Wi-Fi 4/5/6/6E (802.11 n/ac/ax)|
(make educated wireless router/AP upgrade decisions)
(cut through all the marketing hype)
Version 7.0c (updated July 8, 2021)
Wifi speeds vs. broadband speeds:
Wifi speeds have not kept up with increasing Internet speeds. As a result, there has
been a very rapid switch in wifi from
Wi-Fi 4 (2.4 GHz 802.11n) to
Wi-Fi 5 (5 GHz 802.11ac), and now to
Wi-Fi 6 (5 GHz 802.11ax), in an attempt to keep up.
So what new router/AP should you consider buying today?
Router Manufacturers' Marketing Hype:
Don't be fooled by the marketing hype of router manufacturers' advertising
outrageously high aggregate (all bands added together) Gbps wireless speeds (like 7.2 Gbps).
What really matters is realistic speeds achieved by your wifi client devices,
that actually exist today.
The weakest link:
Wifi throughput to a 802.11ac wireless device will likely max out at around 600 Mbps (±60 Mbps)
for 2x2 MIMO, to 1000 Mbps (±200 Mbps) for 4x4 MIMO no matter what 4×4 router is used (when right next to the router).
And the far majority of ALL wireless devices today (smartphones, tablets, laptops, etc)
are still only 2x2 MIMO. So your client device is almost certainly
causing slow wifi speeds (and maybe not your existing AP/router).
The best router/AP VALUE today:
A high quality Wi-Fi 6 router/AP -- or 802.11ac Wi-Fi 5 "Wave 2" (2nd gen chipset)
4x4 router/AP -- supporting beamforming and ALL DFS channels is the way to go right now
(as of May 2021), due to the incredible VALUE.
Also, see the Recommendation and Router Appendix far below.
If you can find a "Wi-Fi 6 Certified" router that
meets your needs, go for it. But, it will be years before there is a sufficient
number of Wi-Fi 6 client devices to make a
Wi-Fi 6 router really worth it (benefits are only for Wi-Fi 6 clients now, of which there
are few), and by then, the next generation of "Wi-Fi 6E certified" routers will
be out -- so just be patient.
So, upgrade, or not?: The only question that really matters is:
What are client PHY speeds now and what will
client PHY speeds be after an AP/router update? Because, if
(the majority of) client PHY speeds will not increase after a router update, what is the
point in spending money on a new router that won't improve PHY speeds?
Wi-Fi 6E is just around the corner: Greatly complicating a decision is that
Wi-Fi 6E is just around the corner (early devices are out now)
that will require (yet again!) new hardware -- existing Wi-Fi 6 devices will not support Wi-Fi 6E.
The goal of this paper: This paper was written to help people understand
current wifi technology, so that YOU can make an educated 'router' upgrade decision
-- because there is WAY too much hype out there (especially about wifi speeds) --
and router manufacturers' are directly to blame.
The issue: Wifi spectrum is a limited, shared resource. Any number of access points
(both yours and neighbors) can all share the
SAME wifi spectrum. But because wifi use has exploded over the last few years
(tablets, laptops, smartphones, TVs, Blu-rays, security cams, thermostats, etc)
the wifi spectrum is way overcrowded. And when combined with ISP Internet speeds
now often times faster than Fast Ethernet (100Mbps), and sometimes even 1 Gbps,
wifi speeds have not kept up.
Industry solution: The industry quickly switched to the much newer
Wi-Fi 5 (5 GHz 802.11ac) spectrum, where
speeds are much faster, due to more available spectrum (over seven times as much)
and new wifi features, such as MIMO and wider (80 MHz) channels. But that requires
a new router, but which one?
BUT, our devices are still speed limited. Why? As will be shown in the
next sections, mainly due to limited 2×2 MIMO support in almost all of
today's wireless (and battery powered) client devices.
But let's also be very realistic. If you have 400 Mbps (or less) Internet speeds,
2×2 MIMO Wi-Fi 5 to your router is almost always sufficient -- and IS fast
enough with a high quality "wave 2" 4×4 MIMO 802.11ac router (an example
of one is seen upper right, and are a great VALUE right now).
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 laptop computer.
Same for most wireless clients. Why? Because that is the speed expected from
these (2×2 MIMO) devices! This section explains in great detail exactly why that is.
You may safely skip to the next section for a shortcut if this section is too detailed/technical for you.
||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 wifi 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.
The following sections detail how the grossly 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 section 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 wifi 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
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 when you are very close (just feet) to the router. But as you move further 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 section.
Understanding Modulation/Coding: Imagine that once a second, 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 wifi.
If you are close to the AP/router, a large number of bits can be conveyed 'at once'. 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 wifi 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 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)
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 wifi channel.
But there may (or may not) be other wifi users (either local, or even neighbors on the
same spectrum) which will decrease your speed by some unknown amount.
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.
Results: 2×2 MIMO devices get a realistic 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 two critical factors that greatly impact and determine
maximum real-world speed for a single client are:
(1) lowest common MIMO level support 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.
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 section).
For a fast shortcut, stay in this section.
With a new modern "wave 2" 802.11ac router, it is 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 raw bitrate
between the device and your AP/router) and take 70% of that PHY speed to estimate maximum
application speed (the next section explains why the overhead is so large).
Then lookup the PHY speed number in the PHY speed tables 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
In a real-world test, on a 2×2 866.6 Mbps 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. So brand new client hardware seems to perform much better than
years old client hardware.
PHY speed is an indicator of: (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, look up the speed in the full
PHY speed tables.
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).
Windows 10: 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 10/8/7: 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).
Windows PHY Speed
Example: Lookup 702 Mbps speed (right)
in the PHY tables far below
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,
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 wifi "Link Speed".
[more info on finding Link Speed on a Mac]
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 wifi driver,
this observation could be very wifi device (and vendor) specific.
Running "netsh wlan show all" (from a DOS CMD prompt) displays a ton of
wireless information, including both Tx and Rx speeds for the wifi interface, but
on my test systems, both speeds are always the same Mbps value (which is not
accurate). It might work better for you?
iOS (iPhone/iPad/iPod): Not known (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
All modern iOS devices are 2×2 MIMO.
UPDATE: Here is a tip I received -- if you happen to use Apple's AirPort WiFi base station:
"Install Apple's 'airport utility' and then open it. Click on the wifi 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."
Android: Go into "Settings / Connections / Wi-Fi", click on the connected wifi network,
and find the 'Network Speed' (example right). Or, it may also be called 'Link Speed'.
Android PHY Network Speed
Example: Lookup 585 Mbps (right) in the PHY tables far below,
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.
Kindle: Under Settings, click on "Wireless", then click on the connected wifi network,
and look for "Link speed".
It appears that Android reports the "Tx PHY" as the 'link speed'. This is
preliminary and needs more research.
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.
Netgear Router 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 Matthew S. for pointing that out).
Comcast Gateway 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.
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. 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.
|Wi-Fi 6 Devices|
(and MIMO level)
|Apple iPhone 11, 12||2×2|
|Dell Laptops (high end)||2×2|
|Wi-Fi 5 Devices|
(and MIMO level)
|Apple iPhone X,8,7,6s||2×2|
|Apple iPhone 6||1×1|
|Apple iPad + Air2||2×2|
|Apple iPad Air||1×1|
|Dell Laptops (high end)||2×2|
|Dell Laptops (low end)||1×1|
|Fire TV (gen 2 and later)||2×2|
|Google Pixel 4,3,2,1||2×2|
|MacBook Pro (some?)||3×3|
Supporting 4×4 MIMO takes a lot more power, and for battery powered devices,
runtime is FAR more important.
You can expect a maximum PHY speed of 866 Mbps, and around
600 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 Dell only offers a maximum 2×2 laptop as of February 2019.
You will see the spec sheets for many modern phones that MIMO is 4×4, but look
closely and notice that this is for cellular, not Wi-Fi.
A final warning: This discussion about 'the' PHY speed of your device is
slightly over simplified, as for every wifi 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 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.
Are you highly technical? Then use the Router deep dive 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.
The bottom line: Under ideal conditions, you can and should expect Mbps throughput
around 70% (±10%) of the client PHY Mbps speed. But in many situations (for tons of various reasons),
overhead can be much higher, resulting in throughput as low as 50% of PHY speed.
What I am beginning to realize is that the PHY speed reported by wifi devices is a best
case value (and may not accurately reflect the actual PHY speed used).
The MCS Spy tool 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).
Wifi overhead is 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 wifi 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, traffic lights, traffic, school zones, weather conditions, etc.
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 wifi. 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.
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 from wikipedia
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
(using a 802.11b 1 Mbps beacon rate)
Half Duplex: There is no separate download spectrum and upload spectrum in wifi
(whereas Ethernet is full duplex - can send and receive at the same time).
Instead, there is only a common spectrum (channel) that ALL wifi devices (router and clients) operating
on that channel must use in order to transmit (and receive).
A striking analogy: You are on a road going 120 mph, but for 40 feet of every mile (or 0.75% of
5280 feet) you are required to slow down to 1 mph. What is your average mph for the mile? -- lower
than you might 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).
So when you are running that download throughput speed test, your device is mostly receiving,
but it is also transmitting (acknowledging data sent)! Using the MCS Spy tool
(Router deep dive appendix below), a PC
downloaded at 120 Mbps, but was uploading at 1.8 Mbps at the same time. 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 wifi is
half duplex, there may be around 1% to 3% (relative) 'overhead' simply due to how TCP/IP
CSMA/CA: The Wi-Fi spectrum is a shared resource. So how does a device know that it is OK to
transmit? Wifi uses something called
(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 wifi 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, and hope for no collisions).
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 wifi devices.
Acknowledgements: Every Wi-Fi packet sent must be 'acknowledged' (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).
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 it mostly works so well). But overhead
can increase dramatically if there are too many devices all wanting to transmit at the
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.
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.
Hidden Node Issue: There is something called the
Hidden Node Problem
that can (potentially) cause a large number of collisions in wifi -- 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.
Wi-Fi hidden node problem
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 wifi and transmit at the
exact same time. If both hidden nodes are sporadically using wifi, 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.
Beamforming overhead: The sounding frames for beamforming adds a tiny
bit of overhead. This 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, 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 (around 3 Mbps), and the second user
will get to use the channel the other 50% of the time (around 433 Mbps).
A final caveat: PHY speed is a very complicated thing. Tx PHY and Rx PHY can not only
be asymmetric (more details below),
but also be highly variable. The 'link speed' your device reports to you
is a highly over-simplified 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.
