IPSHO-Fed: a hybrid federated learning and spotted hyena optimization approach for trust assessment

In the era of rapidly evolving intelligent transportation systems, the vehicle–road–cloud collaborative system emerges as a groundbreaking paradigm that integrates vehicles, road infrastructure, and cloud servers to optimize transportation-related tasks. However, the system’s success hinges on the reliable and secure exchange of data and model training among entities. Trust evaluation becomes paramount to ensure the credibility of information transmitted within the system and safeguard against potential security threats. To overcome this issue, an innovative trust evaluation scheme using the improved PSO-based spotted hyena optimization (IPSHO) algorithm is proposed. In this approach, each entity in the collaborative system is represented as a spotted hyena, and the algorithm’s integration with federated learning enables decentralized model training without sharing raw data. By leveraging the collective intelligence of the system, IPSHO collaboratively refines trust values and ensures personalized trust evaluation at the equipment, data, and model levels. Our simulation environment, implemented in MATLAB, models a realistic network of 100 sensor nodes representing vehicles, road infrastructure, and cloud servers. The entities communicate within a 100 m × 100 m area using secure channels. Through rigorous experimentation, the IPSHO algorithm's convergence behavior, its ability to find optimal trust values, and its performance against malicious nodes and hazardous events are assessed. The results demonstrate IPSHO's effectiveness in achieving accurate trust evaluations in diverse scenarios. Comparison with other optimization algorithms and trust evaluation methods reaffirms IPSHO’s superiority in enhancing trustworthiness assessment. The proposed scheme empowers legitimate nodes with the capability to discern trustworthiness and authenticity of information exchanged, ensuring the integrity and reliability of the vehicle–road–cloud collaborative system.