Trust on wheels – towards trust management in the Internet of Vehicles

Over the past decades, the considerable promising advancements in the telecommunications and automotive sectors have empowered drivers with highly innovative communication and sensing capabilities, in turn paving theway for the next-generation connected and autonomous vehicles. Today, vehicles possess the potential to communicate wirelessly with other vehicles and vulnerable pedestrians in their immediate vicinity in order to share timely safety-critical information (alerts) primarily for the purposes of collision mitigation. Furthermore, vehicles can seamlessly liaise with the traffic management entities via their supporting network infrastructure to become aware of any potential hazards on the roads and for guidance pertinent to their current and anticipated speeds and the travelling course to ensure more efficient traffic flows. This unprecedented evolution has hence led to the promising paradigm of the Internet of Vehicles (IoV) which is generally referred to as an intelligent and inevitable convergence of the Internet of Things, intelligent transportation system, edge and/or fog and cloud computing, and big data that could be intelligently harvested for the cooperative vehicular safety (and non-safety) applications as well as cooperative mobility management. A secure and lowlatency communication is, therefore, indispensable in order to meet the stringent performance requirements of safety-critical vehicular applications.

Whilst the challenges surrounding low latency are being addressed by researchers in both academia and industry, i.e., primarily by developing state-of-the-art radio access technologies (RATs), and intelligent mechanisms for a heterogeneous amalgamation of several existing and new RATs together with their flexible deployment via a highly agile networking infrastructure, it is the security of an IoV network which is of paramount importance. It is indispensable to mention here that a single malicious message is usually capable enough of jeopardizing the entire networking infrastructure and can prove fatal for both the vehicular passengers and the vulnerable pedestrians.

Over the years, a considerable number of security techniques have been envisaged within the research literature for vehicular networking environments which have fundamentally relied on the conventional cryptographic-based solutions employing both public key infrastructure and certificates. Nevertheless, it is pertinent to highlight that the conventional cryptographicbased solutions are not optimal for IoV networks as vehicles are highly dynamic in nature and are distributed throughout the network, the availability of a networking infrastructure cannot be guaranteed at all the times, and cryptographic-based solutions are themselves susceptible to a poor key hygiene, compromised trust authorities, and insider attacks which are insidious and are, therefore, capable of causing a catastrophic damage.

Accordingly, this Ph.D. dissertation employs the emerging yet promising notion of trust as an alternative for ensuring security within IoV networks. Trust itself is a derived quantity and is assigned to each vehicle in accordance with its behaviour. It is contemplated as the currency of interaction between a trustor and a trustee. Nevertheless, weighing these vehicular interactions (often referred to as the trust segments) is indispensable in a bid to enhance the overall accuracy of a trust model and should be ascertained by taking into conformance the prior knowledge and context possessed by the corresponding one-hop neighboring vehicles, i.e., trustors, of a targeted vehicle, i.e., trustee. This dissertation, therefore, envisages a distributed trust management system which ascertains the trust for all the reputation segments within an IoV network and further determines their respective weights by taking into account the salient characteristics (i.e., quality attributes) of familiarity, similarity, and timeliness. An intelligent trust threshold mechanism has been also envisaged which is capable of identifying and evicting misbehaving vehicles from an IoV network in an accurate manner. The performance of the envisaged IoV-based trust model has been investigated in terms of optimizing the misbehavior detection and its resilience to attacks. The experimental results depict that our envisaged IoV-based trust model outperforms other similar state-of-the-art trust models to a considerable extent in terms of optimizing the misbehavior detection and its resilience to attacks.

This Ph.D. dissertation further envisages a scalable hybrid trust-based model which introduces a composite metric encompassing the weighted amalgamation of a vehicle’s computed trust score and its corresponding available resources in order to guarantee that the stringent performance requirements of the safety-critical vehicular applications could be met. Also, a Hungarian algorithm-based role assignment scheme has been envisaged for the selection of an optimal cluster head, proxy cluster head, and followers among the members of a vehicular cluster in order to maximize its overall efficacy. The performance analysis demonstrates the efficaciousness of our envisaged scheme.

Moreover, the notion of machine learning has been exploited to formulate a trust computational framework which possesses a true potential of being a generic algorithm in terms of its applicability to a number of application domains and across a diverse set of trust attributes. This, thus, facilitates in ensuring that the challenges pertinent to quantification of the weights, i.e., for numerous trust attributes, over the course of the trust aggregation process in a bid to ascertain both accurate and intuitive trust values, along with an optimal threshold in order to determine the region of trustworthiness, could be tackled.