Congestion Control In Wireless Multihop Networks

The proliferation of wireless multihop networks has attracted much people's attention on resource allocation problem in such kind of networks. In the whole protocol stack, transportation layer is responsible for end-to-end rate control, which is essentially resource allocation. Efficiency and fairness are critical issues in resource allocation. Conventional transportation layer protocols, such as TCP, are not aware of congestion in the neighborhoods of rely nodes, and tend to adopt over-conservative or aggressive strategy while competing for the shared resource, thus they may suffer efficiency and fairness problems when applied to wireless multihop networks. On the other hand, most of protocols designed for this environment do not consider the unique characteristics of MAC and routing protocols, which results in compatibility problems and performance degradation.Wireless multihop networks are classified into MANET and Mesh, according to their difference in node mobility. This thesis concentrates on the resource allocation problem in wireless mesh networks. Following the Proportional Fairness framework, this thesis analytically models the congestion control problem, and derives a distributed solution to the problem. Based on the distributed solution, a novel and full functional transportation layer algorithm and protocol framework: WAXP (Wireless Ad-hoc Xport Protocol) is proposed to improve the unfairness of existing protocol combinations. Following the idea of cross-layer design, through neighborhood explicit signaling and robust feedback, WAXP sender, rely and receiver cooperatively execute congestion control, which efficiently allocates channel resource and alleviates contention severity on MAC layer. Moreover, WAXP carries all control information in packet headers, thus the advantage of requiring no per-flow state on relay nodes significantly improves the scalability of the protocol.Numerous simulation shows that, without loosing high resource utilization, WAXP adapts to characteristics of lower layers, thus it effectively reduces the probability of MAC collision and route break. On system-wise, WAXP fairly allocates the shared channel resource among all competitors, and enjoys both low packet loss rate and end-to-end delay. Lots of practical concerns embedded in WAXP's design also enhance the viability of the protocol, which makes it promising for fair resource allocation and ready for implementation.