Multi-Cell Massive MIMO Systems With Hardware Impairments: Uplink-Downlink Duality and Downlink Precoding

• Published in: IEEE transactions on wireless communications 2017-08, Vol.16 (8), p.5115-5130
• Main Authors: Zarei, Shahram, Gerstacker, Wolfgang H., Aulin, Jocelyn, Schober, Robert
• Format: Article
• Language: English
• Subjects: Contamination; Covariance matrices; Data exchange; Downlinking; Exact solutions; Hardware; hardware impairment; Interference; Massive MIMO; Mathematical analysis; Matrix; Matrix theory; Mean square errors; MIMO; MIMO (control systems); multi-cell precoding; polynomial-expansion precoding; Polynomials; Resource management; Signal to noise ratio; Training; UL-DL duality; Uplinking; Wireless communication
• ISSN: 1536-1276; 1558-2248
• DOI: 10.1109/TWC.2017.2705709

In this paper, we propose a new framework for uplink-downlink duality in multi-cell multi-user multiple-input multiple-output systems suffering from residual hardware impairments (HWIs) at the base stations and the user terminals. We apply the proposed uplink-downlink duality framework to derive a multi-cell interference and HWI aware minimum mean square error (MCHA-MMSE) precoder which also takes channel state information estimation errors into account. We use results from random matrix theory to derive an analytical expression for the downlink power allocation for the proposed MCHA-MMSE precoder in the large system limit, which only depends on the channel's second order statistics. In contrast to the conventional MMSE precoder, the proposed MCHA-MMSE precoder takes inter-cell interference, pilot contamination, and residual HWIs into account and therefore achieves substantially higher sum rates. The proposed MCHA-MMSE precoder exploits statistical channel knowledge, but does not require data exchange between base stations via backhaul links. In order to reduce the computational complexity, the matrix inversion required for the computation of the MCHA-MMSE precoder is approximated by a matrix polynomial leading to a new polynomial-expansion MCHA-MMSE precoder. Using results from the random matrix theory, we derive closed-form expressions for the asymptotically optimal coefficients of the matrix polynomial, which only depend on the channel's second order statistics.