graphs like the following:


But I find this whole concept a LOT easier to understand by thinking about what
a spectrum analyzer actually displays -- which is that when channel width
doubles, the peak signal strength is reduced by 3 dB (because the same amount of
power is now over twice the bandwidth). But note that this behavior changes with
Wi-Fi in 6 GHz (constant PSD of 5 dBm/MHz).

The bottom line: Wider channels reduces SNR: So no matter your point of view is
(noise floor increases 3 dB, or signal strength decreases 3 dB), the end result
on SNR is the same -- that SNR is reduced by 3 dB when channel width doubles.
And depending upon your distance from the access point, this may slightly
(negatively) impact maximum QAM modulation (PHY speed).

> The easiest way to mitigate this reduced SNR is to always be close to an
> access point.

An IoT observation: This explains why many low-bandwidth IoT devices
intentionally want to stick with 802.11n (which has 20 MHz channels) and the
smallest channel width possible and avoid 802.11ac (which mandates 80 MHz
channel support), because that allows operation at the longest possible
distance. But Wi-Fi 6 adds back in the ability for client devices to only
support 20 MHz channels.




Client forcing 20 MHz channel

Maximizing Wi-Fi range/stability (by sacrificing speed): If being able to
connect to a router in 5 GHz at greatest distance -- at any speed -- is far more
important than having the fastest speed possible, configure that access point to
only use 20 MHz channels. Note that Netgear's name for 'channel width' is 'Mode'
[§C]. Clients should now see the ability to communicate with the access point
from a slightly farther distance (albeit at a much slower speed and on a 20 MHz
channel).

> TIP: Alternatively, some client devices (eg: Windows) allow the '5 GHz channel
> width' to be specified/forced to 20 MHz (normally a client would just use the
> channel width of the access point). This has the huge advantage of allowing
> you select an 80 MHz channel on the router (so most clients get 'fast speed'),
> and then only setting 20 MHz channels for those few (far away) clients that
> need 'extra range'. See example (right) from Windows Control Panel / Device
> Manager / Network Adapters / select Wi-Fi adapter.

Client devices often limit range more (than the router): Most mobile (on
battery) client devices do NOT transmit at the maximum power level allowed (like
most routers). Instead, client devices intentionally transmit at a lower power
level to conserve battery power. The end result is that the client device may
often limit maximum distance [§H] from the router (and not the router itself).

> If your client device is able to see a weak router SSID in the Wi-Fi list, but
> is unable to connect to it, the router-to-client signal strength may be OK,
> but the client-to-router signal strength may be too weak.

A final note: This chapter is more about understanding Wi-Fi. Actually changing
the 5 GHz channel width to 20 MHz does extend range (a little), but kills
performance for clients that are close to the router. Try both the 2.4 GHz and 5
GHz bands and select the one that works best for you (as every location will be
different due to different noise floors). The best solution is to NOT 'extend
range', but rather to just install an AP (wired/Ethernet to your main router)
where it is needed the most.

Learning More:
 * Power Spectral Density and Wi-Fi 6E (juniper.net)


Appendix K: SINR / SNR / Noise floor / SNR sensitivity
Everyone will experience a slightly different 'noise floor' (and interference),
which impact SNR, which determines how fast Wi-Fi works for you.

> SINR: When there is lots of interference (eg: ACI), the
> signal-to-interference-plus-noise ratio (wikipedia.org) (or SINR) becomes much
> more important than just SNR. A transmitted signal must be properly 'heard',
> and the noise floor and interference are both critical for determining how
> well that happens.

In summary: The 2.4 GHz band can (not always, but often times) have an
incredibly high 'noise floor' as compared to the 5 GHz band. So while 'on paper'
the 2.4 GHz signal 'goes farther' than the 5 GHz signal, you will likely
experience much better throughput on the 5 GHz band due to this noise floor
difference (a higher MCS level actually used in 5 GHz vs 2.4 GHz).

> KEY Wi-Fi concept: The PHY speed (and throughput) you experience in Wi-Fi is
> determined not by RSSI, but by how far RSSI is above the 'noise floor' -- a
> value called SNR (signal to noise ratio).



Noise floor: When a router tunes into a Wi-Fi channel and amplifies/processes
it, at some point (with no one on the channel), there is only a 'hiss'. The dBm
level of that hiss is the 'noise floor' (blue line graph right). Often times
this 'noise floor' is 'around' -95 dBm to -100 dBm.

> I have seen the noise floor for a band vary by as much as 23 dB (-105 dBm to
> -82 dBm), and the only thing that changed is physical location (of my travel
> router; different State). That is rather interesting, because it implies that
> where you are located (actually how congested Wi-Fi is around you) can
> actually have a VERY measurable impact on the SNR (and Wi-Fi speeds) you
> experience.
> 
> Please note that the 'noise floor' is not a constant value, but is always
> fluctuating (and sometimes by a lot).
> 
> "Noise is defined as any signal other than the one being monitored" - Source
> (wikipedia.org)
> 
> Technically, there are ways to communicate at signal levels below the 'noise
> floor', like LoRa (wikipedia.com), but that is beyond the scope of this paper.
> Modern Wi-Fi relies upon a signal level well above the 'noise floor'.

The noise floor fluctuates (sometimes a lot): The 'noise floor' is often stated
as a single number, but it is NOT a single number -- it actually fluctuates a
lot. Which means that any stated 'noise floor' numbers are actually an average
(or some other computed value).

Signal level: Now start communicating on the channel and examine the dBm level
of the signal on that channel (green line in graph upper right). The dBm level
of that received signal is called RSSI. Often times between -35 dBm and -70 dBm.

> RSSI stands for "Received Signal Strength Indicator". More Information
> (wikipedia.org). And while RSSI officially is a 'relative' number (to itself),
> with no (official) direct relationship to dBm, RSSI is often converted via
> formulas and displayed as just that (a negative number followed by "dBm").

SNR: The difference between the signal level and the noise level is called the
"Signal to Noise Ratio". More Information (wikipedia.org). SNR units are dB. So
this means SNR is a relative number (not absolute number) that indicates how
'loud' a signal is vs background noise. A signal can only be 'heard' and
understood as a signal by a device only as long as it is adequately above the
noise floor.

> Analogy: Talk normally in an empty room, and you can easily be heard (RSSI),
> because the 'noise floor' is so incredibly low (high SNR). But talk normally
> again (so RSSI signal level is the SAME) in a bar, and you won't be heard,
> because the 'noise floor' is so high (low SNR) relative to the signal level.
> The same thing applies to Wi-Fi. As long as the SNR for a Wi-Fi signal is
> 'good', that signal will be heard and understood.
> 
> The implication of this is that regardless of RSSI (even a poor RSSI), a high
> SNR at the same time means high throughput is very likely, but a low SNR means
> that high throughput is impossible.
> 
> I have seen discussions online state (but have not personally verified) that
> in industrial environments, large electrical motors running can cause the
> 'noise floor' to be so high (meaning SNR is always very low), that getting any
> Wi-Fi to work can be very challenging.

Why is the 2.4 GHz noise floor often much higher (worse) than the 5 GHz noise
floor? Because of the incredible success of 2.4 GHz band. Virtually every single
home is broadcasting 2.4 GHz signals, and the noise floor you see (in your home)
is just the greatly attenuated 2.4 GHz signals from all other homes.

> Because 5 GHz signals attenuate more through free space (and obstacles), that
> is providing a noise floor advantage for 5 GHz. I often see the 2.4 GHz noise
> floor at/around -86 dBm. Whereas for 5 GHz, the noise floor is often around
> -105 dBm.
> 
> ★ The implication of this is that the 5 GHz band (even with 'technically'
> shorter range) may in fact provide much faster throughput then the 2.4 GHz
> band. And I say 'may' because I often find that the 5 GHz band is
> better/faster than the 2.4 GHz band, but there are exceptions.

Wi-Fi SNR is underreported: The 'noise floor', 'signal level', and 'SNR' are all
underreported numbers in Wi-Fi. Some higher end vendors report all of this
information, but many vendors report almost nothing. And the reason it is so
important is because if you are in a 'noisy' environment, even with a strong
signal, you will get very poor throughput. Conversely, a weaker signal but with
a very low noise floor, can still get very good throughput. The implication is
that knowing RSSI is a clue for knowing throughput, but you won't know for sure
until you also know the noise floor (and SNR).