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, you are sharing the SAME channel.
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 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 not being used, but
MAC efficiency is much better than it initially appears.
The bottom line: The AC#### naming convention (AC1900, AC2600, AC5300,
AC7200) and AX#### naming convention (AX6000, AX11000) 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 wifi 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?
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?
iPhone XS Max
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:
MAC Overhead: Take the 5GHz 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.
- Band One: 750 is the maximum PHY speed for MIMO 3×3 in the 802.11an 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.
- Band Two: 1625 is the maximum PHY speed for MIMO 3×3 in the 802.11ac 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).
- Band Three: same as band two.
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
(but see tri-band router section 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 a brand new 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.
MIMO: What is (partly) driving the dramatic increase in wireless (wifi, cellular, etc) capacity
in the last few years is MIMO
(acronym for Multiple Input, Multiple Output),
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 wifi is stuck at 2×2 MIMO for most wireless wifi devices.
4×4 MIMO illustration
Analogy: Think of MIMO as adding 'decks' to a multi-lane highway.
More lanes (capacity) are added without using more land (spectrum).
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 (by both the router and client).
MIMO adds more capacity without using more spectrum!
Example: On a single 80 MHz 802.11ac channel operating at 433 Mbps:
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. And as a 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.
all on the same 80 MHz channel.
- 1×1 MIMO yields 433 Mbps
- 2×2 MIMO yields 866 Mbps (most wireless clients are 2×2)
- 3×3 MIMO yields 1300 Mbps
- 4×4 MIMO yields 1733 Mbps (most new Wi-Fi 5 routers are 4×4)
- 8×8 MIMO yields 3466 Mbps
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: Multiple antennas can also 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.
Beamforming: A 802.11ac "wave 2" technology that 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.
Comcast XB6 gateway - 8×8 MIMO
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 value (a stronger
signal, which translate to better connect speeds for some users)!
Almost all battery powered wireless devices are stuck at 2×2 MIMO for wifi,
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.
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 (July 2020) are (at best) 2×2 MIMO. It is very rare to see a (battery powered) client
device that supports 3×3 (or higher) MIMO.
AX 'Stream' Deception: Router vendors' are now being incredibly deceptive
when it comes to advertising in their new "AX" class of routers. Netgear is using "spatial streams"
to describe their new AX routers
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"
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"
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 hardware.
Reference: A brief look at past legacy wifi generations (and while not
official names, Wi-Fi 1, Wi-Fi 2, and Wi-Fi 3):
802.11 (Wi-Fi 1): PHY data rates of 1 or 2 Mbps using direct sequence spread spectrum (DSSS)
with three non-overlapping 22 MHz channels in 2.4 GHz (1, 6, 11).
|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 ||1999 ||54 Mbps ||Wi-Fi 3 ||5 GHz|
|Third ||802.11g ||2003 ||54 Mbps ||Wi-Fi 3 ||2.4 GHz|
802.11b (Wi-Fi 2): PHY data rates
of 1, 2, 5.5, or 11 Mbps
(how rates are calculated)
using direct sequence spread spectrum (DSSS)
with three non-overlapping 22 MHz channels in 2.4 GHz (1, 6, 11).
802.11a (Wi-Fi 3): PHY data rates 6 Mbps to 54 Mbps (see table right) 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.
But 802.11a really never 'took off' 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).
20 MHz channel
800ns guard interval
More PHY tables
The router industry learned a hard lesson -- that any new router/AP must also be
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.
802.11g (Wi-Fi 3):
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 right) using orthogonal frequency-division
multiplexing (OFDM) with three non-overlapping 20 MHz
channels in 2.4 GHz (1, 6, 11) -- see the next section for 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 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).
The big advance in Wi-Fi 3 was the introduction of OFDM, instantly improving
throughput nearly five times over the prior Wi-Fi 2 (from 11 Mbps to 54 Mbps;
only for AP/clients that support OFDM).
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 wifi 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 section for more details.
The big advance in Wi-Fi 4 was the introduction of MIMO
(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 2.4 GHz is a legacy wireless band that has been replaced by
the much faster 802.11ac in 5 GHz. This section is provided for reference only. You should be using
802.11ac 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 802.11ac (many IoT devices do not).
|Fourth||802.11n ||2009||72 to 217 Mbps ||Wi-Fi 4|
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.
|2.4 GHz wifi channels|
|Channel||MHz center||20 MHz channel|
|microwave ovens: 2450 MHz ±50 MHz|
in the U.S.
Spectrum: There is ONLY 70 MHz of spectrum (2402-2472 MHz) available for wifi to use in the U.S.
in the 2.4 GHz band, supporting only three non-overlapping 20MHz channels.
There are eleven OVERLAPPING 2.4 GHz wifi channels:
In the US, wifi routers allow you to set the 2.4 GHz wifi
channel anywhere from 1 to 11
So there are 11 wifi channels, right?
NO! These eleven channels are only 5MHz apart --
and it actually takes a contiguous 20MHz (and a little buffer MHz between channels) to make
one 20MHz wifi channel that can actually be used. Because of this, in the US, these restrictions
result in only three usable non-overlapping 20MHz wifi channels available for use
(1, 6, or 11; seen right).
The THREE non-overlapping channels:
You CAN set to the wifi 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, the 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
Shared spectrum: All wifi devices on the same spectrum must SHARE
that spectrum. Ideally, all wifi devices decide to operate on either channel 1, 6, or 11 -- the
only 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 wifi channel has PHY bitrate of around 72Mbps, but
due to wifi protocol overhead, you may only get to use around 60% to 80% of that.
In a very 'clean' wifi environment, I have seen throughput around 54.2 Mbps for
a PHY speed of 72.2 Mbps, which comes out to 75% efficiency -- pretty good.
Understanding channel widths: The standard wifi channel width is 20 MHz. So a 40 MHz channel
is TWO 20 MHz channels put together (2× capacity).
256-QAM and 1024-QAM are non-standard
|802.11n PHY Speeds (Mbps)|
20 MHz channel, 400ns guard interval
|0||BPSK ||1/2|| 7.2|| 14.4|| 21.6|| 28.8|
|1||QPSK ||1/2||14.4|| 28.8|| 43.3|| 57.7|
|2 ||3/4||21.6|| 43.3|| 65.0|| 86.6|
|3||16‑QAM ||1/2||28.8|| 57.7|| 86.6||115.5|
|4 ||3/4||43.3|| 86.6||130.0||173.3|
256-QAM and 1024-QAM are non-standard
|802.11n PHY Speeds (Mbps)|
40 MHz channel, 400ns guard interval
|0||BPSK ||1/2||15|| 30|| 45|| 60|
|1||QPSK ||1/2||30|| 60|| 90||120|
|2 ||3/4||45|| 90||135||180|
More PHY speed tables
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 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 wifi 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 wifi signals).
However, for wifi 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'
wifi -- which unless you live in outer Siberia, you WILL 'see' neighbors' wifi signals and
the router will be required to automatically disable channel bonding.
I am curious if this issue had anything to do with why Netgear stopped getting their
routers 'Wi-Fi Certified'?
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.
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 inexpensive laptop
that is only two years old, but only supports 20 MHz channels).
Also, in the real world, things are MUCH more complicated, because many routers
don't always follow 'good neighbor' standards.
Article: "Bye Bye 40 MHz Mode in 2.4 GHz"
Of note is that 40 MHz channels in the 5 GHz band for 802.11n does work (very well).
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 section.
Interference: The entire 2.4 GHz space is plagued by interference (a victim of the
success of the 2.4 GHz band), 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.
which is the entire wifi space, and very likely impacting two of the wifi channels,
and in some cases, even all three wifi channels
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 wifi channels.
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).
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 wifi clients to disconnect from the AP, while in another house, using the microwave
oven only causes a slight slowdown in bandwidth to wifi 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 cam.
Is your house near a busy road? If so, you are likely getting 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
The BOTTOM LINE: The 2.4 GHz band is just WAY too crowed. 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.
In a resort community, with homes very close to each other, a Wi-Fi analyzer app shows well
over 15 2.4 GHz networks within range. At night, Wi-Fi performance (actual throughput) on
the 2.4 GHz band was horrible 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' wifi devices -- don't
assume a wifi 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)!
Understanding where the speed increases in 802.11n (over 802.11g) came from:
54 Mbps in 802.11g becomes 58.5 Mbps in 802.11n by using 52 subcarriers out of 64 (instead of just 48),
which then becomes 65 Mbps by reducing the guard interval (GI) from 800ns to 400ns,
which then becomes 72.2 Mbps via a new QAM modulation,
which then becomes 144 Mbps and 217 Mbps via MIMO.
So MIMO is the key factor for dramatically increased speeds in 802.11n over 802.11g.
The big advance in Wi-Fi 5 was the introduction of 80 MHz channels
(by moving into 5 GHz), 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 (next section). However, very few
devices support Wi-Fi 6 today, and it will be years before most new devices support
The fifth generation of wifi is 802.11ac (2013) on 5 GHz.
a maximum PHY speed of 3.4 Gbps on an 80 MHz channel using 8×8 MIMO
(and fully backward compatible with prior wifi generations).
However, a much more realistic maximum PHY speed is 1.7 Gbps on an 80 MHz channel
using 4×4 MIMO.
In Nov 2020, it should be 'officially' replaced by 802.11ax (Wi-Fi 6).
|Fifth||802.11ac||2013||433 to 1733 Mbps||Wi-Fi 5|
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 wifi 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.
More info from Wikipedia
|5 GHz wifi channels (U.S.)|
|Channel #||20 MHz|
|GAP (160 MHz)|
|GAP (5 MHz)|
Spectrum: There is 500 MHz of spectrum (5170-5330, 5490-5730, 5735-5835 MHz) available for wifi to
use in the U.S., supporting six 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 500 MHz).
The 5 GHz wifi band has six 80 MHz channels (see table right; 42, 58, 106, 122, 138, 155)
BUT ONLY if you have an AP that supports ALL the new DFS channels.
One 80 MHz channel in 5 GHz has more spectrum
than ALL 2.4 GHz channels combined!
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 section.
802.11h defines (1) dynamic frequency selection (DFS) and (2) transmit power control (TPC).
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. This then becomes 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 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.
Cisco blog post on the TDWR issue.
Channels affected are in dark red (right).
