Efficient User Grouping for MU-MIMO Visible Light Communications Based on Block Diagonalization

Abstract

This paper investigates efficient user grouping methods for multi-user multi-input multi-output (MU-MIMO) visible light communication (VLC) systems. Block diagonalization (BD) precoding is considered for interference avoidance. In addition, time division multiplexing (TDM) is applied to perform user grouping when the number of users exceeds the limit of BD precoding based on the number of light emitting diode (LED) transmitters and the total number of users' photodiodes (PDs). User grouping methods are proposed based on pairwise interference considerations among users in the same group. The proposed methods can be implemented through integer linear programming (ILP), which requires less computation than exhaustive search. The numerical results on the average minimum user throughputs over random scenarios indicate that the proposed hybrid method can significantly outperform random user grouping and performs reasonably well compared to exhaustive search. Finally, this study demonstrates that, when BD precoding greatly attenuates the desired user signals, user grouping can help improve minimum user throughputs even though BD precoding can support all users as a single group.