> When I travel, I run the Wi-Fi Analyzer on my smartphone to see how many
> access points my phone can see. I never saw a SSID with a RSSI in the mid
> -90's in the 2.4 GHz band, until one day (in the Florida Keys) I saw a RSSI of
> -95 (seen right). My take away from this is that the 'noise floor' at this
> location must have been VERY low in order for a signal at -95 dBm to not only
> be heard, but understood.
> 
> MacOS: An exception is that MacOS displays Noise dBm right next to Signal dBm.
> Very helpful. Use it.

SNR vs channel width: Each time you double channel width (in 2.4 GHz and 5 GHz;
but not 6 GHz) -- from 20 MHz to 40 MHz to 80 MHz to 160 MHz requires 3 dB more
in SNR to maintain the same modulation/coding. Source (arubanetworks.com). And
if you don't have that SNR headroom, the modulation/coding rate will drop (see
Wi-Fi Channel Width vs Range [§J] for why).

> TIP: What this means is that if a device is having trouble maintaining a Wi-Fi
> connection (and can't be moved closer to the router), and you would rather
> have a much slower PHY connection that stays up all of the time (than a faster
> connection to drops in and out) -- try reducing channel width (from 80 MHz to
> 40/20 MHz).
> 
> Now you know one key reason why many low-bandwidth IoT devices stuck to the 20
> MHz channels of Wi-Fi 4 vs the 80 MHz channels of Wi-Fi 5 (they have no need
> for the higher throughput and would rather have increased range).

SNR sensitivity: Here are the IEEE 802.11 SNR that each modulation/coding
requires. Take these as ballpark figures, as your mileage may vary slightly:



Atheros RSSI is really SNR: In routers when you obtain RSSI from system tools
that use a Qualcomm Atheros Wi-Fi chipset, the RSSI value is the dBm signal
level with the noise floor subtracted out. That is just SNR!

More information:
 * 802.11ac Receiver Minimum Input Sensitivity Test (mathworks.com) -- lists
   IEEE table with sensitivity values
 * 802.11ax receive sensitivity requirements (arubanetworks.com)
 * IEEE 802.11-2012 Table that shows SNR requirements per modulation level
   (wirelesstrainingsolutions.com)
 * IEEE 802.11 EVM specification (rfwireless-world.com)
 * Signal-to-Noise Ratio (SNR) and Wireless Signal Strength (meraki.com)


Appendix L: Router deep dive



Many AP/routers have the ability to 'telnet' (or ssh) into the device and run
Linux commands. This chapter documents some of what I use on Netgear's R7800
router (uses the 'Qualcomm Atheros' chipset; immediately below) and Netgear's
R6250 router (Broadcom chipset; farther below).

> This has been invaluable to analyze the Wi-Fi behavior of a 'locked down'
> devices (what channel width, Tx and Rx PHY speeds, etc the cam uses).

Netgear: Enable telnet: On many Netgear routers, visit
"http://routerlogin.net/debug.htm" (and sign in) and check the 'Enable Telnet'
checkbox. Then "telnet routerlogin.net" (LAN only, not WAN) and use the web
interface password to sign in (if asked for 'login' username, use 'admin'), and
then dive into the deep end...

> TIP: When analyzing a device (like a Ring cam), only have that one device
> connect to the router via Wi-Fi (so all Wi-Fi stats on one Wi-Fi band must be
> from the device being tested). And then on your PC, telnet to the router via
> Ethernet (or the other Wi-Fi band).
> 
> WARNING: Telnet is a text based protocol (not encrypted). So only use it on
> 'research' networks (not production networks) where you trust everyone
> currently connected to the network.


--------------------------------------------------------------------------------


Qualcomm Atheros based AP:

athstats -i wifi0|wifi1: [Atheros] Outputs tons of internal Wi-Fi statistics by
band. By far the most useful are "Rx MCS STATS" and "Tx MCS STATS", which
displays the number of packets sent/received for each MCS index (PHY speed)!
Also, lists a noise floor that varies (but trying to confirm it is accurate).
See header source code (github.com) for a short description of items displayed.

> wifi0 is the 5 GHz band and wifi1 is the 2.4 GHz band or vice-versa, use
> iwconfig (described below) to confirm.
> 
> This is a GREAT way to independently measure both Rx PHY and Tx PHY, with no
> cooperation needed on the part of the device being measured! The MCS index
> reveals the (approximate) PHY speed used (don't know guard interval
> differences), and quickly tells you if there are symmetric or asymmetric PHY
> speeds. Seen right is an example where both Tx and Rx PHY are 'mostly' using
> MCS 7.
> 
> In another example, all MCS numbers were pulled into Excel and produced the
> chart below -- which clearly indicates that the cam is able to receive from
> the router much better (ave 35 Mbps) than the cam is able to transmit to the
> router (ave 16 Mbps).
> 
> > 
> 
> And by only changing Wi-Fi channels, cam upload speeds (the majority of
> everything the cam does) improved significantly (now averages 27 Mbps):
> 
> > 

wlanconfig ath0|ath1 list sta: [Atheros] Lists every Wi-Fi device connected to
the router, with very useful information per device (mac address, channel,
TxRate, RxRate, RSSI, 802.11n mode, channel width, etc). wlanconfig.c source
code (github.com).

> The TxRate and RxRate displayed appear to be kilo (1024) based numbers instead
> of bps (1000) PHY based numbers. So multiply by 1024/1000 to correct back to
> 'bps'.
> 
> RSSI displayed under Atheros is actually SNR, which is a very important number
> to know.
> 
> In one dual-band router tested, ath0 was the 5 GHz band and ath1 was the 2.4
> GHz band. Use iwconfig (described below) to confirm the setup for your router.
> 
> OpenWRT: iwinfo wlan0|wl0|ath0 assoclist: source1 (serverfault.com) source2
> (openwrt.org)

iwconfig: Lists 'wireless' information (if any) for each interface on the
system. Includes SSID name, maximum bitrate and transmit power (in dBm units)
for each Wi-Fi band. See also "ip link show", which enumerates all interfaces on
the router.

> This is incredibly useful to clearly 'see' the difference in transmit power
> (in dBm units) between the different bands/channels in 5 GHz (using DFS
> channels in 5 GHz have a 6 dB penalty). Also of note is that these power
> levels do NOT include antenna gain.

arp: Outputs a list of 'MAC address to IP address' mappings.

iwlist: Get detailed information about interface capabilities

A very cool graphical Atheros "MCS Spy" tool: Check out this MCS Spy
(duckware.com) tool, which displays Wi-Fi MCS index usage in real-time.

> It becomes incredibly obvious after running this tool that not only are PHY
> speeds asymmetric, but that there is no single PHY speed in one direction.
> Rather, the PHY speed is constantly fluttering around. This is especially
> evident when the device being tested (eg: tablet) is moving around (with a
> person walking).

Future Research: Often times, the R7800 reports a noise floor of -105 in 5 GHz
and -97 in 2.4 GHz. Does that explain why RSSI in 5 GHz on a client device is
better than raw calculations (of AP transmit power and free space path loss)
show it should be? Or is the difference fully explained by beamforming alone?



