Energy Efficient Power Allocation for Cell-Free mmWave Massive MIMO With Hybrid Precoder

Abstract

This letter investigates the downlink of a cell-free millimeter wave (mmWave) massive multiple-input multiple-output (mMIMO) system, where many access points (APs) cooperatively serve a user. Although the intensive deployment of APs can dramatically improve the system capacity, it also increases the network energy consumption substantially. To track the non-concave global energy-efficiency (GEE) optimization problem, we decompose it into hybrid precoder design and power allocation design. A novel dynamic subarray with quantized phase shifters (DS-QPS) hybrid precoder is introduced, where each radio frequency (RF) chain only connects to a disjointed subset of antennas. The optimization problem of the number of RF chains is formulated as an eigenvalue maximization problem considering a realistic power consumption model. For power allocation, a new centralized framework is exploited to solve a sequence of simpler power allocation subproblems while still aiming at the GEE maximization by merging with fractional programming, non-cooperative game theory, and gradient-assisted binary search (GABS) algorithm. Simulations show that the joint design is more energy-efficient than the baselines.