Successive Resource Allocation in Multi-User ISAC System through Deep Reinforcement Learning

The rapid convergence of wireless infrastructure and vertical applications has brought the growing needs for integrated sensing and communications. Due to competing purposes and limited radio resources, an effectively designed integrated sensing and communication (ISAC) system has to precisely adjust its resource allocation to communication and sensing. To maximize the value of service (VoS) for ISAC operation with varying concurrent demands and resource conditions, a successive resource allocation scheme for multi-user ISAC systems is proposed. Specifically, the bandwidth and power allocation are formulated as a mixed integer optimization problem by considering the varying user requirements and resource availability. To solve this problem, a deep-reinforcement learning (DRL) based adaptive resource allocation algorithm is utilized for successive ISAC operational gain maximization. Simulation results demonstrate the adaptiveness and effectiveness of the proposed resource allocation scheme under dynamic scenarios.