--------------------------------------------------------------------------------




Broadcom based AP:

Broadcom based devices will have the wl command Command line reference
(dd-wrt.com). Here are some very useful commands...

ip link show: Enumerates all 'interfaces' on the AP.

wl -i eth1|eth2 status: Displays lots of information about the AP, including
SSID, BSSID, supported rates, HT/VHT capabilities, noise floor, etc.

wl -i eth1|eth2 channels: Displays the list of valid channels.

wl -i eth1|eth2 assoclist: Outputs a list of all client MAC addresses currently
associated with the AP.

wl -i eth1|eth2 sta_info x:x:x:x:x:x: Output detailed statistics for ONE client
device connected to the AP (by specifying the client's MAC address). Example
output:

> aid:1
> rateset [ 1 2 5.5 6 9 11 12 18 24 36 48 54 ]
> MCS SET : [ 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 ]
> idle 10 seconds
> in network 183111 seconds
> state: AUTHENTICATED ASSOCIATED AUTHORIZED
> flags 0x1000e03a: WME N_CAP AMPDU
> HT caps 0x3c: GF SGI20
> tx data pkts: 560610
> tx data bytes: 80242252
> tx ucast pkts: 277309
> tx ucast bytes: 45807854
> tx mcast/bcast pkts: 283301
> tx mcast/bcast bytes: 34434398
> tx failures: 246
> rx data pkts: 217168
> rx data bytes: 210097118
> rx ucast pkts: 211051
> rx ucast bytes: 209558247
> rx mcast/bcast pkts: 6117
> rx mcast/bcast bytes: 538871
> rate of last tx pkt: 58500 kbps
> rate of last rx pkt: 117000 kbps
> rx decrypt succeeds: 1087298
> rx decrypt failures: 0
> tx data pkts retried: 5851
> tx data pkts retry exhausted: 246
> per antenna rssi of last rx data frame: -66 -57 0 0
> per antenna average rssi of rx data frames: -64 -57 0 0
> per antenna noise floor: -89 -92 0 0


--------------------------------------------------------------------------------


Quantenna based AP:

This post (snbforums.com) points to using this command:

> qcsapi_sockrpc CMD wifi0 0

with some of the valid "CMD" being:
 * get_tx_mcs
 * get_rx_mcs
 * get_tx_phy_rate
 * get_rx_phy_rate
 * get_noise
 * get_snr
 * get_rssi_dbm


--------------------------------------------------------------------------------




Linux in general:

A very helpful command is:

> iw dev DEVICE station dump

where 'DEVICE' is 'wlan1' (or similar). You will see information output like:

> signal: -73 dBm
> signal avg: -73 dBm
> tx bitrate: 162.0 MBit/s VHT-MCS 4 40MHz VHT-NSS 2
> rx bitrate: 240.0 MBit/s VHT-MCS 5 40MHz short GI VHT-NSS 2
> and lots more info...


--------------------------------------------------------------------------------





Netgear R7800 TCP/IP packet captures:

I travel a lot and often need to packet capture Wi-Fi devices for debugging
purposes. So I travel with a Netgear R7800 configured not as a router, but as an
AP (access point). This also works around a bug that the R7800 has with packet
captures while in 'router' mode.

Method One: Use TCPDUMP on the R7800: I then plug the R7800 (AP mode) into the
local network/router, connect the devices to test to my AP (Wi-Fi), and start
debugging. My steps for obtaining a packet capture on the R7800 (you can
probably use this as a template for obtaining a PCAP on your router/AP):
WARNING: Netgear just released new .74 firmware that breaks TCPDUMP on the R7800
in AP mode, but the steps below still work with the older 1.0.2.68 firmware.
 1. Setup: Connect the AP via Ethernet to the local network/router. I then
    connect my laptop to the AP to confirm Internet access, and find the IP
    address of the AP (tip: use "ping routerlogin.net" -- and use that IP
    address in place of '192.168.1.222' in the steps below).
 2. Enable telnet access: Visit "http://192.168.1.222/debug.htm" (use the
    router's web interface username/password) and ensure that "Enable Telnet" is
    checked.
 3. Telnet to router: Use "telnet 192.168.1.222" and when asked for the
    password, use the router's web interface password. At this point, you should
    be successfully signed into a telnet session connected to the Netgear R7800
    AP.
 4. Start the packet capture: Type the command: "tcpdump -n -s 0 -i ath1 -w
    /tmp/output.pcap -v", using "ath1" to capture the 2.4 GHz band, or "ath0" to
    capture the 5 GHz band. Press 'CTRL-C' to end the packet capture.
 5. Download the PCAP: I keep a micro USB thumb drive plugged into the AP at all
    times, and use "mv /tmp/output.pcap /tmp/mnt/sda1/output.pcap" to transfer
    the PCAP to the USB drive (use "df" to find your USB drive path), and then
    connect to the "\\192.168.1.222" network share (requires that network
    sharing is enabled for the USB drive on the AP; may need to use
    'admin'/password to access) to download the PCAP file to my PC.
 6. Analyze: I analyze the PCAP with WireShark (wireshark.org).

I PCAP to 'local storage' on the R7800, rather than directly to the USB drive,
because local storage is a lot faster than the USB drive (otherwise a lot of
packets will be dropped if tcpdump writes directly to the USB drive). But the
downside of this is that you can only PCAP a couple hundred MB of activity,
which is fine for my work. An alternative is to capture directly to the USB
drive, but only capture the first 100 bytes of each packet (change "-s 0" to "-s
100"). If tcpdump still reports that packets are being dropped by the kernel,
try increasing the kernel buffer size (-B command line option) -- but the '-B'
option is not available on the R7800.

Method Two: A much better PCAP alternative for most: An alternative that also
works very well (and requires no router telnet access) is use the "WAN Port
mirror to LAN port1" option on the R7800 "/debug.htm" web page. Then, plug your
PC wired to port1 on the router and capture directly (on your PC) using
WireShark (capturing the PC 'LAN connection'). In effect, LAN port 1 on the
R7800 still functions (so your PC still works to the Internet), but your PC now
also 'sees' all WAN traffic (as if the R7800 were now a 'hub', instead of a
'switch'), allowing all traffic (heading out the WAN port on the R7800) to be
captured.

> The big advantage of using TCPDUMP on the R7800 itself (far above), is that
> the capture can be done 'remotely' (don't need to be physically next to the
> R7800).
> 
> Conversely, the big advantage of the port mirroring capture technique is that
> it is much easier for most people to use (and a lot less technical), but it
> requires your PC is plugged into port one on the R7800.


Appendix M: WiGig (802.11ad) 60 GHz
WiGig (802.11ad) is effectively dead in routers for Internet access. It just
never 'took off'.




Dubious Netgear 802.11ad marketing

802.11ad: 802.11ad (wikipedia.org), also called WiGig, is being marketed as the
'fastest' Wi-Fi possible, providing speeds "as fast as 4.6 Gbps", for '4K
Streaming, VR Gaming and Backup' (Netgear, right), or for transferring an hour
of HD video in 7 seconds. Source (archive.org/qualcomm.com).

Huge disadvantage: However, the huge disadvantage of 802.11ad is that is has no
range and does not go through walls (or obstacles). It is intended to only be
used line-of-sight in one room and has a range of just a few meters. Source
(radio-electronics.com).

Range: At the same transmit power, 5 GHz has 1/2 (50%) the range of 2.4 GHz, but
60 GHz has 1/25 (4%) the range of 2.4 GHz, as measured in free space, or 'air'.
TIP: See this RF loss calculator (archive.org)

802.11ac: Interestingly, 802.11ac products already exist TODAY that provide 4.3
Gbps -- Arris TG3482G using the Quantenna QT10GU (quantenna.com).

That kind of puts Netgear's marketing hype ("3X faster than 11ac") into
perspective.

Conclusion: 802.11ad may take hold in very specialized situations (laptop docks,
wireless displays, VR headgear, etc), but unless the range issue is addressed,
802.11ad will absolutely NOT become a replacement for Wi-Fi for generalized
internet access for an entire home.

> UPDATE: This is all but confirmed now that 802.11ax (Wi-Fi 6) -- the successor
> to 802.11ac (Wi-Fi 5) -- is out. Wi-Fi 6 effectively kills 802.11ad from ever
> being widely adopted for internet access. Instead 802.11ad IS being used for
> very short distance point to point (laptop computer to dock) communication.
> 
> Also, you know things are VERY BLEAK for 802.11ad in routers when Netgear has
> a tough time pointing out ANY client devices (netgear.com) that actually
> support WiGig!


Appendix N: Beware tri-band marketing hype

UPDATE: Wi-Fi 6E tri-band routers DO make a lot of sense: The key problem with
prior Wi-Fi 5 tri-band routers was that TWO of the bands are the same
overlapping frequency (5 GHz). However, Wi-Fi 6E routers will be tri-band and be
just fine -- because they will cover 2.4 GHz, 5 GHz and 6 GHz -- in other words,
three separate non-overlapping frequency bands.



Linus video on tri-band (youtube.com)

Beware all of the marketing hype surrounding tri-band routers. Tri-band routers
were created by the router industry so that marketing could (yet again) claim
even higher Wi-Fi Gbps speeds for new routers (a single connected client can
never achieve these high speeds).

A router is many devices in one: Remember, a wireless router is: (1) a router,
(2) a switch, and (3) an AP -- all in one box. The AP is almost always dual-band
(2.4GHz + 5 GHz). But the latest marketing hype concerns the speeds of tri-band
routers -- where the AP inside the router is 2.4GHz + 5 GHz + 5 GHz.

