A Context-Aware Hybrid Trust Management Model for the Internet of Vehicles

Abstract

The Internet of Vehicles (IoV) represents an emerging paradigm in intelligent transportation systems, connecting vehicles, roadside infrastructure, pedestrians, and cloudbased services to enable advanced communication and collaborative decision-making. As IoV networks rapidly expand, ensuring the reliability, security, and authenticity of information exchanged between vehicles becomes increasingly critical. Conventional security mechanisms primarily rely on cryptographic methods which offer insufficient protection against intelligent attacks where authenticated nodes behave maliciously. To mitigate these risks effectively, trust management emerges as a vital complementary approach, assessing trustworthiness based on behavioral analysis, historical interactions, and data integrity.This thesis critically investigates existing trust management strategies within IoV, identifying key strengths and limitations in current methodologies. Through a comprehensive literature review, the research categorizes trust models into conventional and machine-learning ones, highlighting gaps related to context-sensitivity and adaptability in highly dynamic vehicular networks. The analysis underscores a need for more sophisticated trust evaluation mechanisms capable of dynamically integrating diverse trust indicators to enhance accuracy and responsiveness.To address these gaps, this research proposes DTM – IoV, a Dynamic Trust Computation Model designed for IoV environments. The envisaged model integrates both entity-centric and data-centric parameters. A distinctive feature of the envisaged model is its adaptive weighting mechanism, which dynamically adjusts the emphasis (weights) between entity trust and data trust according to real-time context and historical interactions. This ensures that trust computations remain relevant and precise even amidst dynamic conditions of IoV environments. The effectiveness of the developed trust model was extensively evaluated through simulation experiments reflecting realistic IoV scenarios. Performance metrics assessed included the model’s effectiveness in distinguishing malicious behaviors from benign activities, responsiveness to fluctuating node behaviors, and ability to maintain stable trust estimations. Simulation results highlighted significant improvements over traditional trust models, particularly, in terms of rapidly identifying malicious nodes, avoiding extreme fluctuations, and consistently maintaining accurate and stable trust values for legitimate vehicles.Overall, this thesis contributes substantially to the trust management domain by providing an in-depth evaluation of current models and introducing an innovative hybrid model optimized for dynamic, heterogeneous IoV networks. The findings and methodologies presented offer valuable insights and foundational advancements, laying the groundwork for future research and practical implementations to achieve robust, secure, and trustworthy vehicular communications.