Re-Thinking 802.11 Rate Selection in the Face of Non-Altruistic Behavior

The use of bandwidth intensive applications in the limited 802.11 wireless spectrum has lead to increased congestion and interference in the mobile space. Sending more data in a timely manner requires a fast transmission speed. The transmission speeds in turn are governed by rate adaptation algorithms that seek to optimize the delivery probability of packets. However, optimizing for an individual node also has a significant impact on other nodes in the wireless network. Despite this impact of rate selection on the larger wireless network, individual nodes tend to be non-altruistic resulting in multi-rate WLANs, thus incurring the "rate anomaly\´\´ problem. In this paper, we posit that rate changes must be justified, i.e. goodput improvement must be commensurate with effective channel cost. We present the concept of rate zones to challenge when rate adaptation is justified with regards to the performance of the larger wireless network. Our experimental, simulation, and real-world results show that on average rate adaptation negatively impacts performance.