Efficient Algorithm for Fairness-Aware Power Allocation in Finite-Blocklength MU-MIMO Downlink

Abstract

This paper investigates power allocation in multi-user MIMO downlink networks under the finite blocklength (FBL) regime. We establish an analytical framework for evaluating individual user throughput, and formulate the max-min fairness power allocation problem to maximize the minimum user throughput. To address the non-convexity of the problem, we propose a direct search algorithm that leverages problem decomposition, significantly improving computational efficiency compared to conventional SCA-based methods. Simulation results demonstrate that the proposed algorithm achieves the same optimal solution as the SCA approach while greatly reducing computational complexity, effectively balancing system throughput and user fairness in practical scenarios.