Sustainable energy management and deep reinforcement learning–based resource allocation in 5G networks for autonomous vehicle communications

Autonomous vehicles (AVs) and 5G wireless networks need low latency, reliability, and energy efficiency. AVs generate massive amounts of heterogeneous, real-time data, requiring efficient energy and resource allocation for large-scale vehicular communications. Vehicles with fast mobility patterns, channel conditions, and traffic intensity challenge typical optimization algorithms. In this study, a Deep Reinforcement Learning–based Energy-Aware Resource Allocation (DRL-EARA) framework for sustainable 5G vehicle networks is presented. The MDP-modeled DRL agent communicates with the network environment to autonomously learn energy-efficient power, spectrum, and user-association rules that maintain service quality. By jointly optimizing energy efficiency, spectrum utilization, and latency, the framework ensures Ultra-Reliable Low-Latency Communication (URLLC) performance, which is critical for autonomous driving. Comprehensive simulations demonstrate that the proposed DRL-EARA model outperforms heuristic and conventional approaches. The reported performance values in the studies are the average network-wide performance, averaged across multiple simulation episodes, when all autonomous vehicles and base stations are faced with congested vehicular traffic. The outcomes of simulations show that DRL-EARA can increase energy efficiency by 18.4%, latency by 32.3%, spectrum utilization by 17.6%, and throughput by 22.2%, while decreasing power consumption by 19.2% relative to state-of-the-art solutions such as ML-5GNO, DRL-ENS, and MADQN-RA. Not applicable.