NEW upcoming channels: In mid December 2019,
the FCC voted
to move forward on allocating an additional 45 MHz to the end of U-NII-3 in 5.9 GHz
to wifi (seen in yellow in table right), that, if passed, would eventually result in
three NEW 20 MHz channels (169, 173, 177).
See also FCC 19-129.
This then creates two additional 40 MHz channels
(167, 175), one new 80 MHz channel (171), and most importantly, one new
non-DFS 160 MHz channel (163). This is important because the
other two existing 160 MHz channels (50, 114) are impacted by DFS restrictions.
Making use of the new channels will almost certainly require new hardware by router
manufacturers (even though new firmware and new FCC permissive filings might work, but there
is no 'profit' in that).
Center Frequency TIP: The 'center frequency' (in GHz) 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 GHz. Or, reverse to compute channel number from center
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.
Range: It is true that the range/distance of 5 GHz is reduced 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.
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.
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 wifi anymore. Of course, all of this depends upon how 'close' your neighbors are.
The Mbps seen at the application level will be around 60% to 80% of the Mbps at the
wifi (PHY) level. This is just due to wifi protocol overhead
(see section on PHY client speed far above).
New Channel Plan: Here is the 5 GHz
802.11 Channel Plan
(see also below) from the FCC.
Of note is that on April 1, 2014 the FCC changed the rules
for usage in the 5 GHz band, to increase availability of spectrum for wifi use.
[summary of the new rules].
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.
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 wifi channels (as frequently as
2.4 GHz channels), which translates directly to less interference (the channel is all yours) and
higher wifi speeds.
Many residential routers have a transmit power around 995 mW. Many (battery powered) wifi
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.
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.
1024-QAM is non-standard
|802.11ac PHY Speeds (Mbps)|
80 MHz channel, 400ns guard interval
|0||BPSK ||1/2|| 32|| 65|| 97|| 130|| 260|
|1||QPSK ||1/2|| 65|| 130|| 195|| 260|| 520|
|2 ||3/4|| 97|| 195|| 292|| 390|| 780|
|3||16‑QAM ||1/2||130|| 260|| 390|| 520||1040|
|4 ||3/4||195|| 390|| 585|| 780||1560|
|5||64‑QAM ||2/3||260|| 520|| 780||1040||2080|
|6 ||3/4||292|| 585|| 877||1170||2340|
|7 ||5/6||325|| 650|| 975||1300||2600|
|↕ typical real-world Modulation/Coding ↕|
|8||256‑QAM ||3/4||390|| 780||1170||1560||3120|
|9 ||5/6||433|| 866||1300||1733||3466|
1024-QAM is non-standard
|802.11ac PHY Speeds (Mbps)|
80 MHz channel, 800ns guard interval
|0||BPSK ||1/2|| 29|| 58|| 87|| 117|| 234|
|1||QPSK ||1/2|| 58|| 117|| 175|| 234|| 468|
|2 ||3/4|| 87|| 175|| 263|| 351|| 702|
|3||16‑QAM ||1/2||117|| 234|| 351|| 468|| 936|
|4 ||3/4||175|| 351|| 526|| 702||1404|
|5||64‑QAM ||2/3||234|| 468|| 702|| 936||1872|
|6 ||3/4||263|| 526|| 789||1053||2106|
|7 ||5/6||292|| 585|| 877||1170||2340|
|↕ typical real-world Modulation/Coding ↕|
|8||256‑QAM ||3/4||351|| 702||1053||1404||2808|
|9 ||5/6||390|| 780||1170||1560||3120|
|- ||5/6||487|| 975||1462||1950||3900|
More PHY speed tables
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).
This modulation is a non-standard extension to 802.11ac, 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 marketing hype.
See Broadcom NitroQAM.
802.11ac Wave 2: The
next generation (wave 2)
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
Cisco Wave 2 FAQ.
Buyer beware: Not all 'wave 2' products will support the restrictive 5 GHz DFS channels!
WiFi certification for 'wave 2' only
devices to support this -- so NOT required.
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 wifi,
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.
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.
(1) sometimes disables client MIMO (where a 2×2 client switches to 1×1; Broadcom chipset)
(2) requires spatial diversity (physical distance) between clients
(3) has significant sounding overhead
(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".
Aruba Networks says "Experience from 802.11ac MU-MIMO in real-world deployments revealed some limitations".
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':
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
wifi. 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)
Understanding where the speed increases in 802.11ac (over 802.11n) came from:
144 Mbps in 802.11n becomes 650 Mbps in 802.11ac by using an 80 MHz channel width (instead of 20 MHz channel width),
which then becomes 866.6 Mbps via a new 256-QAM modulation.
So quadrupling channel width is the key factor for increased speeds in 802.11ac over 802.11n.
The big advance in Wi-Fi 6 was efficiently transmitting to a large number of
users at the same time (but only for new Wi-Fi 6 clients, not prior wifi generation clients).
"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."
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!
STOP - Do NOT buy a Wi-Fi 6 router just yet:
Wi-Fi 6E devices are just starting to come out
(current Wi-Fi 6 hardware will not support)!
Then, consider that the 'fine print' for Wi-Fi 6 routers available now state that
they "MAY NOT" (or even "DO NOT") support all ratified 802.11ax features
(eg: Netgear RAX80 and RAX120). Bottom line: You will NOT benefit
from a Wi-Fi 6 router/AP until most of your wireless devices are Wi-Fi 6 anyway.
Look not for just Wi-Fi Certified, but for the 'Wi-Fi 6 Certified' symbol (seen right).
Wi-Fi Certified Product Finder.
List of Wi-Fi 6 Certified Routers
The sixth generation of wifi is 802.11ax (2019).
a maximum PHY speed of 4.8 Gbps on an
80 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).
|Sixth||802.11ax||2019||600 to 2401 Mbps||Wi-Fi 6|
2401 Mbps speed: The 2401 Mbps maximum PHY speed is for an 80 MHz channel to an 4×4 client.
However, a much more realistic maximum PHY speed is 1200 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.
|802.11ax PHY Speeds (Mbps)|
80 MHz channel, 800ns guard interval
|0||BPSK ||1/2|| 36|| 72|| 108|| 144|| 288|
|1||QPSK ||1/2|| 72|| 144|| 216|| 288|| 576|
|2 ||3/4||108|| 216|| 324|| 432|| 864|
|3||16‑QAM ||1/2||144|| 288|| 432|| 576||1152|
|4 ||3/4||216|| 432|| 648|| 864||1729|
|5||64‑QAM ||2/3||288|| 576|| 864||1152||2305|
|6 ||3/4||324|| 648|| 972||1297||2594|
|7 ||5/6||360|| 720||1080||1441||2882|
|↕ typical real-world Modulation/Coding ↕|
|8||256‑QAM ||3/4||432|| 864||1297||1729||3458|
|9 ||5/6||480|| 960||1441||1921||3843|
|802.11ax PHY Speeds (Mbps)|
80 MHz channel, 1600ns guard interval
|0||BPSK ||1/2|| 34|| 68|| 102|| 136|| 272|
|1||QPSK ||1/2|| 68|| 136|| 204|| 272|| 544|
|2 ||3/4||102|| 204|| 306|| 408|| 816|
|3||16‑QAM ||1/2||136|| 272|| 408|| 544||1088|
|4 ||3/4||204|| 408|| 612|| 816||1633|
|5||64‑QAM ||2/3||272|| 544|| 816||1088||2177|
|6 ||3/4||306|| 612|| 918||1225||2450|
|7 ||5/6||340|| 680||1020||1361||2722|
|↕ typical real-world Modulation/Coding ↕|
|8||256‑QAM ||3/4||408|| 816||1225||1633||3266|
|9 ||5/6||453|| 907||1361||1814||3629|
More PHY speed tables
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' technology into wifi.
This was accomplished by changing to the OFDMA modulation scheme and changing
the wifi 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!
Also, see this interesting
MU-OFDMA vs MU-MIMO
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 wifi 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 (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 wifi 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.
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 get 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 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 hit using wider channel widths.
160 MHz channels: 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.
The other problem with 160 MHz channels is that there are currently only two channels, and they
both intersect with DFS channels (making them both potentially unusable).
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 hope is that the FCC will open up the 6 GHz band for wifi sometime soon.
If the FCC does, 802.11ax will quickly use that band -- and require new hardware/routers!
See Wi-Fi 6E in the next section.
WPA3: For a device to be Wi-Fi 6 'certified', it was
that WPA3 is a mandatory feature.
Outdoor Wi-Fi: 802.11ax changed symbol timings (from 3.2µs to 12.8µs; and
increased GI times), which allows for wifi to operate much better in outdoor environments,
where signal reflections take more time and can cause problems. The increased timings
account for these reflections.
HERE COMES THE HYPE: Manufacturers are touting
incredibly speed claims regarding 802.11ax (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 section).
And with 802.11ax (and everything else the same), the 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 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
(wifi 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 wifi than smartphones), and (3) 1000 mph
was obtained by adding the speeds of multiple cars together (aggregating
multiple wifi bands).
Should I upgrade to Wi-Fi 6? In 2019, absolutely not. In 2020, probably not.
Never buy routers based upon 'draft' specifications. Instead, wait until the 'draft' becomes
'finalized' and then only look into 'final' hardware. And even if 'final' routers are out,
for the 'typical' home, probably not, and not for years. For a business, 'maybe'. If you
have a small to normal number of wifi users connected, Wi-Fi 5 will work just fine. But
if you have a large number of Wi-Fi 6 users, then you may very well see an improvement
by using Wi-Fi 6.
Is there really something that you can't do with 455 Mbps throughput in
Wi-Fi 5 that you can all of a sudden do with a little (10%) more throughput in Wi-Fi 6?
A final word on Wi-Fi 6: Is it possible to get a 38% speed improvement over Wi-Fi 5
to a single wifi client? Yes, but you have to be a Wi-Fi 6 client very close to the Wi-Fi 6
router so that the highest 1024-QAM can be used. And 'at distance', other Wi-Fi 6 clients
will see a speed improvement lower than that (closer to 11%). For Wi-Fi 5 clients, no speed
improvement will be seen. For some people, maybe this small percentage increase matters. But
if ultimate speed matters that much to you, just plug into Gigabit ethernet!
There is very little point in upgrading a Wi-Fi 5 router to a Wi-Fi 6 router until many/most
of the clients connecting to the router fully support Wi-Fi 6. Until that happens, upgrading
a router to Wi-Fi 6 will have very little impact. Vendors are throwing around huge Mbps
numbers that are meaningless (because it is client device capabilities that mostly limits
"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"
"For [most enterprise customers], we recommend installing
802.11ac wave 2 access points today, because of the sheer value of 802.11ac wave 2"
Consumer Reports concludes
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.
