An energy-efficient resource allocation strategy for vehicular networks

Vehicular networks have drawn many researchers' attention for the reason that vehicular networks can change the operation of vehicles into a safer and greener state. Resource allocation is one of important problems in vehicular networks. In this paper, to achieve a good trade-off between energy efficiency and transmission latency, a novel resource allocation method is proposed based on actor-critic reinforcement learning approach. As known, actor-critic reinforcement learning is an acknowledged solution for the problem with continuous valued state and action variables. Further, considering that edge computing technology for vehicular networks, the architecture of fog radio access network (F-RAN) is exploited to formulate the resource allocation problem in this paper. By adopting fog computing, the transmission delay is decreased because the F-RAN nodes with computing ability, are very close to vehicles so that the related information message does not need to be transmitted through the whole vehicular network. Further, to increase access capacity, the proposed method is deployed into non-orthogonal multiple access (NOMA) system.