SFDIC: Spatial Features Distributed Interference Coordination for Massive MIMO Systems

In 5G massive multiple input multiple output (MIMO) system, the main challenges to mitigate inter-cell interference (ICI) are overhead of information exchange and computational complexity. In this paper, we propose an interference approximation method based on spatial features, which can cover the major channel information by low overhead. And based on this method, a novel distributed low-complexity interference coordination algorithm called SFDIC is proposed, which is based on the idea of leader-follower game to avoid strong ICI. The experimental results show that the proposed interference approximation method is strongly consistent with the traditional channel matrix based interference calculation method on the trend, whose correlation coefficient is 0.9098 and Kullback-Leibler (KL) divergence is close to 0. In low and medium-speed scenarios, the SFDIC increases system and edge throughput by more than 20% and 107% than joint space division multiplexing (JSDM) respectively. And these scenarios reduce the sharing overhead by more than 50% simultaneously. In addition, the new channel predicted module based on Koopman operator is incorporated to improve practical feasibility and system performance loss causing by delay for the first time.