But 'dual 5 GHz' is only useful if you are maxing out your current 5 GHz band,
and need to support more (5 GHz only) devices. But are you? This is important to
realize, It DOES NOT make one device faster. Rather, it allows 5 GHz devices
connected to different 5 GHz bands to operate at the same time (two devices
connected to the same band will have the same problem).

RF Interference: When there are multiple antennas in the same AP/router
operating on the same band (5 GHz), near to the same frequency, there WILL be
interference, which can reduce data rates. This mainly happens when one band is
receiving and (at the same time) the second band is transmitting. One mitigation
is a proper separation in channel numbers (the wider the better; at least three
times channel width) because if not, there will be too much interference. The
ideal fix is physical separation of the AP's. And that is why simply using a
second AP (physically separated and located where it is really needed) is the
best solution. Dual 5 GHz AP criticism (youtube.com).

> And since (virtually) all Netgear routers are missing support for channel 138,
> that may limit your options running 'dual 5 GHz'.



Instead of a single tri-band router, just use an AP instead: BUT, if you are
this situation, if is far more cost effective to just add a second AP [§18],
wired/Ethernet to your existing router, rather than throwing everything away and
buying only a tri-band router (for around $500) placed in a central location.
Plus, the advantage of a second AP is that you can physically position it
(separate from the router), on its own channel, exactly where it will do the
most good.

> And physical separation between AP's make a lot of sense. Would you buy two
> AP's (same band) and place them literally on top of each other? Of course not.
> You would spread them around where they are needed. But with a tri-band
> router, that is exactly what you are doing (placing two AP's on top of each
> other).

Don't fall for the marketing hype! If you need a new router and can get a deal
on a router and it happens to be tri-band, go for it. But upgrading from a high
end dual-band to a tri-band because you think it will be a lot faster because
the Mbps rating is much higher is not a good thing.

Tri-band without full DFS channel support is insane: There are vendors that sell
tri-band routers without any DFS channel support! Without DFS channels, there
are only TWO 80 MHz channels available in 5 GHz. A tri-band router must then use
both of the only available channels. You have no choice. That dramatically
increases the likelihood of sharing a channel (and bandwidth) with a neighbor.

Be practical: Don't go out of your way looking for a tri-band router. But if you
just happen to find a capable one at a fantastic price, go for it. Just realize
that you have two AP's in a single location (that you can not physically
separate), and keep the channels for the two AP's as far apart as possible.

Appendix O: Extending Ethernet/Wi-Fi to an Outbuilding
From this paper, you already know the best way to get great Wi-Fi in an
outbuilding is to add an Access Point [§18] in the outbuilding, wired/Ethernet
back to your main router. But what is the best way to extend Ethernet from your
main router to the outbuilding?

The best way is a wired solution (immediately below). But if you can't (or don't
want to) use a wired solution, then look at wireless solutions (further below).

◆ Wired Solutions

The 'not recommended' way: The easy way is to run Cat5+ underground, to the
outbuilding, as long as cable length is under the 100m (330ft) limit. That can
actually work very well, but with any copper wire run outside, you need to worry
about power surges caused by lightning. It will happen at some point -- it is
just a matter of time. If you still want to run Cat5+, keep these tips in mind:

> Use only 'Direct Burial' gel filled cable note: this is different/better than
> just 'outdoor' rated cable. The outside cable jacket is designed to survive
> not only UV exposure, but also contact with the ground and water. Plus the
> inside of the cable is filled with a gel to prevent water ingress, if the
> cable jacket gets nicked/damaged.
> 
> Bury the cable in PVC conduit for a second level of protection. OR (less
> protected) use the edge of a shovel to create a couple inch deep slit in the
> soil and lay the cable in -- this very quickly shallow buries the cable.
> 
> To hopefully prevent damage to equipment during surge events, use (a) shielded
> cable, (b) shielded keystone jacks, (c) shielded patch cords, and (d) an
> Ethernet Surge Protector on each end of the cable Example kit (two devices)
> (amazon.com) protecting to 20KA. Please note that protection devices require
> an actional connection to ground in order to work. Any excessive voltages
> induced into the cable must be shunted to ground at either end of the cable.
> And in the end, the cost of all of this protection is way more than a fiber
> solution.


1 Gpbs Media Converter


Fiber Cable

BEST (1 Gbps): Use Fiber to extend Ethernet: If you need to run Ethernet between
any two buildings, just use fiber cable, buried in conduit, and media converters
(converts fiber to Ethernet) on each end of the fiber cable. On Amazon, there
are very affordable pre-made cables (you select length) and 1 Gpbs media
converter kits. as a ballpark figure, around $90 will get you 100 ft of fiber
cable and two media converters. For example:
 * fiber cable (amazon.com) -- pre-made (ends installed) armored, outdoor-rated,
   fiber cable, in various lengths from 1m (3ft) to 1000m (3280ft)
 * media converters (amazon.com) -- a two-pack kit of 10Gtek brand media
   converters with 10Gtek brand SFP transceivers (1 Gbps).
   
   > Or get two of this mini media converter (amazon.com). But use extreme
   > caution: don't click on 2 'Number of items' as that changes item from SMF
   > to MMF. Instead, add item to cart and change Qty to 2.

Fiber optic cable is immune from EMI (Electromagnetic Interference) -- so there
are never any worries about lightning and induced voltages, because fiber is
non-conductive glass. This is a big advantage fiber has over copper.




10 Gpbs Media Converter

BEST (10 Gbps): This is the same as above (same fiber cable), but instead of a 1
Gbps media converter kit, get two 10 Gbps media converters (amazon.com) --
supporting (a) fiber to 10km and (b) 10GBase-T over Cat6a to 30m (98ft).

Dual strand vs BiDi: Why install dual (two-strand) LC fiber instead of a BiDi
(single-strand) LC fiber solution? -- Because: (a) dual fiber is incredibly
common and only fractionally more expensive than a single fiber cable -- so you
are NOT saving that much by selecting BiDi cable, (b) if you want a 10 Gbps (or
eventually higher) link instead of 1 Gpbs over BiDi, good luck finding a BiDi
kit solution on Amazon. Whereas with dual fiber, those solutions are very common
(amazon.com) right now, and (c) as a bonus, a dual fiber cable can be converted
to two BiDi links, if needed.

Some fiber jargon: Because most Fiber 'home use' is not going to need
'data-center' class speeds, that makes 'Dual LC single-mode fiber cable' the
choice for home use at 1 Gbps and 10 Gbps speeds, due to the very low cost kits
that are available. Some terminology:
 * SFP: "Small Form-factor Pluggable". A fiber cable plugs into the SFP
   transceiver, which in turn plugs into the SFP port on the media converter.

 * 
   
   LC / SC / ST / etc: Refers to the type/size of connector (wikipedia.org). LC
   is very common now, a smaller connector than SC (see right)
 * SMF / MMF: Abbreviations for 'single-mode' fiber (wikipedia.org) and
   'multi-mode' fiber (wikipedia.org). The fiber cable for SMF is less expensive
   than MMF, but transceivers for SSF are more expensive than MMF. MMF is
   limited to hundreds of meters, whereas SMF is not.
 * BiDi: BiDi mean 'bi-directional', using a single fiber strand to Tx and Rx,
   instead of two fiber strands (one Tx; one Rx).
 * Dual LC: refers to the LC connector (seen right) with two LC connectors next
   to each other.
 * Duplex Fiber: a fiber cable with two strands (often one for Tx and other for
   Rx)

Conduit TIPS: In you are going to all the trouble (and time!) to trench and use
conduit, don't just put the one single cable you need into the conduit. Instead,
also add spares and/or backup cables. For example, add Cat5+ as well (and
terminate/test) -- but don't connect to anything. Maybe even add a spare RG6
cable? This way you have backups in case the fiber line goes down. You can
quickly switch to Cat5+ until the fiber problem is fixed or even MoCA over the
dedicated RG6. The entire point is that 'spares' gives you some options down the
road, that you may not be able to see today.
 * Consider using large diameter conduit (like 2" or 3") to make it much easier
   to pull new cable in the future
 * Add a 'pull-line' into the conduit for future use. It is very difficult to
   pull cable in small diameter conduit.
 * See 811-In-Your-State (call811.com). Every state has a free 'mark buried
   utilities' service before you dig/trench.
 * Add a 3" 'buried line caution tape' into the trench a couple of inches above
   the conduit as the trench is refilled with dirt.
 * Still use 'direct burial' rated cables in conduit. The conduit is there for
   extra protection, NOT primary protection.

