A Study on the Effect of Phase Shifter Quantization Error on the Spectral Efficiency Using Neural Network

Abstract

Beamforming (BF) is the inevitable component of the recent communication systems, especially Millimeter wave (mmWave) systems. Thanks to the radio frequency (RF) and digital technologies, BF techniques are implemented in the both digital and analogue domains by using phase shifters (PS) networks. Adopting the digital PS, which has the finite resolution bits, leads to loss in the spectral efficiency (SE). Accordingly, in this paper, we extract the SE loss in a multi-user multiple inputs single output (MISO) system, which would be useful for practical prospective. To this end, we apply machine learning (ML) to extract the SE loss. Simulation results show that the extracted models have the desirable accuracy in the SE loss prediction.