Multi-Relay UAV-RIS Digital Twin-Based Framework for User Fairness and Load Balancing Using Multi-Tasking DRL

As revolutionary enablers capable of dynamically influencing communication link quality, Reflecting Intelligent Surfaces (RIS), particularly when mounted on Unmanned Aerial Vehicles (UAV) have emerged as reliable, cost-effective, and flexible tools for enhancing wireless connectivity across a range of real-world scenarios. Similarly, Digital Twin (DT) technology is increasingly recognized as an enabler for future wireless networks. To mitigate the challenges posed by building obstructions or natural disasters that disrupt radio propagation, this work incorporates multiple UAV-RIS to enhance connectivity and improve link quality throughout the network. This paper proposes a Multi-Relay UAV-RIS-assisted DT (MR-UAV-RIS DT) framework, focused on intelligent network rate management to ensure User Fairness (UF) and Load Balancing (LB) in small cell (SC)-based wireless networks. In the proposed system, dual-connectivity user equipments (UEs) are simultaneously connected to both the macrocell (MC) via relay UAV-RIS, and to Small cell-Base stations (SC-BSs), all operating in the mmWave frequency band to deliver high-speed connectivity. To meet UF and LB goals, an optimization problem is formulated to jointly control the active beamforming for both SC-BSs and the MC, as well as the passive beamforming and 3D positions of the UAV-RISs associated with SC-BSs and relay UAV-RIS linked to the MC. Due to the complexity and high dimensionality of the problem, a DT-based multi-task deep reinforcement learning model (DT-MTDRL) is proposed, built upon the Deep Deterministic Policy Gradient (DDPG) algorithm. Simulation results confirm that the proposed model not only ensures a fair rate distribution among users and SCs but also delivers superior performance compared to existing benchmark schemes, while maintaining robust and adaptive connectivity in dynamic and challenging environments.