UeeVii Wireless Bridge

◆ Wireless Solutions -- Line of Sight

OK: Wireless Bridge: For various reasons, sometimes there is no way to bury a
cable between point A and point B. When this happens, use a wireless bridge. For
example, the UeeVii Wireless Bridge (amazon.com), for around $190. These are now
'commodity items' and you will find a ton of brands and options on Amazon.

> Depending upon model and distance, expect throughput in the 100's of Mbps, and
> be only half-duplex (only one direction at a time).

Mikrotik Wireless Wire

BEST: 1 Gbps Wireless Wire: If you need/want full-duplex 1 Gbps Ethernet
wirelessly (both directions at the same time), check out the Mikrotik Wireless
Wire (amazon.com), for around $230 for a pre-made (devices already paired)
solution, and powered via standard 802.3af / 802.3at PoE -- specifications
(mikrotik.com)

> This kit "replaces a Gigabit ethernet cable with two small devices that
> connect to each other over a 60 GHz wireless link", and claims 100m (330ft) to
> 200m (660ft) 'line of sight' range.
> 
> "The box includes two wAP60G devices that are already paired together, a wall
> mounting kit, straps for pole mounting and also a pair of table stands for
> using the devices indoors."

Longer wireless distances: MicroTik has quite a few other 60 GHz products
supporting line-of-sight distances up to 1500m (4900 ft), and many other vendors
also have 60 GHz products.

◆ Other options

If you (a) can't bury a cable and (b) you don't have 'line of sight', other
options to consider:
 * Try powerline networking [§E8]
 * Is there a way to run an aerial cable overhead?
 * Cellular 4G LTE or 5G "Mobile Hotspot"
 * Satellite Internet


Appendix P: Router/AP Placement


Place the Main Router as 'centrally' as possible

Main Router Placement: Imagine straight lines from the antennas on your main
router to all of your Wi-Fi client devices. Place your main router in a location
that:
 * Minimizes distances: The total length of all imaginary lines are minimized.
   Often times, this means placing your router in as 'centralized' a location as
   possible in the house, and often near where you spend a lot of time, like in
   a living room.
   
   
 * Avoids obstacles: All the imaginary lines to all Wi-Fi client devices should
   avoid as many obstacles as possible (appliances, furniture, walls, bookcase,
   full closets, etc). Ideally, the Wi-Fi signal is only hitting walls and air,
   and nothing else.

> KEY Wi-Fi concept: Place your 'main router' centrally in your home, and out in
> the open.

Do NOT 'hide' your router: You want your router 'out in the open'. Do not place
your router inside a cabinet, or on the floor under furniture. Instead, place
the router so that it is very visible from all directions.

> I was once at a rental house where the Wi-Fi router was in a piece of
> furniture, in a closet (with a door), in a bedroom, located along an outside
> wall of the house. So half the house had a poor Wi-Fi signal and the far half
> of the house had NO Wi-Fi signal. The solution would have been to move the
> main router to the middle of the house (like the living room), and place the
> router out in the open.

Don't overlook also adjusting router 'height': Don't restrict yourself to only
moving the router left/right. Also consider moving the router up/down (eg: on
top of a shelf or hutch). It is all about finding the best location that best
minimizes the most obstacles for 'at range' client devices.

Placing Mesh/Extender Nodes with a 5 GHz wireless backhaul: Place the node so
that the PHY speeds for each of the two paths (router to node, and node to
furthest client device) are approximately the SAME PHY speed.

> The "1/2 rule" for supporting far-away Wi-Fi 5 devices: Discovering the PHY
> speed of the two paths may be very hard, so as a 'general rule', place a
> mesh/extender node roughly half way between the main router and the furthest
> modern Wi-Fi 5 client that you must support -- but also place to minimize
> obstacles! Take advantage of long hallways where there are no obstacles to
> degrade the Wi-Fi signal.
> 
> The "2/3 rule" for supporting far-away older Wi-Fi 4 devices: If you are
> trying to improve Wi-Fi for a far away Wi-Fi 4 device (eg: many Ring cameras),
> try placing a modern mesh/extender node 2/3 of way from the main mesh router
> to the device (camera). This is because you want the 'device to mesh' link to
> have great connectivity (a PHY of around 72 Mbps, which is the max), and you
> want the 'mesh/extender to router' link to have OK connectivity (PHY max
> around 866, but you want anything above 72 Mbps).

Placing Mesh/AP Nodes with a wired backhaul: Since there is a reliable backhaul
to the main router, think of large similarly sized spheres (the Wi-Fi signal)
emanating from each node in your network -- and place all nodes so that (a) each
Wi-Fi client potential location is covered, and that (b) all spheres just barely
touch.

> OR, just place the AP's where they provide the maximum value for the most
> client devices, most of the time. Namely, in the living room, in a kids
> playroom, or in the home office.

Obstacles impact Wi-Fi signal strength MUCH more than air/distance: What
degrades Wi-Fi signal strength the most are the obstacles that the Wi-Fi signal
must pass through, not air/distance. A Wi-Fi signal can easily go hundreds of
feet without any obstacles, but as soon as you start adding walls (and anything
else), signal strength degrades very quickly.

> So if you can place a router so that the Wi-Fi signal avoids as many
> obstacles/walls as possible, that can help a far away client device (like an
> outdoor Ring camera).




Metal objects can interfere with your Wi-Fi signal: I was in a rental house
where Wi-Fi performance was horrible in the 5 GHz band (interestingly, the 2.4
GHz band was OK). The owner had actually even gone to the trouble of installing
a VERY expensive Mesh Wi-Fi system (even though the home at around 2000 sq ft
did not need a mesh system) and even that did not improve Wi-Fi performance.

The ultimate root cause/problem? The Mesh base station (and previously, router)
was placed on the floor and 18" away from a decorative metal stool (seen right),
and that was causing interference.

> KEY Wi-Fi concept: Do NOT place a router/AP on or near metal objects.

Simply moving the metal stool well away from the router and to a different part
of the room instantly and dramatically improved Wi-Fi performance throughout the
entire home.

> I replicated the 5 GHz performance problem by placing a non-mesh Netgear RAX50
> router on top of the stool. Wi-Fi performance was bad, but as soon as the
> stool was removed (replaced by a chair made of wood), Wi-Fi performance was
> great.
> 
> The lesson learned: The objects near/around your router can and will impact
> your Wi-Fi signal.


Wi-Fi hidden node problem

Warning: Hidden Node Problem: Placing a router 'centrally located' often works
best, but it can backfire if you have two client devices both communicating
frequently with that router that are 180° far apart from each other (and Wi-Fi
wise, can't 'see' each other -- example seen right). See Hidden Node Problem
(wikipedia.org).

> When this happens, add an access point [§18] (wired/Ethernet back to your main
> router) to your network to eliminate the 'hidden node'.

TIP: Rotating the router can sometimes help (a little): The Wi-Fi signal range
radiating out of a router is almost certainly not a 100% symmetrical sphere.
Instead, think of a sphere with indentations. This means that simply rotating
the router 90°, 180°, or 270° can sometimes improve the signal strength for a
key far-away device like a Ring camera outside just enough to be helpful. But
beware -- if this technique actually works and helps you, that the 'bad Wi-Fi
signal' will also be rotated the same amount -- and moved to a different part of
the house, or hopefully outside the house, where it does no harm.

TIP: Antenna orientation: Sometimes moving the antennas on the router slightly
can improve the signal strength for a far-away client device. You just need to
try and test.

> Using the MCS Spy tool (duckware.com) to confirm MCS indexes used in
> real-time, I was able to change the MCS level used by a far away device from
> MCS index 1 to MCS index 2 (doubling the speed) simply by moving the antennas
> on my AP slightly.

More information:
 * The Ars Technica semi-scientific guide to Wi-Fi Access Point placement
   (arstechnica.com)
 * ASUS guide on placement of mesh nodes (asus.com)


Appendix Q: What 'Stream' means -- has changed!
Changing definition: Watch out! What "Stream" means -- to router companies --
has changed from Wi-Fi 5 (and before) to Wi-Fi 6 (and later)!