I have seen some reviewers show graphs showing a huge increase in Wi-Fi 6 speeds as compared
to Wi-Fi 5, but that result was obtained by using 160 MHz channels in Wi-Fi 6 vs 80 MHz channels in
Wi-Fi 5. When reviews show numbers too good to be true, scrutinize the details.
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.
I actually think Wi-Fi 6 is going to (eventually) be great. But the industry selling Wi-Fi 6 routers
that are actually 'draft' routers that don't fully implement the Wi-Fi 6 specification,
and are not Wi-Fi 6 certified, is a problem. The router industry has not
self-regulated, and you, the consumer, are paying the price.
Fully "Wi-Fi 6 Certified" routers ARE just starting to come out. Be patient and
don't buy a 'draft' router.
Understanding where the speed increases in 802.11ax (over 802.11ac) came from:
866.6 Mbps in 802.11ac becomes 960.8 Mbps via the switch to OFDMA,
which then becomes 1201 Mbps via a new 1024-QAM modulation. So at distance, 802.11ax is only
11% faster than 802.11ac.
The big advance in Wi-Fi 6E was dramatically increasing available
spectrum/channels (by 7 times for entry level AP/routers), which
should make 160 MHz channels actually usable and commonplace, instantly
doubling throughput over the prior Wi-Fi 5/6 with 80 MHz channels.
Wi-Fi 6E is brand new and devices are just starting to come out!
Wi-Fi 6E: Wi-Fi 6E (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.
There is only 500 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.
The big deal: TONS of new spectrum! The additional spectrum allows for 14
additional 80 MHz channels
(or seven additional 160-MHz channels) in wifi, 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).
And again, once Wi-Fi 6E routers come out, it will take a long time before most
clients are Wi-Fi 6E capable, but you can bet that most higher-end client devices
will immediately and fully switch over to and support Wi-Fi 6E.
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.
The gotcha: New hardware (routers/clients) will be required. Current Wi-Fi 6 devices
don't support Wi-Fi 6E!
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.
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).
Namely, expect Wi-Fi 6E range (in low-power mode) to be around 42% of Wi-Fi 5 DFS
Normal-power mode: In normal power mode, access points are only permitted to use
850 MHz of spectrum (see table 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. Namely, expect Wi-Fi 6E range (in
normal-power mode) to be around 83% of Wi-Fi 5 range.
Automated Frequency Coordination (AFC):
The FCC docs
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 (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 concern: Available channels: 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.
So, it remains to be seen how many channels can actually be used in real-life with AFC
for normal-power devices.
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.
Only time will tell -- until actual Wi-Fi 6E devices become available for testing.
The FCC did not just have 1200 MHz of spectrum laying around unused. Instead, this spectrum is heavily used
by 'incumbent services':
(1) Fixed Microwave Services (FS): You have probably seen these towers around
(google microwave tower),
with a 'dish' pointed in a fixed (horizontal) direction.
(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).
Video showing FS usage in the US (YouTube)
Best use case: The first wave of Wi-Fi 6E devices will likely operate 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 reliable 2 Gbps
PHY connection with 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
Interesting observations about Wi-Fi 6E from this FCC doc:
Wi-Fi 6E was
as an idea/desire on January 3, 2020. Days later, Broadcom
supporting Wi-Fi 6E in 6 GHz. Then on April 24, 2020, the
FCC moved forward
in supporting this
And all other major chipset vendors have also announced support for Wi-Fi 6E.
- 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).
Since all major chip vendors have announced support for Wi-Fi 6E,
we can expect all major router vendors to deliver new hardware supporting Wi-Fi 6E soon.
indicates 6E hardware hitting stores as soon as late 2020 -- and if this actually happens, will the recent Wi-Fi 6
effectively be dead and be immediately replaced by Wi-Fi 6E devices (which should be backward
compatible with Wi-Fi 6)?
Understanding the speed increase: Wi-Fi 6E makes 160 MHz channels
commonplace (vs 80 MHz channels), which will double speeds over Wi-Fi 6.
This section applies to both Wi-Fi 5 (802.11ac) and
Wi-Fi 6 (802.11ax) operating in 5 GHz, but not
Wi-Fi 6E (802.11ax) operating in 6 GHz.
In a nutshell: 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,
greatly increasing the likelihood of sharing that channel with others
(a 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.
There are SIX 80-MHz wifi channels in 5 GHz.
Two channels can always be used (green highlight, right). 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 routers do not support CH 138
|The SIX 5 GHz 80-MHz channels|
DFS = Dynamic Frequency Selection
Why support is important:
Support for all channels becomes critically important to avoid interference (sharing
bandwidth) with a neighbor's wifi. Ideally, every AP/router (yours and neighbors)
should be on a unique/different wifi 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.
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.
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).
See the Router Reference Appendix far below for examples. Some
vendors have NO routers that support DFS channels.
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.
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 220.127.116.11. 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
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).
How to research DFS support for any router/AP (check the FCC filings):
FCC Operating Frequencies show DFS 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 (wifi client). A device is not
allowed to use any DFS channels unless the proper paperwork is filed with the FCC.
- Google 'wikidevi' and the router company name and the router model number (eg: "wikidevi Netgear RAX80").
You should immediately find the
wikidevi web page for that router.
- On the wikidevi web page, find the "FCC ID:" for that router (eg: "PY318200414").
- Google the FCC ID found and click on the first top hit, which should be
- In the resulting web page, look for the "Operating Frequencies" section (seen right).
- 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).
- 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.
Netgear was plain lazy:
Netgear got the R6700v3
certified as a DFS Master
but failed to get the router certified as a DFS Slave.
This matters if you use the R6700v3 as a 'wireless bridge' (to connect 'ethernet only' devices to your main wifi 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
Warning: Just because a router allows 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.
Warning: Not all wifi 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). Wifi clients
have it easy -- just follow the lead of the router. And yet, it is possible
that a wifi 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.
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.
A wifi 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
notice, because the problem device will just connect to the router's slower 2.4 GHz
band (not the fast 5 GHz DFS band).
Someone told me that Roku devices do not support DFS channels. A google search appears
to confirm this, but this needs more research. If true, that is crazy laziness on their part.
SSID: SSID is simply the wifi network NAME. When you connect to a wifi 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 wifi access
point, you want it to use the SAME network name (and password + security), as this allows for
wifi roaming. Your wireless devices simply connect to the strongest wifi signal
with a matching SSID name.
Netgear R7800 wireless setup
You can use different SSID names, but then you don't get wifi 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 names, or different names (often with a "-5G"
appended to the 5 GHz band SSID name)? If named the same, client devices choose which band to connect
to. If named differently, the end user must choose 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 signal
was weaker, but throughput from 5 GHz was MUCH better.
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
can still see your network and find out the network name.
The bottom line: Just use whatever naming method works best for you and your devices.
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).
Channel: ALL wifi access points (in your house and visible neighbors networks)
covering the same frequency must share the wifi 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).
There is nothing special about channel selection. 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.
80 MHz 5 GHz channel 42 is made by
combining four 20 MHz channels
(only one considered 'primary')
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 wifi has overlapping channels and that the only real (non-overlapping) channels
available are 1, 6, 11.
Mesh / Extenders: These devices are a great convenience and do work -- but ONLY use them as a
last resort. Why? Because, by definition, they consume wifi bandwidth/spectrum to accomplish
their job. Instead, I always take the time to find a way to run wired Ethernet and
setup a new/second AP. Your network itself should never consume wireless bandwidth/spectrum
(mesh systems use a wireless backhaul). Instead, reserve wireless for your client devices.
Note that some extenders/mesh devices can be configured to use Ethernet backhaul to the main
router, which is fine.
A full discussion on channel planning is beyond the scope of this paper, but in short, always try
to leave unused spectrum between active 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 (if that is possible).
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.
DNS Servers: I configure all of my routers to use
Google's Public DNS servers
at 18.104.22.168 and 22.214.171.124.
In your router, the setting for DNS servers is usually found in the 'Internet Setup' section.
The default 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, and (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.
Another fast DNS service is provided by CloudFlare,
with DNS servers 126.96.36.199 and 188.8.131.52.
When you make this change, all devices locally on your network will
automatically use the new DNS servers (except for those devices that
manually override the 'get automatically from the router' behavior).
Turn UPnP OFF: There have been so many security vulnerabilities in
"Universal Plug and Plug" in routers over the years, that the first thing you
should 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
maybe turning UPnP back on (or manually fixing what stopped working).
Change the admin password! Change the 'administration' 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 find routers set to default credentials
(often 'admin/password'), which opens up the router to unauthorized changes.
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
Firmware updates: There are often times critical security related firmware updates
published for your router. If your router does not update firmware automatically, stay
on top of security updates.
Wi-Fi is a (time) shared resource. So, the goal for improving things is to get every
wifi 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 wifi time' for other wifi users:
Consider these options...
- Use Ethernet whenever possible
- Improve wireless PHY speeds
1) Use Ethernet whenever possible
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 wifi 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 940 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.
on cheaply adding 10 Gigabit Cat6a/RJ45 to your home network.
b) Best v2: MoCA 2.5 Ethernet: If you can't use/run Ethernet wire, but
have access to RG6 cables 'in the walls', 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). Expect MoCA 2.5 speeds (rated at 2500 Mbps)
to max out 1 Gbps Ethernet
(940 Mbps throughput). But, test in your environment to confirm. More information:
Wiki info on MoCA and
this interesting MoCA 2.0 adapter review (YouTube).
MoCA adapter setup
TIP: If possible, avoid older MoCA versions (like 2.0) that are rated at
lower speeds, and instead use MoCA '2.5'.
For example, Amazon just started selling the
Translite TL-MC84 (two MoCA 2.5 adapters).
c) Fair: Wireless bridge: Many high-end routers can be figured 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 4x4 bridge is connected to
a 4x4 router. Under good conditions, expect 1500 Mbps PHY and 750 Mbps throughput.
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.
TIP: If you have an ISP provided gateway device providing internet service, investigate if
there is already one MoCA adapter 'built-in' to the provided device (there often IS
for Verizon FiOS and Comcast gateways; and likely only MoCA 2.0).
MoCA is a great way to add wired Ethernet to remote locations (up to 16) in a house
that has cable TV wiring, but has no way to add/retrofit CAT5e.
