Cell clustering method based on SINR constraints for network MIMO systems

Abstract

In this thesis, we deal with a new cell clustering problem subject to SINR constraints in downlink network MIMO systems, where multiple BSs cooperate for joint processing as forming a cluster. The considered clustering problem is to find the cluster set that minimizes the maximum number of the cooperative BSs with SINR constraints. Generally, since a downlink problem is more difficult than a uplink problem, we first solve the clustering problem in uplink network MIMO systems. The way to find the cluster set is a cluster extension approach. For this purpose, we prove the SINR improvement resulting from the cluster extension when the maxSINR receiver filter is used. Based on this result, we propose the efficient cell clustering algorithm in the uplink systems, and we can find the appropriate cluster set in a way that extends a cluster. Next, we consider the sum-rate maximization problem with the transmit power constraints to find the transmit beamforming vectors in dowlink systems. This research is for the downlink cell clustering algorithm. Using the uplink-downlink SINR duality, we first derive the dual optimization problem, which is the virtual uplink sum-rate maximization problem with regard to the beamforming vectors and the virtual uplink power vector. Next, we derive the virtual uplink power as the fixed point iteration form, and we can find the virtual uplink power vector and the beamforming vectors using the proposed iterative algorithm. Finally, using the results of the proposed uplink clustering algorithm and the sum-rate maximization transmit filter, we can propose a cell clustering algorithm in the downlink network MIMO systems. Based on the the uplink-downlink SINR duality, we first investigate the feature of the cluster extension in uplink and downlink domain, and we can notice that using the proposed uplink clustering algorithm is the way that the SINR always increases in downlink domain as a cluster is extended. Next, we update the virtual up...