Global-max-volume algorithms for the user selection of the multiuser MIMO downlink systems with multiantenna users and block diagonalization

Abstract

In multiuser MIMO downlink systems, the block diagonalization (BD) precoder suffers from two major drawbacks: the dimensionality constraint on antennas arrays and the inability in handling spatially correlated users. User-selection capable of selecting only the most sum-rate-contributing users is a promising measure to alleviate this situation. One low complexity sub-optimal user selection algorithms is proposed for multiuser MIMO downlink systems with multiantenna users and blocking diagonalization (BD) precoding. It is shown to possess superior sum-rate performance than the existing Frobenius-norm-based algorithm due to the additional consideration of the inner-user channel correlation and the high-rank inter-user channel correlation. Different from the existing Frobenius-norm-based sub-optimal user selection algorithm, the matrix volume of the aggregate channel matrix of the selected users is adopted in our proposed algorithms as the selection criterion. The proposed algorithm, known as the 'global-max-volume algorithm', aims to globally search the maximum-volume aggregate channel matrix and acquires a improved sum-rate performance over the existing Frobenius-norm-based algorithms. Fast algorithms to speed up the computation process of the proposed algorithms are also introduced. © 2011 IEEE.