Dynamic Resource and Power Allocation strategy for ISAC-Aided 5G V2X Networks using Reinforcement learning

The increasing demand for scalability, seamless connectivity, and enhanced situational awareness in Intelligent Transportation Systems (ITS) underscores the importance of Integrated Sensing and Communication (ISAC)-aided 5G Vehicle-to-Everything (5 G V2X) networks for the future of vehicular technologies. Efficient allocation of resources and power is vital to maintaining a balance between high sensing accuracy and reliable communication in V2X networks. The traditional Sensing-Based Semi-Persistent Scheduling (SB-SPS) algorithm, while well-established for resource allocation, is predominantly communication-centric and often neglects the requirements of integrated sensing. This work proposes a dynamic resource and power allocation strategy that optimally adjusts the Modulation and Coding Scheme (MCS), Subcarrier Spacing (SCS), and Transmission Power using Deep Reinforcement Learning (DRL). Tailored to the dynamic nature of vehicular networks, our strategy effectively manages the trade-off between sensing accuracy and communication reliability. Simulation results demonstrate that our DRL-based approach significantly outperforms the traditional SB-SPS algorithm, with notable improvements in Packet Reception Ratio (PRR), Cramér-Rao Lower Bound (CRLB), and Channel Busy Ratio (CBR).