AoI-Aware Power Allocation for ISAC-enabled Vehicular Networks

With the rapid advancement of vehicular networks, integrated sensing and communication (ISAC) has been recognized as a key enabler for future wireless systems, offering a unified framework that supports both data transmission and environmental perception. However, ISAC requires efficient resource allocation to balance sensing and communication (S&C) performance, presenting significant challenges to dynamic vehicular networks. In this paper, we investigate power allocation problem for onboard ISAC systems of vehicles, during vehicle-to infrastructure (V2I) communication, vehicle-to-vehicle (V2V) communication, and sensing process. Our goal is to minimize V2I AoI, prioritize V2V payload transmission, and maximize sensing mutual information (MI) performance. To cope with this issue, we leverage the Twin Delayed Deep Deterministic Policy Gradient (TD3) algorithm within a reinforcement learning framework, which can efficiently learn optimal power allocation strategies. Numerical simulations demonstrate that the proposed TD3-based method achieves superior performance in terms of AoI performance, payload transmission probability, and MI rate outperforming AC and DDPG benchmark approaches.