Finite-Resolution Digital Beamforming for Multi-User Millimeter-Wave Networks

Recent studies have shown that low-resolution analog-to-digital-converters and digital-to-analog-converters (ADCs and DACs) can make fully-digital beamforming more power efficient than its analog or hybrid beamforming counterpart over wide-band millimeter-wave (mmWave) channels. Inspired by this, we propose a computationally efficient fully-digital beamformer relying on low-resolution ADCs/DACs for multi-user mmWave communication networks. Both a generalized (unstructured) beamformer (GB) and a structured zero-forcing beamformer (ZFB) are proposed. For maintaining fairness among all users in the network, specifically tailored objective functions are considered under sum-power constraints, namely that of maximizing the geometric mean (GM) of users' rate and their max-min rate. These computationally challenging beamforming design problems are tackled by developing computationally efficient steep ascent algorithms, which have the radical benefit of relying on a closed-form solution at each iteration. Moreover, to facilitate the employment of low-cost amplifiers at each antenna, the GB design problem subject to the equal-gain transmission constraint is considered, which assigns equal transmit power to each transmit antenna. The proposed algorithms promise a user-rate distribution having a reduced deviation among the user-rates, i.e., improved rate-fairness. Our extensive simulation results show an approximately upto 45% reduction for the GM-rate of a 2-bit ADC (4-bin quantization) compared to the $\infty$-resolution ADC.