Low Complexity Linear Detectors for Massive MIMO with Several Antenna Array Configurations: A Comparative Study

Abstract

In Multiple-Input Multiple-Output (MIMO), there is multi-user interference because of the limited number of antennas at the Base Station (BS). Deploying more antenna elements at the BS to decrease the multi-user interference is a costly solution. Therefore, changing the configuration of antenna arrays at the BS is implemented as an alternative. In this paper, the impact of BS antenna array configurations on the performance of massive MIMO detection techniques is investigated where four different antenna array configurations are implemented at the BS: Uniform Linear Array (ULA), Uniform Rectangular Planar Array (URPA), Uniform Circular Array (UCA) and Uniform Circular Planar Array (UCPA). The performance of the massive MIMO system is based on the millimeter wave (mmWave) channel that depends on the array response vector of the BS antenna array configuration. To the best of our knowledge, this is the first paper to investigate the impact of antenna array configurations at the BS on the performance and the computational complexity of massive MIMO detection techniques. Numerical results show that the implementation of URPA at the BS with MMSE detection algorithm can achieve the best performance of BER=10−4 at SNR = 9dB. The deployment of the UCA at the BS with the successive over-relaxation (SOR), the Gauss-Seidel (GS), and the conjugate-gradient (CG) detection methods provides a performance of BER=10−4 at SNR = 12dB when the number of iterations equals eight. In the case of using the UCA at BS with the Richardson (RI) detection method, the number of iterations is required to be increased to achieve the same performance. To achieve BER=10−4, the detector based on the CG method has the lowest computational complexity while the RI method has the highest complexity.