Multiuser hybrid phase-only analog/digital beamforming with genetic algorithm

Recently, hybrid analog/digital beamforming (BF) system is attracting increasing attention in the study on multiuser millimeter wave systems. In these systems, a relatively large number of antennas are connected to a less number of analog front-end chains so as to reduce both of the hardware complexity and power consumption. This paper investigates on optimizing phase-only analog BF and digital BF weights, with the objective of minimizing the total transmit power subject to signal-to-interference-plus-noise ratio (SINR) constraints. An floating-point genetic algorithm, in which the fitness function is defined as the linear scale of the objective function, is proposed to solve the optimization problem. Chaos crossover probability, chaos mutation probability and elitist strategy are introduced into the genetic algorithm in purpose of improving the operating efficiency as well as increasing the convergent rate. Simulations results under Gaussian channel environment reveal the advantages of the proposed genetic algorithm over the existed iterative algorithm. Besides, under 60GHz channel model, it is shown that the correlation on different subcarriers has profound impact on the performance of hybrid phase-only analog/digital BF.