Distributed Massive MIMO Downlink

Abstract

The feasibility of adopting distributed massive multiple-input multiple-output (D-MMIMO) for mitigating the achievable rate loss incurred by spatially-correlated fading at space-constrained user antenna arrays when served by co-located massive MIMO (C-MMIMO) counterpart is investigated. In order to explore a practically-viable design, the cumulative effects of spatial correlation at both multi-antenna access points (APs) and multi-antenna users, beamforming uncertainty caused by imperfectly estimated channel state information (CSI), and adverse effects of pilot contamination have been taken into account. The achievable user rates of the proposed D-MMIMO downlink operating in a multi-cell set-up are derived for finite/infinite antenna array regimes at the distributed APs. Thereby, the achievable user rates for C-MMIMO counterpart are deduced and compared with those of the proposed D-MMIMO downlink. Through a rigorous analysis and simulation results, we show that the distributed transmissions rendered by D-MMIMO can provide significant performance gains compared to the C-MMIMO counterpart (i) by minimizing adverse effects of spatial correlation via distinct spatial-directions/signatures rendered by distributed transmissions, (ii) by mitigating shadow fading via leveraging macro-diversity gains, and (iii) by reducing the average path-losses through effectively shortening transmission distances. Moreover, we reveal that the proposed D-MMIMO can be exploited to recover a significant portion of rate loss incurred by the lost of asymptotic channel orthogonality due to doubly-correlated channels when both APs and user nodes are equipped with space-constrained multiple-antenna arrays.