Energy efficient three-dimensional user clustering for dynamic power allocation in MIMO-NOMA systems towards 5G/6G network environments

Abstract

The evolution towards 5G and 6G wireless networks demands innovative resource allocation strategies to address unprecedented capacity and reliability requirements, particularly in dense urban environments with significant vertical user distribution. Conventional two-dimensional user clustering and power allocation schemes fail to exploit the full spatial diversity in three-dimensional scenarios such as high-rise buildings, UAV-assisted communications, and multi-floor indoor venues, leading to suboptimal performance and unfair resource distribution. This work proposes the first full three-dimensional (3D) user clustering algorithm for MIMO-NOMA systems that jointly considers user distance, height, and angle-of-arrival in a unified spatial metric—advancing beyond prior partial 3D approaches that use only elevation angle. Unlike sequential methods, we integrate 3D clustering with a Dynamic Power Allocation (DPA) scheme through a joint convex optimization framework that minimizes intra-cluster spatial dispersion while maximizing system sum rate. We formulate a joint optimization framework that minimizes intra-cluster spatial dispersion while maximizing system sum rate through convex optimization. The algorithm complexity is reduced from O(NM 3 ) to O(NM) using Neumann series approximation, enabling real-time deployment. Extensive simulations using 3GPP 3D-UMa channel models demonstrate: (1) 106.0 bps/Hz system capacity (+ 47% vs. 2D MIMO-OMA, + 15% vs. 3D MIMO-OMA), (2) 54.1 bps/Hz/W energy efficiency (+ 48% improvement), (3) 0.978 fairness index (+ 32% improvement), (4) Near-Zero Outage probability at 20 dB SNR, and (5) 99.99% reliability, meeting the URLLC requirements.