The social internet of things (SIoT) enhances the internet of things (IoT) by integrating social networking into device connectivity and communication. Trust models evaluate entities' trustworthiness to enable secure exchanges with reliable peers and minimise harmful interactions. Current methods lack balance and fail to account for changing social relationships and varied node behaviours across services. This research proposes a context-aware and resilient trust computation (CRTC) model that uses direct trust service, degree of participation, social relationship trust, and recommended trust to determine trust scores. Simulation tests and security assessments have confirmed its efficacy. The model categorises nodes with 100% precision, 93% recall, a 96% F1-score, and 94% accuracy with 25% malicious nodes. It resists various trust-based attacks and maintains over 85% service success rates in environments with up to 80% malicious nodes. The model identifies malicious nodes in two transactions and trustworthy nodes in four transactions.
Kumari et al. (Thu,) studied this question.