Enhancement of IEEE 802.11 DCF backoff algorithm under heavy traffic

The contention window size in carrier sense multiple access is crucial to the performance of a wireless network because it affects the network throughput and delay characteristics. The basic distributed coordination function (DCF) access mechanism of 802.11 standard describes multiple backoff window sizes. When a terminal is poised to transmit a packet, the smallest window is selected. If a collision occurs, the terminal selects an alternate backoff window. The selection is made in declining consecutive size until the largest one available at that given time is selected. This phenomenon occurs under heavy traffic conditions in which many terminals are accessing the wireless channel at the same time. Once a terminal successfully transmits its packet, the backoff window size reverts to the smallest size for future transmissions; however, this mechanism performs poorly under heavy traffic. This paper proposes that after a successful transmission, backoff window selection is dependent upon the number of retransmission; in other words, it is adaptive to the traffic condition of the network. This approach can alleviate retransmission rate by spreading the chances of simultaneous transmissions more evenly and with better flexibility relative to the basic DFC access mechanism of 802.11 MAC protocol. The MAC adopted for DSRC inter-vehicle communication applications provides enhanced delay properties,- mainly less delay due to collision-as well as throughput of the network. In our scheme a retransmission attempt of a terminal following the successful transmission of another will perform in one of two scenarios, and both are dependent upon the number of retransmissions occurring during a previous attempt Some terminals that reach a successful transmission after exceeding a certain value n will have to select a larger "minimum contention window," while others will select a smaller one. A mathematical model is developed for this proposed scheme. Our simulation experiments validate the analytical results reported in the model, which was analyzed under various assumptions. Thus, the accuracy of our model is verified.