Trustworthy and Reliable Artificial intelligence for VEhicuLar networks – TRAVEL

Future networks are expected to be a platform of “connected intelligence “solving human and societal challenges. This concept, referred to as "native artificial intelligence (AI)", is regarded as one of the pillars of 6G. In this paradigm, intelligence will be integrated at various levels of the communication network infrastructure to enhance performance and meet the requirements particularly those related to vehicular communications (V2X) such as enhanced reliability and responsiveness. However, the use of AI as black-box models poses significant risks and challenges. It is therefore crucial to understand and be able to trust the decisions made by these models. To address this issue, we plan to develop an explainable AI (XAI) framework that aims to explain the logic behind the black-box model behaviors for applications related to V2X networks. This could be accomplished by leveraging the insights gained from advanced XAI techniques to further refine and optimize the AI-driven solution, such as through enhanced optimization of both data and model components. This strategy will improve the network at various levels PHY layer, Software Defined Networking and Network Function Virtualization, thus ensuring a safe and efficient deployment.