Pattern Division for Massive MIMO Networks With Two-Stage Precoding

In massive multiple-input multiple-output networks with two-stage precoding, the user clusters with serious angle-spreading-range (ASR) overlapping should be divided into different patterns and scheduled in orthogonal sub-channels to achieve optimal performance. In this letter, we propose one graph theory-based pattern division (GT-PD) scheme to deal with the ASR overlapping with a limited number of sub-channels. First, we depict the ASR overlapping as an undirected weighted graph, where the weight of each edge indicates the strength of the ASR overlapping between two connected clusters. Then, we separately denote each user cluster and pattern as a vertex and a color, and transform the pattern division into a graph coloring problem with limited colors. In addition, the GT-PD scheme is developed with the help of the Dsatur algorithm. Finally, numerical results are provided to corroborate the efficiency of the proposed scheme.