Robust UAV-Integrated Active STAR-RIS RSMA Networks: Analysis With Deep Learning Techniques

Abstract

Active simultaneously transmitting and reflecting reconfigurable intelligent surface (A-STAR-RIS) and unmanned aerial vehicle (UAV) can enhance communication channels via reduced multiplicative fading and flexible deployment. On the other hand, rate-splitting multiple access (RSMA) scheme can effectively manage interference in a multi-user setup. In this context, we study the synergistic advantages of these technologies in a robust UAV-integrated A-STAR-RIS RSMA network, deployed in remote and disaster-stricken areas. Specifically, we consider practical impediments such as co-channel interference, hardware impairments, and imperfect successive interference cancellation. We derive accurate expressions for outage probability (OP) and throughput in both delay-limited and delay-tolerant modes over Nakagami-$m$ fading channels. Further, we obtain asymptotic OP expressions to determine the achievable diversity order. We introduce a deep neural network framework that efficiently estimates the complex OP and ergodic sum rate with rapid execution. Our simulations validate these results and demonstrate the network's advantages over traditional relaying systems.