Fair congestion control over wireless multihop networks

Abstract

Over wireless multihop networks, the conventional transmission control protocol (TCP) cannot allocate the bandwidth fairly to flows. This is because flows with different locations competing for the same wireless channel may have very different perceptions on congestion in terms of packet delay and packet loss rate. In this study, the authors introduce a new window control algorithm and a new estimation method of congestion, which is called a delay information-based fair congestion control protocol (DFCCP). A new window control algorithm not only supports starved flows, but also allows starved flows to share the bandwidth fairly. Furthermore, since congestion occurs at multiple nodes over wireless multihop networks, the authors detect the level of congestion based on the summation of the average window sizes of multiple flows. The simulation results show that the proposed algorithm supports the starved flows while providing fairness among flows over wireless multihop networks.