A Convex Quadratic SDMA Grouping Algorithm Based on Spatial Correlation

Abstract

Space Division Multiple Access (SDMA) is a promising solution to improve the spectral efficiency of future mobile radio systems. However, finding the group of MSs that maximizes system capacity using SDMA is a complex combinatorial problem, which can only be assuredly solved through an Exhaustive Search (ES). Because an ES is usually too complex, there are several sub- optimal SDMA grouping algorithms to solve this problem. Such algorithms, however, usually depend on the preceding matrices of candidate SDMA groups and are also considerably complex. In this work, an SDMA grouping algorithm is proposed for the downlink of multi-user multiple input multiple output systems. It is based on the spatial correlation and gains of the MSs' channels in the SDMA group, thus not depending on preceding and having low complexity. The proposed algorithm is formulated as a convex quadratic optimization problem and is efficiently solved by convex optimization methods. It is analyzed considering zero-forcing precoding and it is shown to almost achieve the performance of an ES for the SDMA group that maximizes the system capacity.