> And marketing hype is directly to blame. It is just another marketing ploy to
> make new routers sound 'so much better' than prior generation routers.

Wi-Fi 5 (and before): The number of 'streams' in older routers used to mean MIMO
level. The Netgear R7800 is a 4×4 MIMO router and is was called a "Quad-stream"
router -- supporting 4×4 in 2.4 GHz AND 4×4 in 5 GHz:


Quad stream and 4×4 MIMO Wi-Fi 5 router
Wi-Fi 6 (and later): However, router companies are now adding the number of
streams (MIMO levels) for all bands together to come up with an inflated
'stream' number. The Netgear RAX35 is 2×2 MIMO, but is now called a "4-Stream"
router -- supporting 2×2 in 2.4 GHz and 2×2 in 5 GHz:


4 stream, but only 2×2 MIMO Wi-Fi 6 router
So the first router is "Quad-Stream" and the second is "4-Stream" -- so similar
specs, right? No! The first is 4×4 MIMO (in both bands) and the second is only
2×2 MIMO (in both bands). By the standards that router companies are now using,
the first "Quad-Stream" router is actually now an "8-Stream". What 'Stream'
means has changed!

MIMO level is the key, not 'streams': What really matters to you is maximum MIMO
level for the single Wi-Fi band that most of your client devices will use.
Today, that means you want 4×4 MIMO support for all Wi-Fi bands that the router
supports.

So just do your own research and confirm the MIMO level supported for each band.
Some companies make this information very hard to figure out (not even
disclosing it; it must 'computed' from other disclosed information). Whereas
other companies (like Asus), fully disclose that information under a "Tech
Specs" router section:

> 


Appendix R: Learn More
Learn more on other web sites:
 * WiFiGuys Blog - many interesting blog articles on Wi-Fi
 * SmallNetBuilder - Lots of interesting information (but not updated much
   recently)
 * DeviWiki.com - mirror of discontinued 'WikiDevi'
 * TechInfoDepot - great site for router FCC numbers; chipsets, etc
 * DeviceSpecifications.com - great site for finding smartphone Wi-Fi
   capabilities
 * PhoneScoop.com - tons of specification information on phones (like FCC ID)
 * FAQ - Basic Wireless Concepts - TP-Link (PDF)
 * Dong Knows Tech - many reviews, but heavy advertising
 * WiFi Ninjas - many interesting blog articles on Wi-Fi

Vendor provided FCC compliance information:
 * Ubiquiti FCC ID for all products
 * Netgear notice of FCC compliance (but there are DFS support errors in this
   doc)
 * Eero

Vendor filings with the FCC: A great way to new products that will be out soon.
 * TP-Link Hong Kong FCC filings - 2AXJ4
 * TP-Link Signapore FCC filings - 2BCGW
 * TP-Link China FCC filings - TE7 (no new filings since May 2022)
 * ASUS FCC filings - MSQ
 * Netgear FCC filings - PY3
 * Ubiquiti FCC filings - SWX
 * Apple FCC filings - BCG
 * Linksys FCC filings - Q87 (now under Belkin, as Belkin bought Linksys)
 * D-Link FCC filings - KA2
 * Belkin FCC filings - K7S

Wireless LAN Professionals:
 * Interesting videos on their YouTube channel

Virtual web GUI emulators:
 * TP-Link

UPDATE: Some of the Aerohive posts below have been moved to the very bottom of
this web page.

Book: All other posts were moved into David Coleman's Wi-Fi 6/6E for Dummies
eBook (requires form to be filled out). Alternatively, look at this 2nd source
for the book (no form to fill out).

Aerohive Tech Articles: Aerohive Wi-Fi 6 802.11ax technical blog articles --
sadly, all of these links now appear dead after Aerohive merged with another
company:
 * How Does 802.11ax Address Common Problems With Wi-Fi?
   
 * What Are The Goals of The 802.11ax Standard?
   
 * What Does 802.11ax Change About The WLAN Standard?
   
 * What Is BSS Coloring In 802.11ax?
   
 * How Will Target Wake Time Help Mobile Devices And IoT In 802.11ax?
   
 * Why Is OFDMA One Of The Most Important Features In 802.11ax?
   
 * 802.11ax Frame Aggregation Enhancements
   
 * What Does Multi-User (MU) Mean?
   
 * How Do 20 MHz-Only Clients Operate In 802.11ax?
   
 * 802.11ax and Medium Contention
   
 * What is BSS Color in 802.11ax?
   
 * How Does BSS Coloring Work in 802.11ax?
   
 * Dueling NAVs in 802.11ax
   
 * Will 802.11ax Make 80 MHz and 160 MHz Channels Usable in the Enterprise?
   
 * Does The Number of Spatial Streams in 802.11ax Really Matter?
   
 * What Benefits Does Dual 5 GHz Bring To 802.11ax?
   
 * The Main Ingredient of 802.11ax: OFDMA
   
 * What are OFDMA Resource Units in 802.11ax?
   
 * How Does an 802.11ax AP Allocate OFDMA Resource Units?
   
 * OFDM and OFDMA Subcarriers - What Are the Differences?
   
 * How Does DL-OFDMA Work in 802.11ax?
   
 * Unsolicited Buffer Status Reports in 802.11ax and Wi-Fi 6
   
 * What is UL-OFDMA Random Access (UORA)?
   
 * What is UL-OFDMA Random Access (UORA)? - Part 2
   
 * 6 Things to Expect from 802.11ax (Wi-Fi 6)
   
 * Why OFDMA is the Secret Sauce for Wi-Fi 6
   


Appendix S: Spectral Masks


Screenshot from 'RF-Explorer Handheld Spectrum Analyzer'

The spectrum analyzer graph (seen right) shows the entire 70 MHz spectrum of the
2.4 GHz Wi-Fi band (2402 MHz to 2472 MHz), with a single Wi-Fi client device
broadcasting on 20 MHz Wi-Fi channel 6.

> The X-axis left to right represents increasing Wi-Fi frequency (channels 1
> through 11).
> 
> The Y-axis represents received signal strength (so bottom is the noise floor).

Notice that the client device -- transmitting on 20 MHz channel 6 -- is actually
impacting the entire 70 MHz of the 2.4 GHz Wi-Fi spectrum!

Sideband signal: The 'hump' in the center of this graph represents the main 20
MHz channel 6 transmission signal. And then there is a quick drop off followed
by a much more gradual drop off -- aka the "sideband" signal. This sideband is
unavoidable and just how RF transmissions work.

> There is the strong 'main signal', and then an unavoidable, and much weaker,
> sideband signal, and Wi-Fi devices are required (by law; FCC rules) to stay
> within a well defined 'spectral mask' that limits the amount of sideband
> signal that can be generated (see dotted line in graph below right).


Spectral Mask (dotted line) from ResearchGate paper


The graph above was intentionally captured three feet from the transmitting
device, to show the full sideband signal. Because as a receiver moves farther
away from the transmitting device, the more this entire graph (and especially
the sideband signal) gets 'pushed down' into the noise floor (signal
attenuation), helping to greatly limit the impact of any 'sideband" signal on
adjacent channels.

> And this explains why you don't want radio devices operating on
> similar/adjacent frequencies too close to each other -- you want to avoid the
> interference from sideband signals! This is why you need to be very careful
> using tri-band routers (where two of the bands are on the same band).
> 
> In most cases, where devices are on adjacent channels, but operating far
> enough apart, the sideband signals drop into the noise floor, leaving only the
> main hump above the noise floor.

The real-world impact of Adjacent Channel Interference (ACI): When a RF
transmitting device is operating too closely to the radio receiver operating on
an adjacent channel, the 'sideband' transmissions are no longer below the
receiver noise floor, but rather are above the noise floor, and can cause
significant interference (ACI) in the adjacent channel receiver (lowering SINR),
resulting in reduced MCS levels (throughput). Here is an example of a forced ACI
situation and the impact (in red) that had on receiver channel throughput:


The impact (in red) of Adjacent Channel Interference (ACI) is reduced throughput

> Analogy: You are outside having a conversation with your neighbor (your
> channel). A loud motorcycle one block away (far away channel) is not going to
> impact your ability to have a conversation (you want/need that signal
> attenuation). But if the loud motorcycle comes down your street (an adjacent
> channel), it will greatly impact your ability to have a conversation for a
> short period of time.

