On service time estimation in 802.11 WLANs with heterogeneous traffic sources

Abstract

This work presents a simple method to estimate IEEE 802.11 distributed coordination function (DCF) service time. This is the total amount of time needed to transmit a given frame, which is defined as the duration from the instant a node starts the transmission, until the instant when the transmission task effectively finishes. We are motivated by the fact that IEEE 802.11 DCF does not provide time-bounded transmissions. Thus, it is important to have an estimate of the service time because most of the times a physical connection between two nodes exists, but it is misinterpreted by the upper layers due to the long waiting time to obtain a response from a node. In this case, the estimate of the service time can be used by the upper layer protocols to solve some problems caused by IEEE 802.11, such as routing failures due to timeouts or even TCP connection failures. The congestion time in IEEE 802.11 before each (re)transmission, which is due to the random access mechanism employed in the CSMA/CA, makes it more difficult for service time to be estimated. Our approach needs only local information available at each node and considers a realistic scenario where the traffic sources may have different data generation rates (heterogeneous traffic). The applied methodology is described and we present simulation results to assess its accuracy.