Yes, you are trading one wifi (on the device), for another wifi (on the bridge). But the
PHY speed of wifi on the bridge is (hopefully) several times faster than wifi on the
devices you replaced. Only use this option if PHY speeds go up 2x or more.
d) OK to Bad: Powerline Ethernet: If you need wired Ethernet, but find it impossible to
run a CAT5/6 cable, give powerline ethernet a try. Plug in one next to your router.
Plug the others (yes, many are possible) exactly where they are needed. Look for
1000 Mbps (or higher) adapters. Actual throughput ranges from 20 Mbps to 300 Mbps
for these 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 so bad: (1) because acceptable speed depends
upon your situation and (2) powerline is cheap. So if you get the speeds you
need remotely (even if slow) -- job done.
If both powerline adapters are on the same circuit breaker, you can expect
fast speeds. But as soon as powerline adapters are put on different circuits,
which for most use cases is almost guaranteed, throughput may drop substantially.
TIP: Plug powerline adapters directly into the wall outlet (don't use power strips).
2) Improve wireless PHY speeds
a) Move as many devices as possible from wireless to wired: Avoid spectrum usage
and contention whenever possible. 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).
Everything is about TIME on the channel: Remember, it all comes down to
'time' spent on the wifi channel. Target the devices that spend the most time
on the wifi channel, and conversely, don't worry about (ignore) low channel width, low PHY
devices, that don't use wifi that much (eg: thermostat). The worst 'time' offenders will be
high internet usage devices with low PHY rates -- so target those devices first.
b) Look for the most unused channel (or just change 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, this 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 tranferring data all the time). Often times you just need to change the
channel and test (a lot).
c) Is your Wi-Fi router centrally located and unobstructed:
For smaller homes, place your wireless router at a location that is 'mostly' centrally
located to all of the clients that will use the wireless signal (the most). The goal is to
improve (overall) PHY speed for everyone, but do place the Wi-Fi to benefit the
'heavy users' the most. Also, place so that the router is unobstructed, as signal strength
is reduced as the signal must pass through anything (furniture, walls, etc).
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.
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 wifi channels on the AP/router.
Sometimes, repositioning your router slightly (which direction it faces) can help
signal strength for some devices. In larger homes, plan on a main router with one
or two AP's wired/Ethernet back to the main router.
d) 4×4 MIMO 'wave2' router: If you currently only have a 2×2 'wave1' router,
give a 4×4 MIMO 'wave2' router a try. While PHY speeds very close to the router
may not improve at all (could be maximum PHY speed already), the goal is to increase PHY speeds for
all wifi clients out there 'at a distance'.
Using the MCS Spy tool (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
the AP slightly.
And, of course, if looking at Wi-Fi 6 routers, consider only
"Wi-Fi 6 Certified" Routers, and understand that Wi-Fi 6E
is right around the corner.
e) Add a 4×4 'wave2' AP: If some of your wireless devices are too far
away from your main router, add an access point (must be wired to the main router via
Ethernet) where it does the most good (so faraway devices now get
top PHY speeds). Don't look at only 'access points', but also look into
higher-end routers that often times can be configured as an access point.
Or, move a heavy internet users (children) to their own AP (must be on an unused channel).
Done right, you can double (combined) wifi capacity by adding another AP. The
heavy users gets to max out their Wi-Fi channel, and all other users get to max
out their Wi-Fi channel.
f) A single low-PHY 'heavy' user can slow bandwidth for everyone: A single device operating at
a very low PHY speed, and using the channel a lot, can slow down an entire wifi channel.
Because what is critical is TIME spent on a channel (which increases as PHY speed
decreases). Adding an access point (and hopefully a new channel), and greatly
improving the PHY speed for that one device, can free up time needed by
other wifi clients.
Analogy: Imagine a highway (wifi 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 (wifi devices) wanting to use the highway
(wifi 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).
g) 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 double).
With a PC, find a way to use Ethernet wired to your main router.
h) 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. Each generation
of newer hardware performs just a little bit better.
i) Investigate 160 MHz channels: If your 2×2 client devices support 160 MHz
channels (this was rare, but it is becoming more common), 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 less range (than smaller channel widths) - details.
j) Try to keep 5 GHz reserved 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 the Interstate and simultaneously preventing three other lanes beside it
from being used.
k) Update firmware: Make sure that your router/AP is running the lastest
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
(eg: WAN to LAN performance bug).
But if the dual-band 802.11n device is a lightweight when it comes to wifi
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).
First, analyze the client devices that download/upload the most data. They should be
running at high PHY speeds; and if not, fix.
First, did PHY speed increase?:
Always check the PHY speed 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
to verify overall 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. A good way to test
wifi throughput is by transferring a file from one PC (wired) to another PC (wireless)
and looking at the OS provided network utilization graphs. Or, better yet, use a
dedicated SpeedTest program.
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. What really matters is not potential
(maximum) rated speeds but actual speeds possible.
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. 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 enough (for most people): You can
expect throughput of 455 Mbps (±45 Mbps) on a 2×2 MIMO client device. 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.
Wi-Fi 5 is good enough for 400 Mbps Internet: For the far majority of
people who have Internet speeds 400 Mbps (or less), Wi-Fi 5 is actually
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 section far above).
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 speed tests to confirm that the channel
performs well (not sharing bandwidth with others).
802.11ac beamforming/diversity really works: The one advanced feature in
802.11ac 'wave2' that really does work 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
at which devices connect to your
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'.
Don't go out of your way looking for this feature, but if it just happens to come
with a new router, fine.
WAN 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.
Wi-Fi 6 will initially have low impact: UNTIL the far majority of your wireless
devices are Wi-Fi 6 802.11ax, don't bother with an 802.11ax router.
Wait until 2021, and then, only look at Wi-Fi 6 certified routers (and don't be fooled by
a Wi-Fi 6 router certified for only Wi-Fi 5) -- this way you won't
get stuck with a draft (non-specification) router that does not support all mandatory
Most 'enterprise' installations use only 20 MHz channels: You can almost always get
by with an 80 MHz channel at home, but most 'enterprise' installations still only use 20 MHz
channels, and that reduces/limits throughput (a max around 100 Mbps is typical), but
increases range slightly.
Don't overlook Ethernet:
Wired ethernet 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 wifi usage plummeted (and the only
devices left on wifi were low bandwidth wireless only devices -- like smart thermostats).
Ethernet all of a sudden looks pretty cool when every smart TV in the house can RELIABLY stream
at the same time because NO Wi-Fi is being used!
YOUR client device often limits wifi 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 wifi is download (AP/router transmit power), not upload (client transmit power).
This observation was confirmed by using the
MCS Spy tool, which shows that clients are often
transmitting at a lower MCS level (than the MCS level an AP/router uses to transmit
to the client).
Increased range is NOT always a good thing: I was reading a post by someone
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 you think about it, it makes sense. Everyone on an AP shares that AP's wifi
bandwidth. Period. Which is exactly why you want shorter wifi range and more
AP's in dense environments -- so fewer people per AP means INCREASED wifi bandwidth
per person. The same principle applies to large homes, where you want everyone
evenly connected to several AP's, not everyone connected to one AP.
Have you ever tried to connect to a weak wifi network, only to have your client device
complain that it failed to connect? And then you wonder, 'but my device can clearly
see the wifi 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
Beware reviews testing 'ideal' situations: I have seen many online router reviews
test to a new router that is only feet away, or 'line-of-sight' in the same room as
the router. Of course the router should always get maximum speeds (1024-QAM)
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, etc.
Most vendors implements Wi-Fi using the same chipsets:
The two giants in the consumer router wifi chipset game are Broadcom and Qualcomm. Since these two
companies alone account for over 50% of market share, (almost) every vendor uses their
chipsets. So baring some major bug, all AP/routers within a 'class / generation / wave'
are comparable. So other factors, like vendor firmware/software and quality
of support, 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 7800, which
is just a "Qualcomm Atheros AP161 reference board".
But each new wave/generation of hardware/chips does seem to perform just a little bit
better than prior generations. Is this due to improved signal processing, lower internal
noise, or something else?
Be critical (and smart): There is no point in replacing your router if PHY speeds to your
wireless devices do NOT improve (by at least some reasonable amount).
So, be very critical. Take note of client PHY speeds
before and after a router update. If you see an improvement in PHY 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 PHY speeds actually increase!
Virtually all wifi devices (laptops / tablets / smartphones / smart tv's / etc) today are
STILL only 2x2 MIMO (at best; some are even still at 1×1). And THAT limits the speed
at which those devices will connect to any AP/router (not the max speed of the router).
Because of this, get a router/AP with a mininum of:
Fastest SPEED: If you have a compelling need for a little (11%) more speed AND can find
a "Wi-Fi 6 Certified" router (in your price range), go for it (but the benefit
is only for Wi-Fi 6 clients, not older Wi-Fi clients). Otherwise, a top of the line Wi-Fi 5
router is a great VALUE right now, and will do very well for most situations.
- Wi-Fi 6, or "wave 2" 802.11ac Wi-Fi 5 -
this is the best VALUE you can get today
- 4×4 MIMO -
increases signal reliability for all 2×2 MIMO devices, and ensures faster speeds for the rare 4×4 clients.
- 802.11ac beamforming -
improves signal strength, which increases the range at which devices stay at fast speeds.
- DFS channels -
because if at all possible, you want a channel all
to yourself -- you don't want to share a channel, and therefore bandwidth, with a neighbor.
- Wi-Fi Certified -
guarantees "interoperability, security, and reliability."
Also, watch out for routers certified to a lower specification than expected
(eg: a 802.11ax router certified for only 802.11ac)!
- AP mode -
for routers, look for one that also has an 'AP mode'. Because when you
do upgrade to a newer version of Wi-Fi, you want to reuse the old router as an 'AP' in your
new network (and not have it sit on a shelf).
I find it incredibly hard to justify a very expensive Wi-Fi 6 router for $600 that only
increases speeds by 11% for one or two clients, and provides NO speed improvement
whatsoever for all other clients in the house -- especially when first-gen Wi-Fi 6
hardware has not fully matured to support all Wi-Fi 6 features, and the
next generation Wi-Fi 6E is just starting to come out, which will
(yet again!) require new hardware (both routers and client devices).