Learn More:
 * Google "Wi-Fi Spectral Masks" (google.com)
 * WLAN 802.11ac Spectral Mask (rfwireless-world.com)
 * Wi-Fi: Overview of the 802.11 Physical Layer and Transmitter Measurements
   (tek.com)


Appendix T: Terminology
 * 2.4 GHz / 5 GHz / 6 GHz / 60 GHz: Refers to the wireless frequency
   (spectrum/band) used by Wi-Fi.
 * 802.11n: Wi-Fi 4: The specification for HT (High Throughput) Wi-Fi (mainly)
   in the 2.4 GHz band (also operates in the 5 GHz band).
 * 802.11ac: Wi-Fi 5: The specification for VHT (Very High Throughput) Wi-Fi in
   the 5 GHz band.
 * 802.11ad: The specification for Wi-Fi around the 60 GHz band.
 * 802.11ax: Wi-Fi 6: The specification for HE (High Efficiency) Wi-Fi 6.
 * 802.11be: Wi-Fi 7: The specification for EHT (Extremely High Throughput)
   Wi-Fi 7.
 * ACI: Adjacent Channel Interference.
 * AFC: Automated Frequency Coordination. A part of Wi-Fi 6E that allows an
   access point to obtain a "list of available frequency ranges in which it is
   permitted to operate and the maximum permissible power in each frequency
   range".
 * AP: AP is the acronym for (Wireless) Access Point. This allows your Wi-Fi
   devices to connect to a wired network (which is connected to the Internet).
 * AC####: AC refers to support for 802.11ac (Wi-Fi 5) and #### is the sum of
   the 'maximum PHY network speed' for ALL bands in the router (like dual-band
   or tri-band). This naming convention is very deceptive because it can imply
   faster speeds where no faster speeds exist.
 * AX####: AX refers to support for 802.11ax (Wi-FI 6) and #### is the sum of
   the 'maximum PHY network speed' for ALL bands in the router (like dual-band
   or tri-band). This naming convention is very deceptive because it can imply
   faster speeds where no faster speeds exist.
 * Backhaul: Refers to how a node in the network communicates with the main
   router in the network -- is it 'wireless' or 'wired' (Ethernet).
 * Beamforming: A standards-based (802.11ac/802.11ax) signal-amplification
   technique that results in increased range and speed to a device. Beware
   (avoid) earlier (proprietary) 802.11n beamforming implementations.
 * Cat5+: Shorthand notation referring to any category cable at/above Cat5e. So,
   Cat5e, or Cat6, or Cat6a, etc.
 * Client: Refers to any Wi-Fi device (phone, tablet, console, TV, etc) that
   connects to an access point (AP).
 * DHCP: Dynamic Host Configuration Protocol -- the protocol used by a client
   device to obtain an 'IP Address' when it connects to a network. wikipedia.org
 * Dual-band: Two access points in one. Often band one is 2.4 GHz and band two
   is 5 GHz.
 * DFS: Dynamic Frequency Selection. Routers that use DFS channels in 5 GHz must
   scan for conflicts (TDWR) and get off the channel if a conflict is found.
 * GbE: Gigabit Ethernet
 * GHz: Gigahertz, one billion hertz
 * Guard Interval / GI: The time/space between transmitted Wi-Fi symbols
 * ISP: Internet Service Provider
 * IoT: Internet of Things. A future where every device is connected via Wi-Fi
   to the Internet.
 * LAN: Local Area Network (eg: the wired network in your house, often Ethernet)
 * MAC: Media Access Control. Details (wikipedia.org).
 * MHz: Megahertz, one million hertz
 * MIMO: Multiple-input and multiple-output on the same frequency, where
   'multiple' refers to antennas. Also known as SU-MIMO (single user MIMO).
 * MU-MIMO: MIMO to "multiple users" at the same time.
 * NAT: Network Address Translation - a method for multiple client devices
   (using internal private internal IP Addresses) to share a single Internet
   connection (that only has a single public IP Address). wikipedia.org
 * 'N' Spatial Streams: Refers to T×R:N MIMO, where both 'T' and 'R' equals 'N'.
   For example, 4 spatial streams means 4×4:4.
 * OFDMA: Orthogonal Frequency-Division Multiple Access, a proven technology
   that comes from cellular 4G LTE.
 * ONT: Optical Network Terminal - takes a fiber connection from an ISP and
   provides an Ethernet connection for your router to plug into.
 * PHY: PHY is an abbreviation for physical. For example, 'PHY speed' refers to
   the physical speed at the raw network layer. For every Wi-Fi device, there is
   not just one PHY value, but both a Tx PHY and a Rx PHY.
 * PoE: Acronym for 'Power over Ethernet'. Uses an Ethernet cable to send both
   Ethernet and electrical power to a device. Details (wikipedia.org).
 * QAM: Quadrature amplitude modulation. Details (wikipedia.org). A method of
   converting and sending digital information (0's and 1's) over wires using
   analog signals.
 * Quad-Stream: Often refers to 4×4 MIMO in Wi-Fi 5 devices. But beware that in
   Wi-Fi 6, router companies are starting to 'add' the streams for each
   individual band, and publish that total number.
 * RBW: In spectrum analysis, means resolution bandwidth.
 * RF: Radio Frequency
 * RSSI: Received Signal Strength Indicator. More Information (wikipedia.org).
 * Rx: abbreviation for 'Receive'.
 * SFP: Small Form-factor Pluggable. A port often used in Fiber media
   converters. wikipedia.org
 * SINR: Signal-to-interference-plus-noise ratio.
 * SNR: Signal to Noise Ratio. The difference (in dB) between the signal level
   and the noise level.
 * TCP/IP: Transmission Control Protocol / Internet Protocol -- how computer
   systems connected to the Internet communicate with each other.
 * TDWR: Terminal Doppler Weather Radar. Important due to DFS channel
   restrictions.
 * Tri-band: Three access points in one. Often band one is 2.4 GHz, band two is
   5 GHz and band three is 5 GHz (or 6 GHz Wi-Fi 6E, or 60 GHz 802.11ad).
 * Tx: abbreviation for 'Transmit'.
 * WAN: Wide Area Network (the WAN port on your router is connected to a modem,
   which in turn is connected to the Internet).
 * WAP: Wireless Access Point. A device that provides wireless access to a wired
   network.
 * "Wave 2": An term in Wi-Fi used to define chipset and feature level. "Wave 1"
   is the first generation, supporting core basic features of a new Wi-Fi
   version. "Wave 2" is the next generation of chipsets that adds many optional
   advanced features.
 * Wi-Fi: A 'brand name' for wireless networking created by an alliance
   (wi-fi.org) of companies for IEEE 802.11 wireless technologies, to ensure
   that wireless products from different vendors all work with each other.
   Here's Why It's Called 'Wi-Fi' (huffpost.com).
 * WLAN: Wireless LAN


Appendix V: Version history for this paper
There are far too many minor updates being made to this paper all the time that
will not be documented. However, this appendix was recently created at the
request of readers! -- thanks, keep the feedback coming to document some of the
larger changes being made:
 * 2024/08/04: added POTS (Telephone) to Ethernet [§E9]
 * 2024/07/28: added Powerline networking [§E8] (pulled out of rest of paper)
 * 2024/07/27: added Outbuilding [§O] and Recommended Cable Modem [§E7]
 * 2024/07/25: added Windows tools [§E6] with WI.BAT tool
 * 2024/07/24: added MoCA - Multimedia over Coax [§E5]
 * 2024/07/07: added The hidden cost of 'always on' devices [§E4]
 * 2024/07/04: created 'Extras' section; added Bufferbloat [§E2]
 * 2024/05/05: added comment about four channel 1, 5, 9, 13 plan

