Optimal Resource Allocation for Massive MIMO over Spatially Correlated Fading Channels

Abstract

This paper investigates optimal resource allocation scheme for a typical uplink single-cell massive MIMO system over spatially correlated fading channels. To reduce the pilot contamination effect, we first propose the orthogonal user grouping strategy to partition the terminals into serval groups and assume the users in the same group reuse an identical orthogonal pilot sequence. Employing this strategy, we derive the closed-form expression of the achievable rate for the maximum-ratio combining (MRC) receiver. Then the expression is used to pursue a detailed analysis of the optimal resource allocation scheme so that the system energy efficiency can be maximized. It is found that the optimal training duration is identical to the number of user groups and, hence, can be selected dynamically according to the second order statistical information about the user channels. Numerical results are presented to verify our analysis, and show that the optimal resource allocation can provide significant performance gains compared to the traditional orthogonal training scheme in the high SNR regime.