Multi-Agent Reinforcement Learning Resources Allocation Method Using Dueling Double Deep Q-Network in Vehicular Networks.