Transceiver design for massive MIMO systems: approaches based on matrix completion, beam selection and random pilots

Abstract

Massive multiple-input-multiple-output (MIMO) technology is a key to achieve the promised capacity gains in 5G systems. Massive MIMO systems consist in the simultaneous deployment of a large number of antennas in a base station (BS) to serve many user equipments (UEs). For achieving the full potential capacity of MIMO, accurate knowledge of the channel state information (CSI) at the BS is essential. In frequency division duplexing (FDD) systems, the problem is that the channel feedback load grows linearly with the number of antennas. Then, for practical feedback channels, the overhead to obtain full CSI becomes prohibitively large due to the massive number of antenna elements. Thus, relying on CSI to design the downlink transmission emerges as a bottleneck in FDD-based massive MIMO systems. In this context, first, we develop a framework that uses the matrix completion (MC) technique to reduce the uplink feedback channel overhead exploiting the low-rank channel structure of the channel matrix. The proposed framework is evaluated in two application scenarios: wireless backhauling communications and a multi-user (MU) scenario. Furthermore, we show that the decrease of the reconstruction error is related to the number of BS antennas, and discuss the performance in terms of bit error rate (BER) and goodput. When the number of BS antennas is moderate, an interference problem among UEs allocated for the same time-frequency resource has to be effectively handled. Transmit beamforming is one of the techniques to deal with MU interference. Assuming knowledge of the beamspace channel in a sparse massive MIMO system, we propose a precoder design based on the maximum ratio transmission (MRT) that consists of selecting and optimizing the power of the beams steered to the UEs in order to maximize the signal-to-interference-plus-noise ratio (SINR) at the UE. Considering two different sparse channel models based on independent identically distributed (i.i.d.) and geometric-stochastic beam domain representations, we propose low-complexity heuristics to beam selection and rate adaptation, and discuss the optimal solution for this problem. Simulation results show that our optimal solution can achieve a better performance than the zero-forcing beamforming (ZFBF) scheme. Besides, compared to the linear MRT precoder, the proposed low-complexity heuristics improve the performance of the system in a scenario with channel sparsity, which may be the case in millimeter-wave MIMO channels. Finally, under a multi-cell perspective, we propose a space-time pilot transmission technique based on the space-time random pilot selection (ST-RPS) that mitigates or eliminates the effect of pilot contamination in massive MIMO system. The space-time pilot transmission method uses Bernoulli distribution to decide the transmission. Despite the conceptual simplicity of the ST-RPS scheme, simulation results show that it improves the channel estimation accuracy