Optimal Time Partitioning for Sensing-Communication Trade-Off in V2V ISAC Systems

Integrated sensing and communication (ISAC) in vehicle-to-vehicle networks faces challenges in dynamic resource allocation due to time-varying channels and sensingcommunication trade-offs. This letter proposes GT-MADDPG, a game-theoretic reinforcement learning algorithm that optimizes time-resource partitioning via Nash equilibrium analysis and multi-agent coordination. Key innovations include a hierarchical framework for low-latency decisions and contiguous slot allocation to mitigate interference. Simulations under 3GPP scenarios demonstrate significant improvements in communication reliability and sensing precision compared to state-of-the-art methods, with robustness across varying traffic densities.