Last Post: July 13:
I have been trying to get my head around 802.11ax and OFDMA. Yes, I've read a lot and seen nifty presentations on how cool OFDMA is. However, in the real world there are lots of 802.11ac/n devices around. How is OFDMA supposed to work in a mixed environment?
Case 1: AP has something to transmit to an ax client so the AP will contend for the channel. When the APs NAV reaches zero it will have its turn. At the beginning of the transmission the AP will have to tell how long it is going to use the channel. How does it know? It knows how much it is going to transmit, but will it also start giving out resource units to clients? The AP can't reserve the channel indefinitely or it would starve ac/n clients waiting outside of OFDMA.
Case 2: An ax client has something to transmit. Will the client wait for the channel to clear and reserve it and then only transmit its own data? Or will the AP take control of other resource units? Or will the client wait for the AP to start a OFDMA session and transmit inside that? How would the AP know when to start a session or how long it should be.
As I understand OFDMA can only take place inside OFDM transmissions, which have a definite length. After the announced time the channel must be vacated so older clients can contend for their transmissions.
I haven't found any pointers to interoperability with older standards. When 802.11n or ac came out interoperability was a big subject.
This was buried with no replies.
Please tell me if I am asking something obvious. My biggest concern is that I haven't found any discussion on pre-ax/ax interoperation.
What's wrong with me? :) Every time I try to bring this up it gets buried under an avalanche of spam.
I found my first clue! In http://bowdennetworks.co.uk/downloads/5GHz%20in%20the%20UK%20White%20Paper%20v3.0.pdf Nigel Bowden (CWNE) writes:
One of the significant challenges of 802.11ax will be co-existence with legacy networks (i.e. 802.11ac, 802.11n), due to the very significant differences in modulation technology (OFDMA). Although co-existence will be supported, it will be likely that organizations will deploy dual 5GHz radio access points to support 802.11ax and non-802.11ax devices on different radios. This will obviously place additional demand on the 5GHz spectrum availability for organizations (and their neighbours) that adopt this approach.
This half of a paragraph is all there is in the 36 page white paper. (The subject of the paper is 5GHz in the U.K. so it kind of explains.)
If the solution is to create a new WLAN on parallel radios on adjacent channels, this is big news to me. This will add a new twist to channel planning on 5GHz, but how does this work on 2.4GHz at all? ...and I had imagined we could finally make some actual use of wider channels with OFDMA.. In some cases we'd be back at 20MHz trying to fit everything in.
I finally found my answers in Perry Correll's presentation in WLPC 2018 conf in Prague:
I'm still not impressed with the /ax interop possibilities - but can definitely see the need for "overlapped radio's(?)". :-(
Perry's explanation of Resource Units (RU's) is the best I've seen. Until now, I didn't appreciate the need for all of the complexity. It really is all about "efficiency" with /ax, and not higher rates.
I was a rep to the WFA for almost seven years, so I know a lot about the aches and pains in getting devices certified - especially for devices having to meet their throughput requirements. Most people probably don't think about it, but the WFA's test engine places its own high expectations on throughput. Some products can't even count the number of packets sent out by the WFA test engine, much less make any sense of them. No wonder there are so many un-certified devices out there.
I don't see a lot of cost reductions for certifications until they change the certification rate ($) structure though. Getting a combination n/ac/ax certification, to allow for backwards compatibility, sounds prohibitively expensive.
As far as the PDF link goes - in Appendix 3 of that paper is a link to a weather radar location app for Europe and Africa. Do you happen to know of an equivalent site for North America ?
Thanks Petri, for putting this all together.
Glad you finally got through the spammer fog.
Yeah! OFDMA was borrowed from cellular networks, but the cell network is TDMA. Mixing TDMA with Wi-Fi CSMA/CA is like mixing oil and water. They simply don't mix, but you can create an emulsion. The solution is to use OFDMA inside the ax bubbles. The question is how large are the bubbles and how to control the bubble size. That's what an emulsifier is for. And then it gets complicated, also in chemistry.
I wish they had postponed OFDMA until we get the 6GHz band. Then it would have been greenfield and they could have switched to true TDMA with all of its benefits. I could have waited a few more years. There could have been a third wave of ac if the vendors were afraid of shrinking market. I mean ac wave 2 is not really here. I see mostly n traffic today. In five years ac would be mainstream and the ax products could be honed before hitting the market.
And for the weather radars I found two links:
No frequency information, no contact information. I guess you just look up the closest one, figure out their phone number and call them. I don't get it why don't they make the info as public as possible.
Same goes for Finland. I put together a web page on local weather radars and needed the info. When I finally found the right person to talk to, she was relieved: "Finally somebody asked!" She hoped I would distribute the info far and wide because they were fed up with violators.
Your page is laid out so much better than the NOAA page(s) !
Not a answer to your original question per se, but I found some information that you may find interesting on page 15 of this report.
I've gotten so sick of all the garbage posts on the forums, and the failure of CWNP to do anything about it, that I've pretty much given up on the Forums - but I saw this chart and felt it was a legitimate response to your post.
Notice that with only one /ax user, the performance is worse than /ac.