[Cerowrt-devel] [Make-wifi-fast] Will full-duplex be possible on 802.11 wired air?

Dave Taht dave.taht at gmail.com
Sat Mar 12 14:35:05 EST 2016



Someone, georgia tech I thought, had actually gone and built a chip to
do it. Can't find the link.

I have no doubt that we can construct ever more efficient devices,
what I have grave doubts about is given the billions of existing
devices pee-ing on the spectrum that there will be no way to actually
make an impact.

For christmas, I gave an  unhappy chromecast user the wired dongle.
Stuffing stockings full of that sort of stuff, repacing wifi for wired
wherever possible,
seems like the sanest option. I've been thinking of begging the
landlords around me to let me wire up their buildings and tenants for
ethernet, from 6pm onwards wifi is getting towards unusable.


In terms of complexity in serving multiple stations - as opposed to
full duplex...

The number of DSPs in the 802.11ac mac is well over 400, and MU-mimo,
as defined in the standard - really difficult to implement and use
effectively. The first MU-mimo capable chips (wave 2) are shipping
now, but

Not for the last time, I wish we'd got UWB off the ground. Only needed
236 notches cut out of the spectrum a decade ago to make the existing
spectrum holders happy.

Dave Täht
Let's go make home routers and wifi faster! With better software!

On Sat, Mar 12, 2016 at 11:20 AM, Erkki Lintunen <erkki.lintunen at iki.fi> wrote:
> Last week or so I had a short peek on the lists and read something alike
> 'half-duplex makes wifi feel slow [in addition to the point in
> discussion and when compared to wired]'. Now couldn't find the message
> to reply to the sited one but going on with this one.
> Half-duplex brought me to a news item national IT press was buzzing
> about few months ago: Taneli Riihonen's doctoral dissertation "Design
> and Analysis of Duplexing Modes and Forwarding Protocols for OFDM(A)
> Relay Links" [1].
> 1 http://taneli.riihonen.fi/ "Taneli Riihonen - List of Publications"
> The fuss news media made out of the research was, that the methods
> Riihonen found will solely speed up 5G cellular networks. Amusingly I
> saw a picture a news outlet run out, where Taneli Riihonen was standing
> and in the background a slide showed text "OFDM". I thought what the
> heck, this isn't just about 5G, who cares about 5G. Then forgot the news
> and now got flashback.
> This might be old news/tidbits for the list, but one can't never be
> sure, so passing on as in the back of my head are Dave Täht's words that
> a lot of new research and reading old was made for accomplishment in
> current state of bufferbloat in wired connections.
> Getting to the on-the-air-timings spoken about below, the methods Taneli
> Riihonen found in his dissertation used statistical methods for time
> slotting the "shout out" and "listen to" for duplexing on the air.
> Intuitively this brings me to expect that the dissertation might give
> ideas how to measure timings for tx and rx, and, in the best case
> readily usable, tools how they have proved their theory in the theses.
> Sorry I haven't even glimpsed the 300 pages thesis. I'm only basing on
> slides linked to in the same paragraph as the thesis on the above
> web-page. Unfortunately the slides are in finnish.
> -Erkki
> * Aaron Wood <woody77 at gmail.com> [2016-02-09 06:17:57 +0800]:
>> I've often wanted the same thing:  What's the time-length of given packets
>> (using various transmission rates), and the inter-packet delays, etc.  What
>> _is_ 100% channel utilization, in terms of packets per second of a given
>> size/rate?
>> From a pcap file full of radio-tap-level packets, can the channel usage be
>> computed?  (none of the tools I looked at a few years ago could give me a
>> channel usage indication from an analysis of actual packets (with rates and
>> timestamps)).
>> -Aaron
>> On Tue, Feb 9, 2016 at 4:02 AM, Dave Täht <dave at taht.net> wrote:
>> > Much more readable than the spec!
>> >
>> > http://chimera.labs.oreilly.com/books/1234000001739/ch03.html
>> >
>> > I still keep hoping for a comprehensive list (or a tool) for timings for
>> > every possible operation across all the 802.11 standards. Trying to
>> > figure out how long things take "on the wire" makes my brain spin.
>> >
>> > _______________________________________________
>> > Cerowrt-devel mailing list
>> > Cerowrt-devel at lists.bufferbloat.net
>> > https://lists.bufferbloat.net/listinfo/cerowrt-devel
>> >
>> _______________________________________________
>> Make-wifi-fast mailing list
>> Make-wifi-fast at lists.bufferbloat.net
>> https://lists.bufferbloat.net/listinfo/make-wifi-fast
> _______________________________________________
> Cerowrt-devel mailing list
> Cerowrt-devel at lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/cerowrt-devel

More information about the Cerowrt-devel mailing list