Inter-confidentiality protection of agent communication in multi-agent system based applications

Abstract

Mobile agent interaction is usually vulnerable to attacks from within and outside the agent’s execution environment. Also, the mobility property of mobile agents earns them the opportunity to migrate from one security domain to another. Intranet/LAN with connection to internet do, from time to time, experience agent visitation either for malicious purpose or for legitimate mission. To protect legitimate agent communication against attack by visiting agent, we propose a technique that restricts migration of the visiting agent and isolate it to a neutral host where its mission could be achieved. We refer to this technique as restriction-based access control mechanism (ResBAC). The proposed mechanism employs certificate authentication, re-defining visiting agent itinerary path and visiting agent isolation to accomplish the aforementioned objective. The performance of the proposed mechanism is evaluated using scenarios to determine the strength of the mechanism in term of its ability to protect agent communication against the three major threats: man-in-the-middle attack, replay attack, and passive eavesdropping.