In this paper,an efficient resource allocation scheme based on deep reinforcement learning( DRL) for nonorthogonal multiple access( NOMA)-based vehicle platooning cellular vehicle-to-everything( C-V2X) networks was pr...In this paper,an efficient resource allocation scheme based on deep reinforcement learning( DRL) for nonorthogonal multiple access( NOMA)-based vehicle platooning cellular vehicle-to-everything( C-V2X) networks was proposed. Based on the DRL algorithm,a multi-objective optimization problem is formulated to minimize the age of information( AoI) and improve energy efficiency( EE) and transmission efficiency of cooperative awareness message( CAM). Deep Q-network( DQN) and the twin delayed deep deterministic policy gradient( TD3)algorithms are constructed to achieve the optimization. To enhance training efficiency and model generalization,the traditional sampling method of the experience replay buffer is abandoned,and a dual three-layer network based on a neural network experience replay mechanism is proposed. Based on the proposed DRL algorithm,reward function,AoI,EE,and transmission efficiency of CAM are investigated. The results demonstrate that the proposed DRL algorithm outperforms the existing algorithms.展开更多
基金supported by the National Natural Science Foundation of China (62001166)the Provincial Science and Technology Program of China (2023KFKT002)the Natural Science Foundation of Hebei Province of China(F2024201053)。
文摘In this paper,an efficient resource allocation scheme based on deep reinforcement learning( DRL) for nonorthogonal multiple access( NOMA)-based vehicle platooning cellular vehicle-to-everything( C-V2X) networks was proposed. Based on the DRL algorithm,a multi-objective optimization problem is formulated to minimize the age of information( AoI) and improve energy efficiency( EE) and transmission efficiency of cooperative awareness message( CAM). Deep Q-network( DQN) and the twin delayed deep deterministic policy gradient( TD3)algorithms are constructed to achieve the optimization. To enhance training efficiency and model generalization,the traditional sampling method of the experience replay buffer is abandoned,and a dual three-layer network based on a neural network experience replay mechanism is proposed. Based on the proposed DRL algorithm,reward function,AoI,EE,and transmission efficiency of CAM are investigated. The results demonstrate that the proposed DRL algorithm outperforms the existing algorithms.
