Performance analysis of low complexity multi-user MIMO scheduling schemes for massive MIMO system

Massive MIMO is a promising technology for the 5G mobile communications system and contributes to the enhancement of multi-user MIMO transmission. However, the user scheduling of multi-user MIMO transmission requires a high computational complexity because the matrix operation is performed to estimate SINR, which is used as a user selection criterion. In this paper, we investigate low complexity multi-user MIMO scheduling schemes that estimate the SINR without the matrix operation. The schemes are the polynomial approximation scheme, which approximately estimates the SINR by channel correlation between users, and the single-user based scheme, which uses the SINR of single-user transmission instead of the SINR of multi-user MIMO transmission. Simulation results show that the complexity of the polynomial approximation scheme is reduced to a few tenths of that of the matrix operation based scheme without throughput degradation. It is also shown that the complexity of the single-user based scheme is reduced to a few thousandths of that of the matrix operation based scheme with little throughput degradation as the number of spatially multiplexed users increases.