Federated Learning for Digital Twin-Based Vehicular Networks: Architecture and Challenges
Emerging intelligent transportation applications, such as accident reporting, lane change assistance, collision avoidance, and infotainment, will be based on diverse requirements (e.g., latency, reliability, quality of physical experience). To fulfill such requirements, there is a significant need to deploy a digital twin-based intelligent transportation system. Although the twin-based implementation of vehicular networks can offer performance optimization. Modeling twins is a significantly challenging task. Machine learning (ML) can be a preferable solution to model such a virtual model, and specifically federated learning (FL) is a distributed learning scheme that can better preserve privacy compared to centralized ML. Although FL can offer performance enhancement, it requires careful design. Therefore, in this article, we present an overview of FL for the twin-based vehicular network. A general architecture showing FL for the twin-based vehicular network is proposed. Our proposed architecture consists of two spaces, such as twin space and a physical space. The physical space consists of all the physical entities (e.g., cars and edge servers) required for vehicular networks, whereas the twin space refers to the logical space that is used for the deployment of twins. A twin space can be implemented either using edge servers and cloud servers. We also outline a few use cases of FL for the twin-based vehicular network. Finally, the paper is concluded and an outlook on open challenges is presented.