An Artificial Intelligent-Driven Semantic Communication Framework for Connected Autonomous Vehicular Network

Semantic communication will considerably enhance transmission efficiency by exploring and only transmitting semantic information. However, most of the previous work in this field is limited to particular applications such as text, audio, or images and does not consider task-oriented communications, where the effectiveness of the transmitted information must be taken into account for completing a specific task. This paper focuses on developing a semantic communication framework for a high altitude platform (HAP)-supported fully connected autonomous vehicle network. A system model is proposed in which the traffic infrastructure (TI) transmits its semantic information to the macro base station (MBS) whenever it observes a connected and autonomous vehicle (CAV). The semantic information has been extracted using a convolutional autoencoder (CAE) as the encoder of CAE gives a smaller representation of the input data. Then, after receiving the semantic concept, the MBS decides on an appropriate action for the CAVs. A proximal policy optimization (PPO) algorithm in the MBS for interpreting and making a decision for the semantic concepts. Simulation results show that the proposed method can reduce up to 63.26% of the communication cost.