Best VALUE: Get an mid-range Wi-Fi 6 router, or a top of the line Wi-Fi 5 AP/router (4×4:4 ,4 streams, 802.11ac "wave 2") that
supports beamforming and ALL six 80 MHz 5 GHz channels (42, 58, 106, 122, 138, 155)
Comcast TIP: If you have Comcast for your Internet and they are already providing you with a
'gateway', contact Comcast and say that you want their new best
XB6 Wireless Gateway,
which is an 8×8:8 MIMO (eight stream!) 802.11ac device, which supports data throughput of 1 Gbps.
Please note that the XB6 comes in two models. The TG3482G, which does NOT support DFS channels
(but later revisions do).
And the CGM4140COM, which DOES support DFS channels.
Comcast XB6 Gateway
UPDATE: Comcast has come out with a newer 3rd generation version,
called the XB7 supporting Wi-Fi 6 (FCC ID G954331X) -- you are eligible
for the XB7 if you subscribe to Comcast 300 Mbps (or higher) Internet speeds.
WARNING: But someone wrote to me stating that on the XB7
that it is no longer possible for the 'end-user' to select which Wi-Fi channel to
use, as that is now all 'automatic' and behind the scenes.
Wi-Fi 6: The
Netgear RAX50 is a mid-range
Wi-Fi 6 highly-rated AP/router for around $217, with 4×4 MIMO for 5 GHz only,
and supports all DFS channels.
Second best: Same as above, but select an AP/router that supports MOST
Wi-Fi 5 Router: A gem of an older high-end "Wave 2" router is the
actually supporting all DFS channels (via recent firmware).
Very widely used, with top marks in reviews
(but strangely, not Wi-Fi Certified!).
Usually available on Amazon for around $170. A great value, but
getting very hard to find.
Wi-Fi 5 AP: One very reasonably priced ($158) AP is the
4×4 "wave 2" 802.11ac AP and offers incredible value. Install where
it does the most good, and wired/Ethernet (via PoE) to your main router.
Try to buy an AP/router that is "Wi-Fi Certified" -- and avoid
draft specification devices.
Almost all Netgear routers that support DFS channels are in this category --
because Netgear only supports five (42, 58, 106, 122, 155) of the six DFS channels
(leaves out channel 138). See the Router Reference Appendix far below.
The ugly: Stay away from any AP/router that supports only the TWO standard
80 MHz channels (42, 155) and NO DFS channels (58, 106, 122, 138).
And in the consumer router market, there are a LOT of these.
Refer to the Router Reference Appendix for many examples.
Other: Don't forget to look into 'Enterprise' grade AP/routers. Ubiquiti sells a
line of 4×4 "UniFi AC" access point products that DO support ALL 5 GHz
channels, and are very reasonably priced.
For example, the
which is a 4×4 Wave2 AP for only $158 on Amazon.
A final thought: For most people, one great router centrally located 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 AP (wired to your main router) dedicated
and located nearby to those unique devices -- and this makes the most sense
when there is an unused wifi channel available.
One caveat - COST: The cost of the latest and greatest consumer-grade
(not even enterprise-grade) Wi-Fi 6 routers approaching $600 is insane. You get nowhere near that 'value'
given that virtually all client wireless devices are MIMO 2×2 limited.
You would (likely) be far better off spending that money on three 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 80 MHz channel. Distribute the AP's around so that everyone
in the house gets the maximum PHY speed possible!
TIP: The hidden YEARLY cost of electricity can really add up:
As a very general rule, the yearly cost
in electricity for any 'always on' device roughly equals wattage.
An old Netgear WNR1000v3 uses 4 watts ($4/year).
A Netgear R6250 uses 10 to 14 watts ($10-$14/year).
A Netgear R7800 uses 7 to 14 watts ($7-$14/year).
You can expect newer routers to use even more (Netgear RAX120 has a 60W
power adapter, but how much is actually used?).
A 100-watt light bulb uses 100 watts ($100/year)!
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 often under the control of your ISP
(you are NOT able to update/change firmware). 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.
Goal: With a modern (2×2) Wi-Fi 5/6 client device (phone, tablet, etc)
connecting to a modern (4×4) Wi-Fi 5/6 router, you should be able to easily
see and verify a 866 Mbps PHY connection (or better) between the two devices, when standing
right next to the router. Then as you move away from the router, PHY speeds
Disconnect/Reconnect wifi: You might be surprised how often simply disconnecting
from wifi on the client and reconnecting to wifi 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 wifi, 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.
Verify PHY speed: Go to your wireless device and
check the PHY speed
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.
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.
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 5 client, I would
expect to see a very strong signal with a PHY speed of 866 Mbps to a Wi-Fi 5/6 router.
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 wifi 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,
which then reveals a ton of information about how you are connecting.
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.
Verify router capabilities:
Install the WiFiAnalyzer (open-source) 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.
WiFi Analyzer Access Points
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.
Use the WiFiAnalyzer (open-source) 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.
Q: I upgraded my ISP Internet speed from 60 Mbps to 120 Mbps and my wired Internet speeds dropped!
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.
Q: I am connected to my wifi router at 866 Mbps, but a speedtest shows only 500 Mbps?
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 devices, causing that link to operate
at the problematic 100 Mbps speed instead of the desired 1 Gbps speed.
Due to wifi protocol overhead, the expected throughput at the application level is
around 60% to 80% of the physical (PHY) wifi 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 wifi router, but I can only connect at 650 Mbps (PHY) from my smartphone?
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?
Maybe, but maybe not.
Always try several different speed test programs, like
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?
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.
Q: Why don't I get fast wifi speeds from phone upstairs to my router downstairs?
to research versions of wifi supported by your smartphone (802.11n vs 802.11ac, etc).
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 further
away you move from the router. Sorry, but that is just how wifi signals work.
Q: Why is my client PHY speed stuck at 86.6 Mbps (or 173.3 Mbps)?
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 wifi client connecting
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?
Try a different channel on the router, or try updating wifi 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 wifi drivers on the laptop resolved the problem.
Q: My wireless Internet is horrible at video calls, what can I do?
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 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?
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"
for a discussion and instructions on how to disable. 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 speeds tests are sporatic and the modem diagnostic page shows that
one "Channel ID" has a lot of 'uncorrectable' errors. What is wrong?
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.
Also, the section above on how to improve wifi speeds
TIP: Digging deeper on Windows: If you are on a Windows computer, go into
a DOS cmd.exe window and type the following command. What is really helpful
is to see technical information about all SSID within range of the computer issuing
this command (a very simple and basic 'site survey'):
The table below lists only those routers/AP that: (1) support five or six 5 GHz
80-MHz channels (DFS channel support), and (2) have 4×4 MIMO.
Below the table is a list of routers eliminated
for NOT meeting these criteria.
Wi-Fi 6 routers that are not "Wi-Fi 6 Certified" are gray.
|RAX200 ||4×4||6||BCM43684, draft/ax|
|RAX120 ||8×8||6||QCN5054, draft/ax|
|RAX80 ||4×4||5||BCM43684, draft/ax|
|RAX45/50 ||4×4||6||BCM43684, draft/ax|
|Netgear (Extenders in AP mode)|
|EX7300v2 ||4×4||5||FCC=6, FW=5|
|Ubiquiti (PoE AP)|
|EnGenius (PoE AP)|
|GT-AX11000 ||4×4||6||BCM43684, draft/ax|
|RT-AX88U ||4×4||6||BCM43684, AX|
|Aironet 1850 ||4×4||6|
|Archer AX11000 ||4×4||5||BCM43684, draft/ax|
|Archer AX6000 ||4×4||6||BCM43684, draft/ax|
Don't forget that: (1) many residential routers can be configured as an AP,
(2) many commercial/enterprise AP's have reduced range (over residential
routers in AP mode) on purpose since they are designed to be installed
many at a time.
Routers/AP first eliminated from the table due to no DFS channel support:
Routers/AP then eliminated from the table due to lack of 4×4 MIMO support:
- Netgear: RAX20/RAX15, R8000/R7900, R7500v2, R7000/R6700
- Asus: GT-AC5300, GT-AC3100, RT-AC88U, RT-AC87U, RT-AC68U, RT-AC66R, Blue Cave
- TP-Link: Archer AX1500, Archer C5400X, Archer C5400, Archer C4000, Archer C3150V2
- Motorola: MR2600, MR1900, MR1700
- Linksys: MX5300, EA9500, EA9350, EA9200, EA8500, EA7500, EA7300
- D-Link: DIR-895L/R, DIR-890L/R, DIR-885L/R, DIR-882, DIR-879, DIR-878, DIR-867, DIR-859
- Amped Wireless: RTA2600-R2, RTA2600
- TRENDnet: TEW-829DRU, TEW-827DRU
- ZyXEL: Armor Z2
- Google: Google Wifi
- Eero: Pro 2nd gen (UPDATE: Eero now supports DFS with new firmware as of May 2020)
802.11ac chipset information:
- Netgear: RAX40, R7000P/R6900P, Orbi RBR50, Orbi RBR40
- Ubiquiti: UAP-AC-PRO, UAP-AC-EDU
- Synology: RT1900ac
- Linksys: EA9300, EA8300, WRT32X, Velop
- TP-Link: AX3000
- Cisco: CBW140AC
802.11ax chipset information:
- QCA9984 - Qualcomm 4×4 802.11ac Wave 2 + MU-MIMO + 160MHz
- BCM4366 - Broadcom 4×4 802.11ac + MU-MIMO
- MT7615N - MediaTek 4×4 802.11ac Wave 2 + MU-MIMO
Router Naming Convention: The common router N###/AC####/AX#### names
are obtained by adding up the maximum speed for all bands and then rounding.
Here are some common ones:
- QCN5054 - Qualcomm 802.11ax
- BCM43684 - Broadcom 4×4 802.11ax
- WAV654 - Intel 802.11ax
2.4 GHz speeds are cited for a 40 MHz channel.
5 GHz speeds are cited for either an 80 MHz channel (802.11ac) or an 160 MHz channel (802.11ax).
|Name ||2.4 GHz ||5.0 GHz ||5.0 GHz|
|MIMO||Mbps ||MIMO||Mbps ||MIMO||Mbps |
|N150 ||1×1|| 150 || || || || |
|N300 ||2×2|| 300 || || || || |
|N450 ||3×3|| 450 || || || || |
|AC1200 ||2×2|| 300 ||2×2|| 866 || || |
|AC1750 ||3×3|| 450 ||3×3||1300 || || |
|AC1900 ||3×3|| 600 ||3×3||1300 || || |
|AC2600 ||4×4|| 800 ||4×4||1733 || || |
|AC3200 ||3×3|| 600 ||3×3||1300 ||3×3||1300 |
|AC5300 ||4×4||1000 ||4×4||2166 ||4×4||2166 |
|AX3000 ||2×2|| 574 ||2×2||2402 || || |
|AX4300 ||2×2|| 459 ||4×4||3843 || || |
|AX5400 ||2×2|| 574 ||4×4||4804 || || |
|AX6100 ||2×2|| 400 ||2×2||2402 ||4×4||4804 |
|AX11000 ||2×2||1148 ||4×4||4804 ||4×4||4804 |
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 wifi settings, but what does that mean?
