Increase performance of home WiFi

Psycho · Thu Dec 03, 2020 2:12 pm

Hello,
I am experiencing bad bad performance on my home WiFi. I know, both devices are "older" a bit, however, the WiFi is slower than expected.

Topology can be seen on the attached image, both devices are connected & configured uring CAPsMAN. As seen from the wireless interface configuration, I am using G/N only 20MHz channel on 2412 (channel 1). I have tried to fiddle with antenna gain a little bit, but no results so far.

Observations:

In rooms 2 and 3 signal seems to be OK, however, the speed is extremely low -> I have a spare 951, I will try to place it in room 2

When I go from room 4 to room 1, my laptop gets disconnected for approx. a minute. In the opposite direction (1->4) it works fine

Speed measurement (cable only for reference)

speed-measuremenet.PNG

Notes:

The link between 2011 and 951 is 100 Mbit only (I cannot upgrade to 1 Gbit, cables are in the wall (I am aware of this bottleneck).

Ingress from ISP is 150/10 Mbit, I can observe around 10 Mbit upload all the time - this is not a big issue.

My WiFi is called SUSHI[/
- HW info:
  - RB2011UAS-2HnD | 6.42.3
  - RB951-2n | 6.47.1
  Any suggestions on how to improve wifi throughput?

troffasky · Thu Dec 03, 2020 3:31 pm

Start by testing internally with a tool like iperf, rather than to the internet.

mkx · Thu Dec 03, 2020 5:46 pm

Well, using internet speed test servers doesn't explain low throughput in some rooms. And doesn't explain the huge difference between different services in same room. However some good signal measurements would. The published chart doesn't tell much without knowing what happened at particular moments. OP should measure wireless signal (in dBm) and see if signal strength is an issue (anything lower than -70 dBm is not good). If signal strength is decent, then scanning for interference is next thing to do. Check channel occupancy ... for all channels occupied by AP. If there are some (or even many) other APs using channel which overlaps the one used by OP's AP, then this is a problem.
And make sure that RB2011 and RB951 are using different non-overlapping frequency channels.

Psycho · Thu Dec 03, 2020 7:16 pm

Thanks for the answers and corrections!

I have made local measurements with iperf. Results are confirming my hypothesis about bottleneck in WiFi between20 - 30 Mbit.

Two sets of measurement has been done - first is using my desktop with W10 and one laptop with unknown WiFi card (however it is one year old so I suppose modern standards).
The second is using the same laptop with W10 and Macbook from late 2019.

I will do a signal measurement and also switch one AP to channel 6 (room 1) while keeping the other one set to channel 1.

Maybe I am wondering about 2.4 GHz limits and upgrading to 5 GHz.

btw, who is OP?

R1CH · Thu Dec 03, 2020 7:57 pm

2.4 GHz is usually pretty bad except in remote places, way too much interference. You should also enable WMM if you want 802.11n to work.

Psycho · Thu Dec 03, 2020 8:44 pm

I have done a small "site survey" to map signal levels across the flat. Confirming huge issues in room 2.
Note: Signals are in dBm and should have the prefix "-".

+ I have enabled WMM, thanks.

LuckyMan · Thu Dec 03, 2020 11:19 pm

Thanks, it helped me a lot!

Psycho · Fri Dec 04, 2020 11:01 am

I have done channel analysis near the routers and I have to say: channels are pretty busy :D

I am considering 5 GHz more and more.

gotsprings · Fri Dec 04, 2020 5:41 pm

I say it all the time... need better wifi... change vendors.

Mikrotik routing... awesome.
Mikrotik wireless... they released a Development Software recently to finish a standard ratified in 2016.

For anyone keeping track or can't add... that's 5 years out of date.

https://en.wikipedia.org/wiki/IEEE_802. ... ess_points

maigonis · Fri Dec 04, 2020 11:37 pm

I have done channel analysis near the routers and I have to say: channels are pretty busy :D

I am considering 5 GHz more and more.

