Proceedings of the 2nd International Workshop on Security in Mobile Multiagent Systems

Abstract

This report contains the Proceedings of the Second Workshop on Security on Security of Mobile Multiagent Systems (SEMAS2002). The Workshop was held in Montreal, Canada as a satellite event to the 5th International Conference on Autonomous Agents in 2001. The far reaching influence of the Internet has resulted in an increased interest in agent technologies, which are poised to play a key role in the implementation of successful Internet and WWW-based applications in the future. While there is still considerable hype concerning agent technologies, there is also an increasing awareness of the problems involved. In particular, that these applications will not be successful unless security issues can be adequately handled. Although there is a large body of work on cryptographic techniques that provide basic building-blocks to solve specific security problems, relatively little work has been done in investigating security in the multiagent system context. Related problems are secure communication between agents, implementation of trust models/authentication procedures or even reflections of agents on security mechanisms. The introduction of mobile software agents significantly increases the risks involved in Internet and WWW-based applications. For example, if we allow agents to enter our hosts or private networks, we must offer the agents a platform so that they can execute correctly but at the same time ensure that they will not have deleterious effects on our hosts or any other agents / processes in our network. If we send out mobile agents, we should also be able to provide guarantees about specific aspects of their behaviour, i.e., we are not only interested in whether the agents carry out-out their intended task correctly. They must defend themselves against attacks initiated by other agents, and survive in potentially malicious environments. Agent technologies can also be used to support network security. For example in the context of intrusion detection, intelligent guardian agents may be used to analyse the behaviour of agents on a firewall or intelligent monitoring agents can be used to analyse the behaviour of agents migrating through a network. Part of the inspiration for such multi-agent systems comes from primitive animal behaviour, such as that of guardian ants protecting their hill or from biological immune systems.