Skip to Content.
Sympa Menu

wg-pic - Re: [wg-pic] VoIP over WiFi story

Subject: Presence and IntComm WG

List archive

Re: [wg-pic] VoIP over WiFi story


Chronological Thread 
  • From: "ATD Munro, Electrical & Electronic Engineering" <>
  • To:
  • Subject: Re: [wg-pic] VoIP over WiFi story
  • Date: Fri, 11 Jun 2004 10:05:53 +0100

Hi All,

I'm glad that this author wasn't in attendance at last year's fall
member meeting.

http://story.news.yahoo.com/news?tmpl=story&cid=1738&e=1&u=/zd/20040609/
tc_zd/129119

Ouch! As counterpoint, I have been in Greece and made a number of
calls from X-Lite on a WiFi PocketPC, through the Greek commodity
network, across the pond, to a PSTN gateway in the States. They were
flawless.

I also had some pure IP calls with Dennis that were productive, but
not exactly flawless (we were mysteriously dropped several times).

Sometimes it works, sometimes it doesn't... I had several good calls at the Fall meeting last year and some that failed. I'm not surprised that the reporter cited above had a bad deal at the VON conference; nor that Ben got lucky on occasion.

Have I bored you all to death about this before? If so apologies and skip the next paragraph...

The 802.11 MAC has a mode (PCF) that is intended to help with delay-bounded traffic, but it is not available in most equipment on the market. Even if it was, (maybe the VON reporter was put in a channel with a PCF-enabled AP), it won't work unless all devices on a channel are using it. I think most chipsets in common products implement PCF and will interoperate with others but there are no hooks (as far as I know - please correct me) back into the driver in the o/s and the applications to configure all the parameters to schedule the traffic. The 802.11e extensions (EDCF and HCF) are supposed to fix this, (and are also implemented in recent chipsets for 11b, 11g, and 11a), but, IMO, 11e is only half a solution and probably as bad as native 802.11. As far as I can tell, it does the job on paper (see recent conferences ad nauseam) but, like PCF, there's no means of getting to it. Then there are no QoS policies for providing a framework for managing VoIP in competition with other traffic (PCF, EDCF and HCF will add sufficient delay to make other services unusable if there is any voice). And all this before you get hammered by interference from nearby co- and adjacent-channel systems.

While anyone can get lucky from time to time (as Ben and I did), the moment there is any competing traffic or interference, VoIP is likely to stop. You would really need to dedicate a channel for VoIP but this implies a level of signalling and management that we don't have yet (or may never have). There is not enough spectrum in either the 2.4 or 5 GHz bands allocated for 802.11 to support a significant amount of voice anyway (i.e. 100s of users), and you would need a very strong lobby from operators to get more, especially when they are all trying to make money on 3G, 802.16, 802.20, etc.

I think expectations ought to be set very low for real-time services using WiFi without significant effort being spent on mechanisms that manage the resource better - not just access to a single channel but across all the channels as well.

Regards,

Alistair

--
Prof. Alistair Munro, Toshiba Professor of Communications Networks
Dept. Electrical & Electronic Engineering, Bristol University
Merchant Venturers Building, Woodland Road, Clifton,
Bristol BS8 1UB
E-mail:

Tel.: +44-7974-922442



Archive powered by MHonArc 2.6.16.

Top of Page