Graph Clustering Based User Grouping for Multi-Cell Massive MIMO Communications

Massive MIMO technology can meet the increasing demand for ultra-high transmission rate in modern wireless networks. However, interference among users sharing the same resource block (RB) reduces the transmission rate, and restricts the capabilities of massive MIMO systems. In this paper, we propose a multi-cell user grouping method based on graph clustering to enhance the system performance. We provide a beam based statistical channel model and establish a user grouping problem aimed at maximizing the sum rate. We analyze user grouping criteria that consider channel power and directions for users on the same RB. Subsequently, we employ unsupervised learning to develop a density-cut-based graph clustering algorithm (DGCA). Simulation results validate that our algorithm achieves a higher sum rate than the benchmark methods without extra computational complexity.