Clustered linear precoding for downlink network MIMO systems with partial CSI

Cell collaboration, also known as downlink network MIMO, has been proved to suppress the adverse effects of inter-cell interference (ICI), and as a result, to improve the performance of cell edge users. In this paper, we propose a novel clustered linear precoding scheme applicable to network MIMO systems using only partial channel state information (CSI) to enhance the quality of service and the edge user throughput. More specifically, by using a novel block diagonalization (BD) method, we first construct the precoding matrices that jointly eliminate ICI and maximize the sum-rate for a given input covariance matrix. Assuming per-BTS power constraints (PBPCs), optimal allocation schemes are further developed to optimize the sum-rate. Numerical results show that the proposed method can achieve a similar sum-rate to that of the BD with full CSI knowledge. More importantly, such a benefit can be obtained by using a small cluster size without the need of global optimization. The comparison in terms of the average signal-to-interference-plus-noise ratio (SINR) between the proposed technique using partial CSI and that with full CSI is also made. It is then revealed that there is no difference in SINR, and as a consequence, the BER performance, between the two schemes.