A partial CSI based Affinity propagation Dynamic clustering algorithm

The precondition of the whole known channel state information (CSI) is impractical in realistic system. To deal with this problem, a partial CSI based Affinity propagation Dynamic clustering algorithm was proposed in this paper. In this algorithm, degree of wiliness to cooperate (DWC) was used to design the input matrix and a cooperation fact was introduced to adjust the number of clusters in clustering results. Assuming every BS has the partial CSI information of the whole system, during each iteration round, Responsibility and availability are propagated to accumulate the coordination incidences among BSs. After that, it constructs several clusters of dynamic size. Simulation results show that the capability of this scheme is almost the same as the whole CSI based benefit-tree clustering algorithm. And the convergence iteration rounds increase slowly as the system size enlarging. This exhibits high convergence speed, which is well suited for implementation in realistic system.