Penalty Function Method for Peer Selection over Wireless Mesh Network

Appropriate peer selection from the discovered file holders plays a vital role for peer-to-peer (P2P) file sharing over wireless mesh networks (WMNs). When peers choose their own utility-maximizing strategies for coalition and peer formation, the solution is always sub-optimal. Peer formation, based on only application layer information, also results in inefficient use of network bandwidth. When multiple recipient-peers try to access the same file from same source-peer simultaneously, contention may occur on the shared wireless channel. On the discovery of multiple source-peers, corresponding recipient-peer may choose optimal source-peer in favor of increased network throughput. We formulate the joint peer selection and utility maximization problem as a mixed integer nonlinear programming (MINLP) framework. We also propose penalty-based heuristic genetic algorithm (GA) to solve the MINLP. The results show that our favorable-peer selection strategy results in higher aggregate throughput by selecting optimum source-peers with better load distribution and minimum interference.