Bi-Directional Training with Rank Optimization and Fairness Control

The fifth-generation (5G) wireless communication system is expected to provide increasing throughput with advanced transmitting and receiving techniques. This paper considers the problem of maximizing the downlink data rate in a multi-cell system by jointly designing transceiver filters, transmission ranks, and user associations. The base stations (BSs) are equipped with massive multiple-input multiple-output (MIMO) antennas, and can cooperatively serve the user equipments (UEs) in the network. A global optimization problem is formulated to maximize the downlink weighted sum rate (WSR). We first propose a centralized solution by assuming the existence of a central controller and global channel state information (CSI). We use the re-weighted l1 method to approximate the problem, and then solve it using the fractional programming (FP) algorithm. Since the centralized algorithm only requires local channel and signal information, we decompose it and propose a distributed algorithm based on the idea of bi-directional training (BIT). By iterative uplink and downlink training, the precoders, receivers, and transmission ranks are adaptively optimized without explicit channel estimation. An uplink power control scheme is proposed to improve the fairness among the UEs. Simulation results are presented to evaluate the performance of proposed algorithms.