Towards Energy-Efficient RIS-UAV Assisted ISCC Network: An HDRL Approach

Integrated sensing, communication and computing (ISCC) is a key enabler for 6G networks, the deployment of unmanned aerial vehicles (UAVs) greatly expand ISCC coverage. However, complex channel conditions exacerbate energy and time costs, furthermore, existing frameworks struggle to balance energy-time trade-offs in dynamic environments due to limited optimization granularity. To meet these challenges, we propose a reconfigurable intelligent surface (RIS)-aided UAV-assisted ISCC network, leveraging an innovative hierarchical deep reinforcement learning (HDRL) structure. Unlike conventional single-agent DRL, HDRL decouples the original problem and adopts sub-agents to optimize UAV beamforming matrix, RIS phase shifts, UAV trajectories, and computation offloading hierarchically. Simulations show 36%-56% energy and time cost savings compared to single-agent DRL baselines. This demonstrates HDRL’s scalability in high-dimensional ISCC control, advancing real-time resource allocation for ISCC networks.