Abstract: The channel assignment could affect the performance of wireless mesh network (WMN) greatly. In order to solve the problem of channel assignment, a game-based channel allocation algorithm was proposed. The algorithm modeled the nodes as the players, whose strategies were the allocations of channels, and the entire network’s throughput was regarded as the target of the utility function. The physical meaning of utility function is the successful transmission traffic given a traffic demand matrix. In order to optimize the utility function, the player could maximize the throughput of the network. Then an improved algorithm named GBCA-TP was proposed, which aimed at the insufficiency of GBCA. Simulation results based on NS2.34 show that GBCA and GBCA-TP have quite good performances on convergence, loss rate and throughput.