5GHz is a way to go nowdays, not because of MT WiFi quality. 2.4GHz have way too noise, interfierance and low troughtput.

kalamaja · Fri Dec 04, 2020 11:56 pm

Try this, it has helped me a lot in noisy 2.4Ghz environments:
/interface wireless set [ find default-name=wlan1 ] adaptive-noise-immunity=ap-and-client-mode

bpwl · Sat Dec 05, 2020 12:23 am

I see all sorts of "indirect" measurements, good enough to test if all is working fine, but not specific enough to identify the cause why it is working OK or not.

The information is in the Mikrotik, very easy to access and gives a very clear indication what is going on with every wifi connection.
The information is in the "Registration" tab of "wireless", you must add more columns or use "detail mode". (CLI "/interface wireless registration print stats")

rx-rate=, tx-rate="65Mbps-20MHz/1S/SGI" -> interface rate -bandwidth/# of streams/short guard used
frames=, hw-frames= -> ratio defines # of retransmits
signal-strength=, signal-to-noise=38dB signal-strength-ch0=-74dBm signal-strength-ch1=-70dBm -> signal should be above -75 dBm, SNR larger than 25 dBm
tx-ccq=, p-throughput= -> CCQ is quality of signal, % of transmission that is successfull.

This at least is about WIFI ONLY characteristics.
When this is in perfect condition then;
Expect on 2.4 GHz an interface rate of 144Mbps-20MHz/2S/SGI
Expect on 5 GHz an interface rate of 400Mbps-40MHz/2S/SGI

The free airtime is seen with "Freq Usage". (100-Freq Usage) is the free time.

Payload data rate is 50% of (interface rate * CCQ * (100-frequsage)) if in one direction (half duplex) , 25% if in full duplex.
So the 400 Mbps of 5 GHz with 100% CCQ and full free airtime is the same as a 100Mbps ethernet cable in full duplex load.
Repeating the wifi signal will half the available airtime and throughput.
So repeated 2.4GHz (wifi-wifi) with perfect connection (144Mbps-20MHz/2S/SGI),(100%CCQ, 0 Freq usage) will allow up to 36 Mbps half duplex. (TCP is always partly full duplex for the ACK's)

Psycho · Sun Dec 06, 2020 12:52 pm

@bpwl I did best I can. thanks for your post.

I was unable to find any information in Wireless/Registration, probably because I am using CAPsMAN to manage my APs.

Below are stats with 3 devices connected to remote CAP:

> caps-man registration-table print stats 
 0 interface=CAP-951 ssid="SUSHI" mac-address=31:.. tx-rate="57.7Mbps-20MHz/1S/SGI" rx-rate="1Mbps" 
   rx-signal=-72 uptime=21m36s710ms packets=156,221 bytes=11029,19123 tx-rate-set="OFDM:6-54 BW:1x SGI:1x HT:0-7" 
   eap-identity="" 

 1 interface=CAP-951 ssid="SUSHI" mac-address=21:.. tx-rate="65Mbps-20MHz/1S" 
   rx-rate="58.5Mbps-20MHz/1S" rx-signal=-46 uptime=3m6s680ms packets=572,886 bytes=77093,1009994 
   tx-rate-set="OFDM:6-54 BW:1x SGI:1x HT:0-7" eap-identity="" 

 2 interface=CAP-951 ssid="SUSHI" mac-address=F1:.. tx-rate="6.5Mbps-20MHz/1S" 
   rx-rate="6.5Mbps-20MHz/1S" rx-signal=-39 uptime=3m2s770ms packets=8562,7125 bytes=6121590,2016848 
   tx-rate-set="OFDM:6-54 BW:1x SGI:1x HT:0-7" eap-identity=""

When I connect to the main 2011, stats can be found in Wireless/Registration and are the following:

