A Formal Approach to prevent Attacks on Mobile Agent Systems

One of the most important issues in mobile agent systems is the security aspect. For the purpose of ensuring security, we provide in this paper a formal model for secure mobile agent system. This model supports the specification of numerous security policy types which control the behaviour of system entities and protect them, as far as possible, from attacks that may occur. In order to verify whether a security policy could overcome a given kind of attack, we present a formal approach to this end. Thus, we propose a precise and formal representation of possible attacks that may occur in a mobile agent systems. According to this representation, we define a set of proved theorems which describe under what conditions a policy succeeds to prevent those attacks. The proposed specifications have been formalized using Z notation and checked using the Z-EVES tool.