 * 2024/01/20: navigation aids: (a) sticky headers and (b) click on header goes
   back to top
 * 2024/01/19: font size readability improvements for non-desktop
   phones/tablet/etc devices
 * 2024/01/15: added Supporting wiisfi.com [§X] and Ethernet [§E1]
 * 2024/01/10: added 'Wrapping it all up' [§22]
 * 2023/12/25: updated channel width vs range [§J] to add what a spectrum
   analyzer sees
 * 2023/12/14: added an 'Understanding the math' section to PHY tables [§F]
 * 2023/12/07: added Spectral Masks [§S]
 * 2023/10/31: combined first two chapters; added A quick overview of Wi-Fi [§2]
 * 2023/10/30: moved this paper to https://www.wiisfi.com
 * 2023/10/28: Changed Router/AP reference to Investigating Router
   Specifications [§B]
 * 2023/10/22: Added chapter numbers to in-document links
 * 2023/10/14: Added an example section to Access Points [§18]
 * 2023/07/05: Changed ASUS recommendation from Asus TUF-AX5400 to RT-AX88U Pro
 * 2023/05/20: Added 'metal object' issue to router placement [§P]
 * 2023/04/10: Added 'What 'stream' means' [§Q]
 * 2023/02/23: Added 'the easy way' to How to improve Wi-Fi speeds [§16]
 * 2023/02/20: Added Mesh Networks [§19] and Access Points [§18]
 * 2023/02/19: Changed TP-Link recommendation [§21] from Archer AX73 to Archer
   AX80
 * 2023/02/12: Added Router/AP placement [§P]
 * 2023/02/08: Emphasize need for 4×4 MIMO router in recommendation [§21]
   section
 * 2023/02/06: Added this version history appendix
 * 2023/01/29: Added Wi-Fi Range Extenders [§17] chapter
 * — tons of changes over the years that were not documented —
 * 2018/01/??: moved this paper to https://duckware.com/tech/wifi-in-the-us.html
 * 2017/10/??: started as a part of
   https://www.duckware.com/tech/ring-floodlight-cam-review.html

> Show more...


Appendix W: Call for changes in the Wi-Fi industry / rants
A call for changes in the Wi-Fi industry (and a rant or two)...

Clearly spell out ALL specifications: Router companies must clearly state the
full specifications (Wi-Fi standard, max PHY speed, QAM-level, MIMO level, MHz
channel width, etc) for EACH version of Wi-Fi and band supported by a router.
Don't force consumers to 'figure it out'. Instead, clearly spell it out on
specification web pages. For example:

> Wi-Fi 2 802.11b 2.4 GHz: 11 Mbps
> Wi-Fi 3 802.11g 2.4 GHz: 54 Mbps
> Wi-Fi 3 802.11a 5 GHz: 54 Mbps
> Wi-Fi 4 802.11n 2.4 GHz: 300 Mbps (64-QAM, 2×2, 40 MHz)
> Wi-Fi 4 802.11n 5 GHz: 300 Mbps (64-QAM, 2×2, 40 MHz)
> Wi-Fi 5 802.11ac 5 Ghz: 3467 Mbps (256-QAM, 4×4, 160 MHz)
> Wi-Fi 6 802.11ax 2.4GHz: 574 Mbps (1024-QAM, 2×2, 40 MHz)
> Wi-Fi 6 802.11ax 5 GHz: 4804 Mbps (1024-QAM, 4×4, 160 MHz)

Publish realistic throughput speeds: Router companies must publish the realistic
throughput speed that a typical Wi-Fi client device of that Wi-Fi generation can
expect to achieve. This would mean today, a throughput speed for a 2×2 MIMO
client.

Devices must display ACCURATE Tx and Rx PHY speeds: Clearly a Wi-Fi client knows
exactly what both Tx/Rx PHY speeds are, as it is both sending and decoding the
Wi-Fi signal. The industry must change from reporting a single overly optimistic
"Link Speed" in Wi-Fi clients to instead reporting both REAL transmit and
receive PHY speeds. Of note is that Ubiquiti routers and access points do report
both speeds for Wi-Fi clients. UPDATE: And now, so does Windows 11 (but there
are accuracy issues).

> There is NO justification for seeing great (inflated) PHY speeds when Wi-Fi is
> idle, and then seeing lowered PHY speeds in the middle of a speed test. This
> seems very intentional -- and wrong! Instead, display PHY numbers that
> accurately represent reality. Maybe the statistical 'mode' value the occurs
> the most of PHY for the last 100 or so non-management frame packets.
> 
> Better yet, display both a Tx and Rx histogram of the MCS levels used over
> both the last second and the last so many seconds. The Wi-Fi driver has this
> information. That would be a killer feature for assessing Wi-Fi signal quality
> -- because it makes the 'unseen' Wi-Fi signal visible. Especially helpful
> because it would make Wi-Fi interference visible as MCS levels used would
> slide lower, then slide back to normal.

Wi-Fi preferred band: When the SSID name is the same for multiple Wi-Fi bands,
it is up to the client device to choose which band to ultimately connect to.
Client devices sometimes do a horrible job of selecting the best band (the band
that results in the highest throughput). There needs to be a way on client
devices, when setting up a Wi-Fi connection, to specify which band is the
preferred band (like 5 GHz). Namely, if the band is present, the client device
must connect to the preferred band.

> Why? Because I would prefer to name the SSID for all bands the same, but can't
> because I often need to force an IoT device to use a particular Wi-Fi band.
> And currently the only way to accomplish that is by naming the SSID for each
> band uniquely.

Netgear's router web interface is in the dark ages: Netgear's web interfaces has
not been updated in a very long time, and now compared to its competitors, is in
the dark ages. The Netgear web interface is very hard to use: (a) the UX is
horrible the white on white color scheme used makes it impossible to see what is
static text and what is a field that can be changed, and (b) basic features that
the competition has, don't exist.

> For example, on TP-Link, I can turn a port forwarding rule on/off via a single
> click on a slider button -- because the rule has an on/off setting. But for
> Netgear, port forwarding rules are always on. Forcing you to delete the rule
> to turn it off, and re-enter the full rule to turn it back on. It is just not
> user friendly at all. Netgear, get your act together and modernize the router
> web interface.

Vizio TV's and missing Ethernet ports: Does anyone else notice that newer Vizio
TV's in configuration screens display an Ethernet MAC address, but then don't
actually provide an Ethernet port on the back of the TV? Interesting.

Appendix X: Supporting wiisfi.com


"Ready Player One" Nolan Sorrento saying "We can sell up
to 80% of an individual's visual field before inducing seizures"

NO ADVERTISING: I wrote this paper, and continue to update it on a regular
basis, to help educate people about how Wi-Fi works. My goal is not to make
money if it were, this paper would be chock-full of ads, like some other web
sites out there, but rather, my goal is education.

> I find this "Ready Player One" movie clip (youtube.com), 25 sec (seen right)
> to be a hilarious, gritty, and biting commentary on the very real annoying
> trend of 'too many ads' on web sites today.

★ How to support wiisfi.com: Please, no active donations needed. Instead,
support wiisfi.com passively -- by following any of the Amazon.com links to
products in this paper or the link below when you shop for anything on Amazon.
Why? Because if you follow a wiisfi.com link to Amazon and purchase anything
even a product not directly linked to, I, as an Amazon Associate, might earn a
little bit from your qualifying purchases.

> Shop Amazon - support wiisfi.com (TIP: drag link to your desktop to use later)
> 
> How it works: Amazon affiliates only get credit for qualifying purchases when
> you (1) click on an affiliate link, (2) add a product any product, really into
> your shopping cart, and (3) at some point right then, or even later, purchase
> the item in your shopping cart. There is NO credit if you click on an
> affiliate link, research, close the web browser, and later open a web browser,
> add products, and purchase. Closing the browser session ends any affiliate
> association.

You can confirm that the above Amazon link is 'working' when you see the
'wiisfi.com' tag in the opened web browser URL location (sample below). Then
search for a product, and add it to your cart.

> 

Thank you for your support.

Appendix Y: Contact the author



I have done my best to make this paper as easy-to-understand, no-nonsense,
informative and accurate as possible -- so that YOU can make your own educated
Wi-Fi upgrade decisions. But it has grown FAR larger than initially intended.
Did you find an error, a typo, or have a suggestion on how to improve this
paper? Did this paper help you? Do you disagree with any recommendation? Let me
know...

Use this web form (duckware.com) to contact the author of this paper, Jerry
Jongerius.

How you can support wiisfi.com [§X]



--------------------------------------------------------------------------------

Some images and text are copyrighted by their respective owners and are
reproduced
in this paper under Title 17 Section 107 (Fair Use) of the copyright code.
The rest of this document is Copyright © 2018-2024 Jerry Jongerius
As an Amazon Associate, I earn from qualifying purchases.