EVD-Based Detection for Multi-Cell Massive MIMO Network

This paper focuses on detection for multi-cell Massive MIMO network. As the column vectors are independent and identically distributed with each other, channel fast fading coefficient matrix is the eigenvector matrix for the covariance of received signal when the number of antennas at base stations (BSs) tends to infinity. Thus, we can get a set of normalized base vectors by eigenvalue decomposition of the covariance matrix for received signal. Based on this close relationship, an EVD (Eigenvalue Decomposition) based detection algorithm, which could solve the pilot contamination problem completely when the number of antennas at BS tends to infinity, is proposed. Simulation results show that the bit error rate (BER) could be reduced by two orders of magnitude when signal-to-noise ratio (SNR) is 40dB, with EVD-based detection algorithm, comparing with the traditional maximum-ratio combining (MRC) detection.