On the Performance Characteristics of WLANS:Revisited

Anand Balachadran, Geoffrey M. Voelker, Paramvir Bahl, P. Venkal Rangan

I love this paper.It basically takes all the findings that simulation-guys have been writing tons of papers about and says: this is bullshit, it has nothing to do with real world.

Now, what are the most interesting contributions... first it says that TCP and DCF make good friends, specially when the wireless terminals are clients downloading from the net. In this situation, the download throughput remains constant, no matter how many STAs enter into the network (up to 100).

HOWEVER, when the wlan reaches the saturation point, jitter and delays rise suddenly. This is not good for MM :-)

And this is funny: the automatic rate-adaption in 802.11 actually worsens the performance of .11 . It goes like this: when congestion occurs and a STA loses two consecutive packets due to collisions, the station changes to a lower rate effectively increasing the congestion.

And my conclusions: The DCF-TCP is the perfect team for web browsing, and file downloading from the net. However, as soon as new applications appear such as MM streaming, VoIP, P2P,... dcf can be a disaster. Would .11e be powerful enough to alleviate this situation?