It means 80/40/20 MHz channel width! It does NOT change or adjust MIMO level
support (which is always on).
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.
The first step in setting up an AP/router is selecting a wifi channel that you think
is the 'most unused channel'. The second step is to verify that you are getting
the expected throughput -- 'around' 70% (±10%) of the PHY speed.
TIP: Reboot all of your test equipment before running a performance test. 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.
Use this SpeedTest program to easily both
test download / upload Mbps speeds between any two computers on your network,
which means it becomes very easy to test maximum download and upload
Wi-Fi throughput speeds. Just configure one test computer using Ethernet
and the other test computer on wifi, and run the Mbps throughput test
between the two computers.
802.11ac client with 866 Mbps PHY speed
Downloading from 192.168.1.14 port 33333...|
83,924,836 bytes in 1001 ms = 670,727,960 bps
84,090,176 bytes in 1000 ms = 672,721,408 bps
84,614,464 bytes in 1001 ms = 676,239,472 bps
84,137,504 bytes in 1000 ms = 673,100,032 bps
83,893,568 bytes in 1001 ms = 670,478,065 bps
Uploading to 192.168.1.14 port 33333...
63,125,888 bytes in 1000 ms = 505,007,104 bps
62,434,276 bytes in 1000 ms = 499,474,208 bps
61,928,032 bytes in 1000 ms = 495,424,256 bps
61,123,228 bytes in 1000 ms = 488,985,824 bps
61,451,072 bytes in 1001 ms = 491,117,458 bps
TIP: For the test, you only want ONE the computers on Wi-Fi, not both. Because
if both are on Wi-Fi, you will get a bad result (as both computers are using Wi-Fi
at the same time).
This speed test program is invaluable because it works on your local LAN and avoids
using your Internet connection, which may not max out your wifi speeds. Ideally,
your LAN is 1 Gbps, which should allow very accurate wifi download/upload speed
measurements. This is important as Tx PHY and Rx PHY for wireless can
sometimes be very different.
TIP: Run the speed test program LAN to LAN between two computers first, to confirm
that both your computers and your LAN can handle 1 Gbps speeds. A result of around
940 Mbps for your 1 Gbps LAN would be a good result.
Router WAN to LAN/WLAN throughput speed test:
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.
There are actually two PHY speeds for any wifi 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.
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.
So what is 'Link Speed'?: So is the 'link speed' displayed by wifi 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 wifi 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 wifi 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 wifi client link speed, the best way to see this asymmetry is in
throughput speed tests.
The best way to notice and see asymmetric PHY speeds is to place a
client device 'at distance' away from the AP/router being used. The
closer a client device is to the AP/router, the less you will notice
The bottom line: The PHY number that clients report is not yet very clear.
Is it Tx PHY, Rx PHY, or a comination 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.
A call for change in the industry:
Clearly a wifi client knows exactly what both Tx/Rx PHY speeds are,
as it is both sending and decoding the wifi signal.
The industry must change from reporting a single
"Link Speed" in wifi clients to instead reporting both transmit and receive PHY speeds.
Of note is that Ubiquiti routers and access points do report both speeds for wifi clients.
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 below, I found
that MCS 7 (650 Mbps) was being used for download and MCS 4 (390 Mbps) was being used
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.
PHY speed tables:
PHY tables for Wi-Fi can be found in this online
Google docs spreadsheet. 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 wifi 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
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 something to look into.
Other sources of PHY information:
mW: Signal strength in Wi-Fi is all about the mW (milliwatt, or 1/1000 watt). Most
wifi devices (routers, clients, etc) have a power output somewhere between 25 mW and 1000 mw.
But most devices receiving the wifi signal only see a signal strength of 'around' 0.00001 mW
to 0.0000001 mW.
Signal strength 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
(more details in the next section). 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 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.
mW to dBm: To convert from mW to dBm:
dBm to mW: To convert from dBm back to mW:
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.
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'.
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.
Multiplying mW by 10 equals adding 10 dB to dBm|
Dividing mW by 10 equals subtracting 10 dB from dBm
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.
For those who want to know, the actual 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 between two power levels (the difference between
two dBm -- 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 section are log base 10, or 'log10(#)'
- dBm - Wikipedia information on dBm
The short answer: The BEST option to improve Wi-Fi range (and more importantly, get
great speeds at range) is to install another AP exactly where it is needed
(and wired/Ethernet to your main router).
Try both bands: If your primary goal is long range (and not fastest speed),
first try connecting to your router's 2.4 GHz band SSID. 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.
Clients 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.
Client Device Range sorted by 5 GHz Range
|Range to an R7800 AP|
|Device||2.4 GHz CH 1||5 GHz CH 36|
|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|
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?
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).
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 distance', clients will see asymmetric PHY speeds.
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).
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 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
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.
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).
The Ring Video Doorbell Pro has a transmit power in 2.4 GHz of 82mW and an antenna gain
of 1.08 dBi (source).
Since both antennas are used in both directions, that is a essentially a 'wash' for
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).
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!
R7800 to Galaxy S6 long range
5 GHz smartphone example:
A Samsung Galaxy S6 is very similar, and results using
the MCS Spy tool
can be seen right, and confirm the very asymmetric PHY speeds (as 'router to client'
power is much higher than 'client to router' power).
So on paper, the 2.4 GHz band has more range than the 5 GHz band, and in real-world
testing, that happens most of the time, but not all of the time. Once in a while,
the 5 GHz band provides just as much range (and much better throughput). You just need
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).
12 dB hit (6 dB DFS + 6 dB 5 GHz)
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.
Diversity / Beamforming: A couple of unknowns that needs more research is
how much impact antenna diversity and beamforming have. They do have an impact,
but exactly how much? I have read a couple of dB (as in 1 dB or 2 dB),
but have not verified.
Of note is that the FCC discusses a 'maximum' gain when doubling antennas
of 10×log(NANT/NSS) dBi, which for
2×2 clients and 4×4 AP's, would be a maximum diversity gain
of 3 dBi, so presuming an actual gain of around 1.5 dBi seems reasonable.
"Can't I easily increase range by using high-gain antennas on my router?":
Yes, very likely.
But increased range in wifi is actually a 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 wifi devices in that extended range! Meaning the likelihood
of seeing neighbors' routers and sharing a wifi 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 wifi signal strength and still maintain good bandwidth".
The answer is: by installing another AP exactly where it is needed,
wired/Ethernet to your main router.
"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.
Also, high-gain antennas are not without consequence. They work by altering
the 'shape' of the signal. Namely, instead of sending the wifi 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.
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 wifi range extenders?": In general, don't use them at all.
By definition they consume wifi bandwidth to perform their job (every packet is sent
over Wi-Fi twice). Instead, an AP wired/Ethernet to your main router is
Less range in an AP can mean higher throughput: Counterintutively, 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).
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 in both 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 may not be true. I frequently see a big jump in
performance switching to the 5 GHz band. The ultimate cause is noise
floor differences. It does not matter if 2.4 GHz has a stronger signal
strength if that also means a higher noise floor!
A final note that explains some strange observations:
The asymmetric Tx power levels between a client and AP can cause strange
observations. 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.
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 very significant 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.
Wi-Fi signal strength decreases VERY quickly with distance, even in free
space (with no obstacles). But why?
How does mW signal power relate to distance?:
The discussion in the section above was all about raw mW power levels,
but how do power levels relate to distance traveled by a wifi signal?
Naively, twice the mW power means twice the distance, right? NO. To understand why not,
we must first understand the
which states that changes in wifi 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.
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).
Formulas: A change in distance squared is inversely proportional to the
change in power:
TIP: You don't have to memorize the Inverse-square law formula. Instead, just
remember this 'sphere' analogy and you can easily derive the relationship between distance
and signal strength.
And simplifying slightly, we get:
(new_dist/old_dist)2 = (new_pow/old_pow)-1|
(new_dist/old_dist)2 = old_pow/new_pow|
TIP: The best way to use this formula is to fill in the three terms that you do know, and
then solve for the one remaining unknown term.
And cross multiplying the above formula, you get the following formula
that makes a lot of intuitive sense. The 'old' side of the equation (left side) is a fixed
value. Then on the 'new' side (right side), you can change either term (power/distance), but
there must be a corresponding (and inverse) change in the other term (distance/power)
to keep the equation balanced:
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
is 3 dB twice, or 6 dB.
old_pow × old_dist2 = new_pow × new_dist2|
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.
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 further (distance) can the higher power router
communicate with wifi clients? Trick question, because virtually always, the answer
is no change at all, because the wifi 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).
BUT, if you have a wifi 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:
Netgear R7800 to Samsung Galaxy S6 on 2.4 GHz band
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 wifi connections is between -40 dBm (pretty close to a
router) and -65 dBm (any further away and lower throughput may be very noticeable).
But, in Example 1 above, how much should dBm improve on wifi clients, hopefully resulting
in slightly better PHY speeds (download) from the router to wifi clients, but not a better PHY speed
from wifi clients to the router (upload). Answer: Up to 10×log(975)-10×log(216)
dB, or 6.5 dB.
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.
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.
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 further 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.
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:
(1) the range of the Wi-Fi signal, and
(2) signal strength/quality for all clients.
|20 MHz || ||100% ||×1|
|40 MHz ||-3 dB ||71% ||×2|
|80 MHz ||-6 dB ||50% ||×4|
|160 MHz ||-9 dB ||35% ||×8|
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).
Wider channels reduces range: Each time you double Wi-Fi channel width
(20->40, 40->80, 80->160) you decrease Wi-Fi range by around 30%, or signal
strength by 3 dB. This is a key reason many low-bandwidth IoT devices intentionally want to
stick to the smallest channel width possible and avoid 802.11ac (which mandates 80 MHz
channel support) and stick with 802.11n (which has 20 MHz channels), because that allows
operation at the longest distance possible.
