Improving Security in Multi-Agent Systems by Using a Novel Reputation Model

-Multi-agent systems like Peer-to-Peer networks are widely used for content sharing and searching. However, privacy and security are the concerns with such networks as they are vulnerable to attacks due to open, dynamic and anonymous nature. To protect communications over such networks many models came into existence that are based on the reputation or trust of the peers that interact with each other. However, they can not cope with the unpredictable behavior of malicious peers. To overcome this problem and improve QoS (Quality of Service) Das and Islam presented a dynamic trust computation model. In this paper we implement that trust model and build a prototype of multi-agent system to show the proof of concept. The experimental results revealed that the application facilitates dynamic trust computation and achieves secure communication and quality of service by balancing load among the agents. Index Terms – Peer-to-Peer networks, trust, trust management, malicious agents, load balancing