Interference Mitigation Using 3D Building Blockage for Space-Air-Ground Integrated Networks

This study re-examines non-line-of-sight (NLoS) channels, proposing a novel approach that leverages three-dimensional (3D) building blockage to mitigate interference signals, thereby enhancing space-air-ground integrated network performance. Unmanned aerial vehicles (UAVs) benefit from mobility, allowing the adaptable formation of line-of-sight (LoS) and NLoS channels by considering building blockage. Accordingly, a mathematical model is presented to determine whether the interference channels from the UAV to satellite nodes (SNs) are blocked by buildings. We then formulate a joint optimization problem involving scheduling, transmit power, and trajectory to maximize the minimum throughput of ground nodes (GNs), ensuring the minimum required throughput for the SNs. We employ various optimization techniques to solve the formulated nonconvex problem and find that this approach requires significant computational complexity and its performance is sensitive to initialization. To address these challenges, we propose an integrated approach, combining an unsupervised learning-based deep learning (DL) framework for determining initial values with subsequent refinement through optimization. Simulation results provide useful insights into employing building blockage for interference mitigation. Notably, the UAV avoids direct access to GNs in areas that can form the LoS interference channels to the SNs and stays in NLoS areas to serve all GNs, preventing severe interference with the SNs. The integrated approach exhibits superior performance with a much faster convergence time compared to the optimization approach by improving the strategy inferred by our DL model with optimization methods.