Improve WiFi real-time performance in the presence of foreign WLANs

0

I want to use IEEE802.11 WiFi in infrastruture mode to communicate in both directions between an access point and several stations.
The amount of traffic between the stations and the access point is not very high and thus should not cause significant delays due to collisions between my stations and the access point in the CSMA/CA protocol.

However, the problem is that this setup could be operated in environments with several other WLANs present independent of my setup and not under my control. There could be heavy traffic in those WLANs and it is not predictable, which channels in the 2.4GHz band they will use, nor according to which of IEEE 802.11b/g/n they will operate. They may not even leave any free channel to use for relatively undisturbed transmission for my setup. Therefore, naturally I expect that my devices will contend with those other devices and will be subject to frequent collisions and thus increased delay times due to the backoff mechanism in the CSMA/CA protocol.

My goal is to have the least possible delay times for the WiFi traffic in my setup to have best possible performance. Ideally there would be a real-time guarantee for one request-response message exchange cycle between a station and an access point.

Is there any good way to reduce delay times by avoiding collisions with traffic from other WLANs (switching to 5GHz is not an option btw.)?
As far as I understand I will not be able to profit from a IEEE802.11e QoS implementation such as EDCA, since those only manage channel access within my WLAN. However, within my WLAN all messages have equal priority. Did I understand this correctly?

In case I am in the wrong stack exchange, please let me know!

oh.dae.su

Posted 2018-01-17T20:54:28.920

Reputation: 288

I think you misunderstood 802.11 QoS. It lets you mark some traffic as higher priority (VO/VI) , allowing your clients to use shorter inter-packet wait times when sending that traffic, which means they get on the air sooner / more often than any normal priority (BE) or low priority (BK) traffic from ANY network in radio range. – Spiff – 2018-01-17T21:44:57.240

Ok, thanks for the hint. If I understand you correctly, this actually means that IEEE802.11e would help in my scenario to reduce delays caused by any other network in radio range. On the other Hand, if the other Networks would use 802.11e QoS and mark high priority traffic, I could suffer from even worse delays. Do I underand this correctly? – oh.dae.su – 2018-01-17T22:07:35.450

Yes, marking your latency-sensitive traffic as VO should definitely help you, and Yes, if you leave your latency-sensitive traffic marked as BE and someone else in radio range has VO traffic, their VO traffic will make your BE latency worse. – Spiff – 2018-01-17T22:13:28.817

Answers

1

My goal is to have the least possible delay times for the WiFi traffic in my setup to have best possible performance. Ideally there would be a real-time guarantee for one request-response message exchange cycle between a station and an access point.

You have essentially ruled out unlicensed radio traffic as a solution that fits your needs.

In the United States, the FCC has a variety of documents available, one example of which provides a brief overview of the kinds of interference you'll see on unlicensed radio bands like 802.11 Wifi uses (both 2.4Ghz and the 5Ghz bands); there include but are not limited to other wifi networks, Bluetooth (2.4Ghz), voice radios, other data transmissions, toy radios, other intentional radiation, and unintentional radiation (microwaves, lights, other electronics), and so on and so forth.

You can do whatever you like within WLAN, but a bad microwave oven nearby is going to toast your signal, and on an unlicensed frequency, you're pretty much stuck with it.

Never mind 802.11 deauth attacks.

Never mind anyone else doing whatever they want on 2.4Ghz or the 5Ghz bands within regulatory bounds in your jurisdiction, quite possibly using bigger antennas than you.

If you really want performance guarantees over a radio link and you have the budget, while you won't get a real-world actual "real-time guarantee" or even "least possible delay times", but with a licensed wireless setup in your jurisdiction (hire a contracter expert in this area; they can fill out the paperwork - including local jurisdictional issues - and do the install for potentially very reasonable fees), you have serious regulatory recourse to interference.

In all other cases, either take radio off the table, or take your performance requirements off the table.

The laws of physics compel you!


Edited to add: If you insist on using Wifi, per the comment.

  • Quit using 2.4Ghz; move to 5Ghz at a bare minimum. 2.4Ghz is too crowded, and has not enough non-overlapping channels
  • Understand the regulatory (legal) environment you're working in
  • Use radios with highly directional, highly noise isolating antennas (low sidelobe, high front-rear isolation). Two very inexpensive examples are the Ubiquiti Isostation and the Ubiquiti Prismstation
  • Perform frequent RF site surveys with directional antennas, and relocate your equipment to keep the worst interference away from your antenna's best lobes, i.e. out to the sides

Anti-weakpasswords

Posted 2018-01-17T20:54:28.920

Reputation: 481

Thanks a lot for your detailed answer. So I understand that real time guarantees are off the table using WiFi in the 2.4 or 5 GHz bands. Still I am wondering about one thing: How about providing the least possible delay times using WiFi in the 2.4 GHz band. From Spiff's comment I get that using priorities for my frames via IEEE802.11e would help me in that respect. Is there any other QoS measure that could reduce possible delay times, when dealing with neighbouring WiFi networks? Thanks a lot in advance! – oh.dae.su – 2018-01-20T22:55:21.700

@Jakuzzi81 Edited answer, but you really, really need to get off Wifi. I'm ignoring 801.11e on the basis that it appears you are expecting serious contention with OTHER wireless networks, not WITHIN your own, and thus you hit this issue " The HCCA provides the means to deliver time-bounded traffic, but requires all stations within the range of the HC to follow its coordination." - stations not on your wifi network will obviously not follow your coordination, and usually not support 801.11e

– Anti-weakpasswords – 2018-01-22T06:54:47.447

Highly appreciate your detailed explanations. Currently a switch away from 2.4GHz Wifi is not possible. But you have provided me enough arguments to convince my colleagues that either they have to lower their expectations regarding real-time performance of WiFi or that we need to actually look at different solutions for communication. One last thing: Basically, all wireless communication is susceptible to jamming of the frequencies used and therefore cannot guarantee real-time behaviour ultimately, correct? – oh.dae.su – 2018-01-23T13:41:53.317

@Jakuzzi81 - Don't call it jamming. Jamming is intentional interference, and in the U.S., strictly against FCC rules. All radio communication is susceptible to interference. As an unlicensed part of the spectrum, while all devices emitting in that spectrum must follow FCC rules, they also all must accept all unintentional interference, be it from microwaves, kid's radios, cordless phones, other 802.11 devices (Rokus, etc.), cars, Bluetooth, wireless cameras, high-power backbone links, etc. etc. etc. and so on and so forth, never mind natural sources. And, yes, no realtime guarantee – Anti-weakpasswords – 2018-01-24T03:01:35.510