Interference Management in Space-Air-Ground Integrated Networks With Fully Distributed Rate-Splitting Multiple Access

Abstract

Despite the allure of ubiquitous, high-speed, and low-latency connectivity offered by Space-Air-Ground Integrated Networks (SAGINs), the co-existence of Low Earth Orbit (LEO) satellites and Unmanned Aerial Vehicles (UAVs) within the same frequency band poses significant challenges in interference management. Traditional optimization approaches, requiring seconds or even minutes for beamforming design, simply cannot keep pace with this dynamic environment. This work addresses these challenges by proposing a Fully-Distributed Rate-Splitting Multiple Access (FD-RSMA), which enables efficient cross-system interference management in SAGINs with statistical Channel State Information (CSI) at the Transmitter (CSIT). Building upon FD-RSMA, we study the precoder design of LEO satellites and UAVs along with common rate allocations of RSMA to maximize Weighted Ergodic Sum Rate (WESR). To handle channel randomness, we employ a Sample Average Approximation (SAA) approach. Furthermore, a Deep Learning (DL)-based precoder design algorithm, called GruCN, which marries the advantages of Gate Recurrent Unit (GRU) and Convolutional Neural Network (CNN), is proposed to efficiently tackle the non-convex optimization problem. Numerical results demonstrate the effectiveness and efficiency of our proposed DL-assisted FD-RSMA. Compared to conventional RSMA approaches, FD-RSMA improves up to 20% of WESR performance, while the GruCN achieves around 50% higher WESR performance and up to four orders of magnitude lower processing time than the conventional optimization approaches.