Study of WiFi behavior at a large meeting
I was at the IETF last week so blogging was light. One of the features of an IETF is that pretty much every one of the 1,500 attendees is using WiFi. So an IETF is a pretty good test of how WiFi performs under heavy use.
This study by Columbia University provides a good explanation of why the performance of these networks is often somewhat patchy.
Basically the problem is that the drivers have a somewhat naive approach to roaming and fail to take account of the cost incurred when switching between access points. The driver will hop to a marginally stronger access point even though this causes the user's open conections to drop.
The good news is that it is all fixable. But will it get fixed?
Comments
URL is broken but I was able to find the PPT that probably made the PDF here:
http://www1.cs.columbia.edu/~andreaf/downloads/presentation/IETF-Measurement.ppt
Good presentation and fascinating about the large WIFI study.
Posted by: Rob Bergin | January 3, 2007 12:17 PM