Large dimensional analysis of cooperative multicell precoding with local individual csi

Abstract

Channel state information (CSI) is critical for the performance of multicell cooperative transmission. This paper compares two cooperative multicell precoding methods, namely, centralized precoding with global CSI and distributed precoding with local individual CSI. For both cases, the BSs have access to the data of all the users. The only difference is that in centralized precoding, the precoding is based on the global CSI shared in a central control unit, while in distributed precoding, each BS performs precoding based on its local CSI. The comparison is made under the framework of large dimension analysis based on random matrix theory. The performance gap between these two approaches can be revealed and quantified theoretically in closed form. The gap converges to a constant value in the large number of antennas regime, which we refer to as the price of local CSI. We can obtain some insights into the impact of limited CSI on the cooperative multicell transmission performance. Numerical results are provided to confirm the theoretical results and to show performance of cooperative multicell transmission with local CSI almost decrease linearly with the increased ditributedness.