Maximizing Wi-Fi range (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
(Netgear's name for 'channel width' is 'Mode').
Clients should now see the ability to
communicate with the access point from a slightly further distance
(albeit at a slower speed and on a 20 MHz channel).
TIP: Alternatively, some clients 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'.
Client devices often limit range:
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 from the router
(and not the router itself).
If your client device is able to see a weak router SSID in the wifi 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 section is more about understanding Wi-Fi.
Changing the 5 GHz channel width to 20 MHz does extend range (a little). But
if it is range you need, just connect to your router's 2.4 GHz band instead,
which 'should' provide more range (than even a 5 GHz 20 MHz channel),
or install an AP (wired/Ethernet to your router) where it is needed.
Noise floor: When a router tunes into a wifi channel and amplifies 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). 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 you
Signal level: Now start communicating on the channel and examine the dBm level
of the signal on that channel (green line in graph right). The dBm level of that
received signal is called RSSI. Often times between -35 dBm and -70 dBm.
"Noise is defined as any signal other than the one being monitored"
Technically, there are ways to communicate at signal levels below the
'noise floor' (like LoRa), but that is beyond the scope of this paper.
Modern Wi-Fi relies upon a signal level well above the 'noise floor'.
RSSI stands for "Received Signal Strength Indicator"
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"
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, because the 'noise floor'
is so incredibly low (high SNR). But talk normally in a bar, and you won't be heard, because
the 'noise floor' is so high (low SNR). 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.
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 of this is that regardless of RSSI (even a great 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 wifi to work can be very
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 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,
SNR vs channel width: Each time you double channel width -- from 20 MHz to 40 MHz
to 80 MHz to 160 MHz requires 3 dB more in SNR to maintain the same modulation/coding
and if you don't have that SNR headroom, the modulation/coding rate will drop
(see Wi-Fi Channel Width vs Range for why).
MacOS: An exception is that MacOS displays Noise dBm right next to
Signal dBm. Very helpful. Use it.
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).
Here are the minimum IEEE 802.11 dBm values that each modulation/coding should be able to be
supported. If a router/AP can do better (support the modulation/coding at a worse
RSSI), that is OK. These are minimum values:
Now you know one key reason why many low-bandwidth IoT devices stuck to 20 MHz channels
Here are the minimum IEEE 802.11 SNR that each modulation/coding requires. Take
these as ballpark figures, as your mileage may vary:
Atheros RSSI is really SNR: In routers when you obtain RSSI from
system tools that use a Qualcomm Atheros wifi chipset, the RSSI value
is the dBm signal level with the noise floor subtracted out. That is just
Many AP/routers have the ability to 'telnet' into the device and run Linux commands.
This section 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; further below).
This has been invaluable to analyze the wifi 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 wifi (so all wifi stats on one wifi band must be from the device
being tested). And then on your PC, telnet to the router via Ethernet (or the
other wifi 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 wifi statistics
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
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.
wlanconfig ath0|ath1 list sta: [Atheros] Lists every wifi 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
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 wifi channels, cam upload speeds (the majority of
everything the cam does) improved significantly (now averages 27 Mbps):
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'.
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 wifi band. See also "ip link show", which
enumerates all interfaces on the router.
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:
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.
A very cool graphical Atheros "MCS Spy" tool: Check out this
MCS Spy 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
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).
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, 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:
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:
points to using this command:
with some of the valid "CMD" being:
qcsapi_sockrpc CMD wifi0 0
Linux in general:
A very helpful command is:
where 'DEVICE' is 'wlan1' (or similar). You will see information like:
iw dev DEVICE station dump
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 (using TCPDUMP):
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.
I then plug 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 184.108.40.206 firmware.
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.
- 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).
- 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
- 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.
- 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.
- 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)
to download the PCAP file to my PC.
- Analyze: I analyze the PCAP with WireShark.
A better 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' (no local PC required). Conversely, the big advantage of
the port mirroring capture technique is that it is much easier (and a lot less technical)
but requires your PC to be plugged into the R7800.
WiGig (802.11ad) is effectively dead in routers for Internet access. It just never 'took off'.
802.11ad: 802.11ad (also called WiGig)
is being marketed as the 'fastest' wifi 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.
Dubious Netgear 802.11ad marketing
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.
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'.
[RF loss calculator]
Interestingly, 802.11ac products already exist TODAY that provide 4.3 Gbps
(Arris TG3482G using the
That kind of puts Netgear's marketing hype ("3X faster than 11ac") into perspective.
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 wifi 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 it IS being used for very short distance point to point
(laptop computer to dock).
Also, you know things are VERY BLEAK for 802.11ad in routers when
Netgear has a tough time pointing out ANY client devices
that actually support WiGig!
UPDATE: I expect future Wi-Fi 6E tri-band routers WILL make a lot of sense:
The key problem with tri-band routers today is that TWO of the bands are the same
frequency (5 GHz). However, I fully expect that future Wi-Fi 6E routers will be tri-band
and be just fine -- because they will cover 2.4 GHz, 5 GHz and 6 GHz
(three separate non-overlapping frequency bands).
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.
And since (virtually) all Netgear routers are missing support for channel 138, that may limit
your options running 'dual 5 GHz'.
Instead of tri-band, use a second AP instead:
BUT, if you are this situation, if is far more cost effective to just add a second AP
($158 for a high end Ubiquiti nanoHD Wave2 AP) -- 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.
The bottom line: Add wired Ethernet everywhere you can in a new home,
like to AP locations and all streaming device locations, etc.
LAN/Phone Structured Wiring Cabinet
Gray Cat5e = 4-line Phone / room
Blue Cat5e = LAN1 + Internet / room
Yellow Cat5e = LAN2 + Internet / room
Red Cat5e = Thermostats (not used)
CATV Structured Wiring Cabinet
Black RG6 = CATV / room
White RG6 = spare RG6 / room
Red Cat5e = Internet to each CATV
But isn't the future all wireless?
If so, then so is slow Wi-Fi and packet loss! Yes, wireless continues
to improve, but the simple fact is that wireless can't come close to the
incredible speed and reliability of wired connections.
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!
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 wifi 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 wifi 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 CAT6a everywhere! To every
(a) telephone jack location,
(b) CATV location,
(c) wired LAN location,
(d) thermostat location,
(e) doorbell location,
(f) security camera location,
(g) access point locations,
(h) desk location,
(i) game console location, and
(j) any other location you can think of,
run the wires needed for that service, but also add spare CAT6a as well. The cost of doing
this during initial construction is minimal compared to the inconvenience of trying to add
wires after-the-fact (which often times, is impossible).
Possible alternative?: Consider adding strategic 'smurf cable' (empty
conduit) runs to many locations.
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.
TIP: There is a lot of wire coming out of China that is CCA
(copper clad aluminum) that is pure JUNK, and actually
violates building codes. Make sure the wire you install in the walls is
quality 'solid bare copper' wire (not CCA) and not stranded.
CAT5 is versatile:
You can run almost anything over CAT5 cable, with the right adapter
(USB / VGA / HDMI / audio / etc).
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 wifi is used for streaming). Also, since
every room had two wired Ethernet connections, several were
repurposed for AP's.
My experience: I built a new house in 2005-2006, and at that time, here
was my structured wiring checklist:
(a) Every room/desk phone location got one Cat5e (four phone lines) and one spare Cat5e.
(b) Every room/desk Internet/LAN location got two Cat5e (each connected
to a different switch in the structured wiring cabinet - for redundancy).
(c) Every CATV location got two RG6 and one spare Cat5e.
(d) Every thermostat location got a spare Cat5e.
(f) CATV demarc location got one RG6 and one spare RG6.
(g) Phone demarc location got one Cat5e and one spare Cat5e.
(e) Possible satellite dish demarc location got two RG6.
In hindsight, the changes I would have made:
The bottom line: It is very inexpensive (and easy) to add Cat-5e when
a house is being built. However, it is almost impossible (and costly)
to add Cat-5e 'after the fact'.
- I should have run spare Cat5e to:
(a) doorbell locations,
(b) security camera locations,
(c) security alarm base station location,
and (d) all potential Wi-Fi AP locations.
- 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.
- 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.
- Plan on having 'expansion room' in the structured wiring cabinets for
future changes (for me, changes: Internet to all TV locations,
VoIP telephone, PoE security cameras, etc).
- 2.4 GHz / 5 GHz / 6 GHz / 60 GHz: Refers to the wireless frequency (spectrum/band)
used by wifi.
- 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.
- 802.11ax: Wi-Fi 6: The specification for HE (High Efficiency) Wi-Fi 6.
- 802.11ad: The specification for Wi-Fi around the 60 GHz band.
- 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 wifi
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.
- Beamforming: A standards-based (802.11ac) signal-amplification technique that
results in increased range and speed to a device. Beware (avoid) earlier
(proprietary) 802.11n beamforming implementations.
- Client: Refers to any wifi device (phone, tablet, console, TV, etc) that connects
to an access point (AP).
- 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.
- 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. See Wikipedia.
- 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.
- '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.
- PHY: PHY is an abbreviation for physical. For example, 'PHY speed' refers to
the physical speed at the raw network layer. For every wifi 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.
- QAM: Quadrature amplitude modulation
A method of converting and sending digital information (0's and 1's) over wires using
- Quad-Stream: Refers to 4×4 MIMO.
- SNR: Signal to Noise Ratio. The difference (in dB) between the
signal level and the noise level.
- 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).
- WAN: Wide Area Network (the WAN port on your router is connected to a
modem, which in turn is connected to the Internet).
- "Wave 2": An 802.11ac term used to define chipset and feature level. "Wave 1" was the first
generation, supporting core basic features. "Wave 2" was the next chipset that added many optional
- Wi-Fi: A 'brand name' for wifi created by an alliance
of companies for IEEE 802.11 wireless technologies, to ensure that wireless
products from different vendors all work with each other.
- WLAN: Wireless LAN
Aerohive Tech Articles:
Aerohive Wi-Fi 6 802.11ax technical blog articles -- sadly, all of these links appear dead
after Aerohive merged with another company:
Learn more on other web sites:
FCC information and filings:
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 for Dummies eBook
(requires form to be filled out). Alternatively, look at this
2nd source for the book
(no form to fill out).
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 router/AP upgrade
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 contact form to contact the author of this paper, Jerry Jongerius.
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 © 2018-2021 Jerry Jongerius
As an Amazon Associate I earn from qualifying purchases.