The Influence of Broadcast Traffic on IEEE 802.11 DCF Networks
|Ομιλητής||Rodolfo Oliveira, Universidade Nova de Lisboa, Portugal|
|Τίτλος||The Influence of Broadcast Traffic on IEEE 802.11 DCF Networks|
|Ημερομηνία||Τρίτη 13/11/2007, ώρα 16:00|
|Διεύθυνση||Ιάσονος 10, Βόλος|
web site: http://tele1.dee.fct.unl.pt/people/rado.html
The performance of the IEEE 802.11 protocol depends on multiple factors. One of them is related with the relative amounts of broadcast and unicast traffic in the total load due to the coexistence of the different transmission schemes applied to each of these types of traffic. In this talk we will present an analytical model to compute the 802.11 probability of a successful transmission of a frame and the average transmission delay assuming the presence of both unicast and broadcast traffic. Several realistic issues are addressed, as pre and post transmission backoffs and finite MAC buffers. Broadcast frame transmission has a swifter algorithm. When compared to broadcast, unicast frame transmission exhibits a more reliable scheme to avoid a frame from being discarded when it collides, but creates a trade off as it can introduce larger transmission delays. Simulation results are presented and compared to the analytical computations validating the model’s accuracy. Interesting results related with the influence of the amount of broadcast or unicast traffic on the network’s performance for non-saturation and saturation operating zones are discussed for different data transmission rates. This is particularly relevant as most of the ad-hocrouting algorithms rely heavily on broadcast. Results of the model application will be presented for a Beaconing Scheme (BS) where each network node periodically broadcasts messages (beacons) to obtain and maintain an image of the network physical topology. The model will be used to improve the BS robustness face to the communication process drawbacks. Each node uses the information given by the medium access control (MAC) layer to dynamically adapt the Beaconing Timeout value. Several simulation results evaluate the adaptive method performance using the IEEE 802.11b DCF protocol. We conclude that the BS accuracy can be improved using the proposed model. Finally, a brief state of the art on modeling the CSMA/CA based networks is given, to introduce some recent papers where other theoretical tools are suggested (e.g. game theory).