A Verifiable Discrete Trust Model (VDTM) Using Congruent Federated Learning (CFL) for Social Internet of Vehicles

Abstract

The Social Internet of Vehicles (SIoV) connects cars that are nearby and uses different types of infrastructure to connect people with shared interests. A public, open tool, such as the cloud, is used to share information about things like tolls, traffic, weather, and more. When people share social information, the risks of data leaks and trustworthiness are still not dealt with. This article presents a Verifiable Discrete Trust Model (VDTM) that uses Congruent Federated Learning (CFL) to make social information-sharing tools more trustworthy. The proposed trust model ensures pre- and post-sharing trust verification of the communicating vehicles. Trust is verified as a global identity factor due to the inconsistency between sharing occasions. The CFL is accountable of checking forward and backward trust between the times before and after sharing. In this learning, the congruency is zero-variance detection on both occasions of information sharing. The learning does this check over and over to make sure there is discrete trust in information-sharing times between vehicles, between vehicles and infrastructure, or between vehicles and platforms. The identified trust is valid within the specific interval broadcasted during request initializations. Depending on the trust level, the sharing interval is authenticated using forward and reverse private keys. Therefore, the vehicle's trust results from the maximum information integrity observed in the pre-and post-sharing interval. For the maximum vehicles considered, the proposed model leverages the trust index by 8%, information sharing by 7.15%, and reducing key overhead by 9.35% and time consumption by 7.76%.