Abstract: Multi-channel multi-radio (MCMR) has been widely used in wireless mesh networks for improving the network performance. Two primary problems are faced in existing channel assignment algorithms. One is that it is impossible to achieve global optimization because both the time and space complexity are high. The other problem is that those algorithms can not be scaled flexibly and, thus, cannot be applied to large networks. To solve the above problems, this paper models the channel assignment problem with particle swarm optimization model by utilizing its advantages of fast convergence and low cost. Based on the network message exchange and interference model, a particle swarm optimization based channel assignment algorithm (PSOCA) is proposed. This algorithm aims at minimizing the fitness function with constraints of radios, channels, interference and so on. Through intensive simulations, the algorithm proposed is proved feasible, both the network throughput and packet drop ratio are remarkably improved in comparison with other similar algorithms.