> /interface wireless registration-table print stat

interface=wlan1 mac-address=F1... ap=no wds=no bridge=no rx-rate="130Mbps-20MHz/2S/SGI" tx-rate="72.2Mbps-20MHz/1S/SGI" packets=65676,40558 bytes=82658488,20721003 frames=36773,40560 frame-bytes=83349089,20477891 
   hw-frames=44517,41318 hw-frame-bytes=104410186,22253242 tx-frames-timed-out=0 uptime=1m5s last-activity=10ms signal-strength=-53dBm@6Mbps signal-to-noise=45dB signal-strength-ch0=-55dBm signal-strength-ch1=-58dBm 
   strength-at-rates=-60dBm@1Mbps 44s800ms,-53dBm@6Mbps 620ms,-54dBm@HT20-1 37s960ms,-53dBm@HT20-2 18s980ms,-54dBm@HT20-3 210ms,-54dBm@HT20-4 130ms,-53dBm@HT20-5 5s630ms,-53dBm@HT20-6 10ms,-53dBm@HT20-7 1s100ms tx-ccq=94% 
   p-throughput=63990 last-ip=10.0.80.23 802.1x-port-enabled=yes authentication-type=wpa2-psk encryption=aes-ccm group-encryption=aes-ccm management-protection=no wmm-enabled=yes tx-rate-set="OFDM:6-54 BW:1x SGI:1x HT:0-15"

Psycho · Sun Dec 06, 2020 12:54 pm

@kalamaja Thanks for the response - I will try to configure and verify this configuration!

bpwl · Sun Dec 06, 2020 1:34 pm

Not so much to see. I don't know where the rest of the CAPsMAN information can be found.
frames/hw-frames don't point to extreme interference on the RB2011.
However you own SUSHI channels do overlap. This will introduce co-channel wait times. The adaptive-noise-immunity, very good for weak background signal suppression, will not help for this co-channel wait.
One SUSHI is even very strong (>-30dBm) , testing very close or too strong AP ?
Clients only connect at 1S (single stream). This may be due to client hardware, but also limits expected PHY to 72Mbps, with a max data rate of arround 35 Mbps.
The PHY at 6.5Mbps is rather problematic, and this with a strong signal of -39dBm. This does not even take the interference into account, unless the interference is dropping packets and as such forces the PHY to select a lower rate (lower more robust MCS). This can be caused by interference, diffraction (by wall), reflection (by wall and metal surfaces), distortion (by wall and other objects and by using a too strong AP signal).

anav · Sun Dec 06, 2020 3:35 pm

If the title of your post is a general question, then I can state I have answered it by using this device vice capac, the tp-link eap245. Or the tp-link AX10 for that matter. Plug it in, it works, good for your stress levels and bpwls ;-))))

mkx · Sun Dec 06, 2020 4:10 pm

The PHY at 6.5Mbps is rather problematic, and this with a strong signal of -39dBm.

I think that stats in registration table are dynamic. During low (or no) client activity it will show basic connection rates which should rise to maximum achievable rate only if there is some activity between AP and station.

bpwl · Sun Dec 06, 2020 6:04 pm

Yes. Correct. PHY is always low without traffic (and CCQ is quite bad).. But usually it stays high for a while when it gets at higher rates with traffic.

I assumed " rx-rate="6.5Mbps-20MHz/1S" rx-signal=-39 uptime=3m2s770ms packets=8562,7125 bytes=6121590,2016848 " had enough traffic to show better values. I may be wrong.
But OK, you need to see it life to judge properly. Even while walking around to find the good and bad spots.

Psycho · Wed Dec 09, 2020 12:21 pm

Thanks all of you, guys...

I have decided to buy better APs. I bough two D-LINK COVR-1102 and now I get much higher speeds and better coverage. But I still keep rb2011 as the main router :)

Thank you very much!

Increase performance of home WiFi

Increase performance of home WiFi

Re: Increase performance of home WiFi

Re: Increase performance of home WiFi

Re: Increase performance of home WiFi

Re: Increase performance of home WiFi

Re: Increase performance of home WiFi

Re: Increase performance of home WiFi

Re: Increase performance of home WiFi

Re: Increase performance of home WiFi

Re: Increase performance of home WiFi

Re: Increase performance of home WiFi

Re: Increase performance of home WiFi

Re: Increase performance of home WiFi

Re: Increase performance of home WiFi

Re: Increase performance of home WiFi

Re: Increase performance of home WiFi

Re: Increase performance of home WiFi

Re: Increase performance of home WiFi

Re: Increase performance of home WiFi

Who is online