Efficient Algorithms for Constant-Modulus Analog Beamforming

The use of a large-scale antenna array (LSAA) has become an important characteristic multi-antenna communication systems to achieve beamforming gains such as in designing millimeter-wave (mmWave) combat severe propagation losses. In applications, each element be driven by radio frequency (RF) chain for the implementation fully-digital beamformers, significantly increasing hardware cost, complexity, and power consumption. Therefore, constant-modulus analog (CMAB) becomes viable solution. this paper, we consider scaled (SAB) or architecture design system parameters solving two variants beampattern matching problem. first case, both magnitude phase are matched given desired whereas second only is matched. Both problems cast variant least-squares (CLS) We provide efficient algorithms based on alternating majorization-minimization (AMM) framework that combines minimization MM frameworks conventional-cyclic coordinate descent (C-CCD) solve problem case. also propose new modified-CCD (M-CCD) approach. For all developed prove convergence Karush-Kuhn-Tucker (KKT) point (or stationary point). Numerical results demonstrate proposed converge faster than state-of-the-art solutions. Among algorithms, M-CCD-based have when evaluated terms number iterations AMM-based offer lower complexity.