Identification of misbehavior detection solutions and risk scenarios in advanced connected and automated driving scenarios

Abstract

The inclusion of 5G cellular communication system into vehicles, combined with other connected-vehicle technology, such as sensors and cameras, makes connected and advanced vehicles a promising application in the Cooperative Intelligent Transport Systems. One of the most challenging task is to provide resilience against misbehavior i.e., against vehicles that intentionally disseminate false information to deceive receivers and induce them to manoeuvre incorrectly or even dangerously. This calls for misbehaviour detection mechanisms, whose purpose is to analyze information semantics to detect and filter attacks. As a result, data correctness and integrity are ensured. Misbehaviour and its detection are rather new concepts in the literature; there is a lack of methods that leverage the available information to prove its trustworthiness. This is mainly because misbehaviour techniques come with several flavours and have different unpredictable purposes, therefore providing precise guidelines is rather ambitious. Moreover, dataset to test detection schemes are rare to find and inconvenient to customize and adapt according to needs. This work presents a misbehaviour detection scheme that exploits information shared between vehicles and received signal properties to investigate the behaviour of transmitters. Differently from most available solutions, this is based on the data of the on-board own resources of the vehicle. Computational effort and resources required are minor concerns, and concurrently time efficiency is gained. Also, the project addresses three different types of attack to show that detecting misbehaviour methods are more vulnerable to some profile of attacker than others. Moreover, a rich dataset was set up to test the scheme. The dataset was created according to the latest standardised evaluation methodologies and provides a valuable starting point for any further development and research.