A 3D Network of Novel Nodes Sum Rate Maximization for 6G BD-RIS-Assisted Multi-Cell Transportation
TL;DR
A deep reinforcement learning–based joint orchestration algorithm (DRL-JOA) employing graph attention networks is proposed to solve the sum rate maximization problem in beyond diagonal reconfigurable intelligent surface (BD-RIS)-assisted multicell transportation networks within a space-air-ground integrated network framework.
Abstract
Sixth generation (6G) wireless networks require three-dimensional coordination of heterogeneous network components for intelligent transportation systems (ITS). This article addresses sum rate maximization in beyond diagonal reconfigurable intelligent surface (BD-RIS)-assisted multicell transportation networks within a space-air-ground integrated network framework. Unlike conventional diagonal RIS, BD-RIS enables interelement signal coupling through nondiagonal scattering matrices, providing additional beamforming degrees of freedom. The authors formulate a joint optimization problem encompassing trajectory planning for unmanned aerial vehicles, base station power allocation, and BD-RIS phase configuration. To solve this nonconvex problem, they propose a deep reinforcement learning–based joint orchestration algorithm (DRL-JOA) employing graph attention networks. Simulation results demonstrate that the proposed framework achieves 47.3% sum rate improvement over diagonal RIS and 62.8% over non-RIS baselines, while reducing intercell interference by 34.2% in dense multicell environments.
