[Asterisk-Dev] Re: 802.11b a contraindication?

[John Todd]

At 4:55 PM +0100 on 4/13/04, Chris Wilson wrote:
>Hi James,
>
>>  Chris> What sort of statistics do you get over long periods? High
>>  Chris> packet loss?  (>1%) High jitter (>1ms)? Have you tried IAX2
>>  Chris> with jitter buffer over the link, instead of SIP, and if so did
>>  Chris> it help?
>>
>>  It was too unuseable to do anything at all.  But I only have the
>>  ata186 and the laptop on that lan (at least that can do voip) so
>>  testing iax2 is not possible for now.
>
>What about ping statistics, did you manage to capture those?
>
>>  I got the studdering effect.  At least from the notebook to the ata;
>>  I cannot say how it was in the other dir...
>
>That's interesting, as Cisco's 7940/60 phones are very good at coping with
>packet loss and jitter, I'd assumed that ATA-186s were too. What software
>were you running on the laptop? Would be interesting to know if
>ATA->laptop was good but laptop->ATA was poor, or they were both poor.
>
>>  It was as if either every other packet or so were dropped or as if
>>  every packet were truncated. 
>
>Well, they couldn't be truncated, but they might well be lost or arriving
>with a lot of jitter.
>
>What AP are you using, and what card in the laptop?
>
>>  I am using 100baseT now for the laptop's link and can do near-toll
>>  quality calls (using iax2/ilbc over the v90) at night and almost
>>  as good during the day.  (I've also tried G732.1 from the ata
>>  tunneled over iax between my *s and over sip to a remote harware
>>  gateway; that also works well.)
>
>Nice to know, but not surprising.
>
>I would really appreciate some more discussion on this list of how to get
>good audio performance over wireless links. I've even considered writing a
>special proxy to send all UDP packets N times to eliminate packet loss,
>and buffer them to eliminate jitter.
>
>Cheers, Chris.
>--
>_  __ __     _
>  / __/ / ,__(_)_  | Chris Wilson -- UNIX Firewall Lead Developer |
>/ (_  ,\/ _/ /_ \ | NetServers.co.uk http://www.netservers.co.uk |
>\__/_/_/_//_/___/ | 21 Signet Court, Cambridge, UK. 01223 576516 |
>

As a point of reference, I've had my 7960 working quite well over 
802.11 for weeks at a time.  My home network traffic is very light, 
so I've not tested in a "production" environment with many SIP 
endpoints on the same AP (well, I have, but it wasn't "controlled" so 
wasn't very impressive) but I know from personal experience that it 
does work quite well with a single session to the point where I 
actually forgot that I was using 802.11 and completely confused 
myself when I unplugged what I _thought_ was the uplink cable on the 
hub and kept on hearing the hold music on the remote phone.

I have experience with more dense situations, and there are practical 
limits.  I'm still under NDA for some of those, but the hardware 
wasn't anything you can't buy off the shelf.  I've seen >10 sessions 
simultaneously without things breaking down.  Mind you, I was using a 
bridged 802.11 network where a bridge unit was talking to a single AP 
(one channel, one pair of tx/rx stations) so this perhaps will start 
to break down as multiple talkers appear on the same frequency.

I agree that a more specific understanding of interactions between 
802.11[b,a,g] and SIP RTP sessions would be worthwhile if it could be 
found or generated and posted to this list.  Additionally, what would 
be more worthwhile would be a similar IAX2 study.   I'll put this on 
my long and un-cheery list of "things to be tested", right beside 
satellite latency effects on IAX2...

JT