Joint Optimization of Energy Efficiency in Multi-RIS-Assisted DFRC System : A Reinforcement Learning-Based Approach

In this paper, we investigate energy efficiency optimization in multi-reconfigurable intelligent surface (RIS)-assisted dual-function radar and communication (DFRC) systems. We employ the system energy efficiency as the optimization objective that utilize zero-forcing (ZF) algorithms to eliminate multi-user interference, and apply radar signal-to-interference-plus-noise ratio (SINR) and communication sum-rate as constraints. To address the low energy efficiency of traditional integrated sensing and communication (ISAC) systems, we propose a deep reinforcement learning-based joint optimization Dueling Double Deep Q-Network-Water-Filling (D3QN-WF) algorithm to resolve the non-convex optimization problem. This algorithm pursues stable optimal solutions by jointly optimizing phase shifts of multi-RIS elements, bandwidth allocation among users, and transmit beamforming configurations. Simulation results validate the effectiveness of the proposed scheme and demonstrate the energy efficiency enhancement achieved through RIS deployment in DFRC systems.