Pursuit Learning-Based Joint Pilot Allocation and Multi-Base Station Association in a Distributed Massive MIMO Network

Pilot contamination (PC) interference causes inaccurate user equipment (UE) channel estimations and significant signal-to-interference ratio (SINR) degradations. Pilot allocation and multi-base-station (BS) association have been used to combat the PC effect and to maximize the network spectral efficiency. However, current approaches solve the pilot allocation and multi-BS association separately. This leads to a sub-optimal solution. In this paper, we propose a parallel pursuit-learning-based joint pilot allocation and multi-BS association. We first formulate the pilot allocation and multi-BS association problem as a joint optimization function. To solve the optimization function, we use a parallel optimization solver, based on a pursuit learning algorithm, that decomposes the optimization function into multiple subfunctions. Each subfunction collaborates with the other ones to obtain an optimal solution by learning from rewards obtained from probabilistically testing random solution samples. A mathematical proof to guarantee the solution convergence is provided. Simulation results show that our scheme outperforms the existing schemes by an average of 18% in terms of the network spectral efficiency.