Novel Trust Evaluation Model for Heterogeneous Hybrid Multi-Agent Cooperative Networks

In recent years, the use of multicluster hybrid collaborative networks consisting of agents with different functionalities has become increasingly common. These networks are used for various purposes, such as large-scale collaborative operations and emergency rescue missions. However, operating in an open and dynamic collaborative network environment exposes each cluster management node and the underlying agents to potential intrusion attacks, while their capabilities are constantly evolving. To ensure effective intercluster collaboration, this paper addresses the issue of trust evaluation in agents capable of performing collaborative tasks across different clusters. A novel trust evaluation model called ICEITrust is introduced, which has two main features. First, it incorporates an integrated trust evaluation for individual agents and the clusters in which they operate. Second, it combines explicit trust scores based on direct trust and recommendation trust with implicit trust scores derived from historical data and contextual information. Experimental results validate the effectiveness of ICEITrust in improving the success rate of intercluster task allocation. The results also highlight the superiority of ICEITrust, in particular its fast recovery performance, when faced with dynamic changes in the network environment. When compared to traditional and current trust evaluation models, ICEITrust demonstrates its effectiveness in mitigating trust-related challenges and facilitating secure and efficient intercluster collaboration.