Abstract: Reputation and trust management systems have been useful in situations that involve interaction between mutually distrusting parties to function correctly and to fulfil their purposes. Trust Management System based on reputation gained popularity in recent time for estimating the trustworthiness and predicting the future behaviour of nodes and other network entities in a large-scale distributed system. Trust and reputation have gained importance in diverse fields such as economics, evolutionary biology, distributed artificial intelligence, grid computing, and agent technology, among others. Many researchers have proposed different approaches to compute reputation/trust value to evaluate the trust worthiness of the entities involved. Approaches mainly differ based on the system model they have used, entity for which trust worthiness is computed and area of application. This paper explores the trust model based on reputation to provide security to both MA and executing host. In order to establish safe and secure communication between agent and hosts, each must be trusted that it would not harm other when it is given the access in a system. Paper presents a new way to compute reputation value of both host and mobile agent based either on past experience or experiences of other trusted and known entities and third party. Reputation of host is evaluated by the previous experience of the MAs being executed on the host and also by using intrusion detection mechanisms.

Keywords: Mobile Agents (MA), Mobile Agent Systems (MAS), security, Intrusion Detector System.