基于深度强化学习的耦合相移STAR-RIS辅助ISAC系统联合优化方法OA

To address the problems of limited half-space coverage in conventional reconfigurable intelligent surface(RIS)-assisted integrated sensing and communication(ISAC)systems,as well as the severe coupling between communication and sensing resources and the difficulty of joint optimization in dynamic scenarios,this paper proposes a deep reinforcement learning-based joint optimization method for coupled phase-shift STAR-RIS-assisted ISAC systems.First,a STAR-RIS-assisted ISAC system model considering the mobility of sensing targets is established.Under the constraints of base station transmit power,communication quality of service,radar sensing threshold,and STAR-RIS energy conservation and coupled phase-shift,a joint optimization problem is formulated to maximize the weighted sum of communication rate and sensing rate.Furthermore,the problem is modeled as a Markov decision process,and a joint optimization strategy based on the soft actor-critic algorithm is proposed to realize the adaptive collaborative design of base station beamforming and STAR-RIS coefficients.Meanwhile,the receive filter optimization problem is transformed into a Rayleigh quotient problem.Simulation results show that the proposed method outperforms benchmark schemes in terms of both convergence speed and overall communication-sensing performance.