An Edge-Based Digital Twin Framework for Connected and Autonomous Vehicles: Design and Evaluation

Abstract

Connected and Autonomous Vehicles (CAVs) will be provided with multiple sensing and connectivity options as well as embedded computing and decision-making capabilities. The resulting technological landscape paves the way for the deployment of a plethora of innovative applications involving different stakeholders, such as insurance companies, car repairs, car manufacturers and public authorities. In such a context it is crucial to collect data in an efficient manner, not to burden the vehicle itself and the network infrastructure, while also providing an interoperable data sharing among all the involved players. The Digital Twin (DT) concept can play a key role to properly retrieve, store and share data as well as to exploit them to monitor, predict and improve the vehicle safety and driving experience. This work proposes a comprehensive framework which encompasses the presence of an edge-based DT interacting with the vehicle and the remote applications. It leverages properly specified interfaces and semantic models for different types of data provided by on-board sensing and learning capabilities. A Proof-of-Concept (PoC) has been developed to assess the practicality of the proposal and its performance in terms of communication and computation footprint under a variety of settings.