Efficient adaptive random network coding for video content dissemination in NR-V2X networks

High-bandwidth content, particularly video and video-like data, has become increasingly common in recent vehicular networks.This encompasses everything from real-time high-resolution maps and sensor data shares to vehicle overtaking videos and infotainment streams.However, due to the high mobility of vehicles and the large volume of video data, roadside units encounter challenges in efficiently disseminating specific video content to multiple vehicles within strict deadlines.To address this issue, we propose the use of adaptive random network coding (ARNC) for multicasting video content in new radio (NR) vehicle-to-everything (V2X) networks to maximize the quality received by target vehicles.However, ARNC requires frequent feedback from vehicles via the physical sidelink feedback channel, resulting in significant overhead in NR V2X networks.To strike a balance between video quality and feedback costs, we introduce a partial feedback ARNC (PARNC)-based scheduling scheme and establish a utility function to evaluate its performance.We also employ a deep reinforcement learning algoriTheorem to optimize the PARNC design, thereby achieving a locally optimal solution.Extensive simulations validate the PARNC performance against other benchmarks.The results show that PARNC outperforms alternative schemes, particularly when the utility function emphasizes transmission performance and when feedback costs arełinebreak manageable.