Energy-efficient resource allocation in RIS-UAV MEC networks with enhanced compute capacity

Abstract

Reconfigurable intelligent surface (RIS)-assisted unmanned aerial vehicle (UAV) mobile edge computing (MEC) networks provide flexible coverage extension and on-demand computation at the wireless edge. However, existing resource allocation (RA) approaches for RIS-UAV MEC systems focusing only on signal-to-noise ratio (SNR) interference models, and do not effectively utilize MEC computational capacity, resulting in suboptimal energy efficiency (EE) and limited compute performance. To address these limitations, this work introduces a random forest-Grasshopper optimization algorithm (RF-GOA) framework designed for energy-efficient RA with enhanced compute capacity in RIS-UAV MEC networks. The RF module predicts traffic load and large-scale channel state information (CSI) to guide power distribution, sub-band allocation, MEC CPU-frequency scaling, and RIS phase-configuration modes. The system is formulated as an EE-maximization problem that couples transmit power, bandwidth, RIS configuration, UAV mobility, and MEC compute cycles under delay, power-budget, mobility, and LoS/NLoS constraints. A GOA-based metaheuristic performs global exploration over mixed discrete-continuous variables, while convex refinements ensure feasible and convergent solutions. Simulation results show that the proposed RF-GOA framework achieves significant performance compared with conventional benchmark schemes in terms of energy efficiency, compute utilization, and scalability. These results confirm that the proposed framework provides an effective solution for intelligent and adaptive resource allocation in RIS-UAV-MEC systems.