A Security Evaluation Model for Multi-Agent Distributed Systems

Abstract

A new phase of the development of distributed system is now under way in which mobile software agent can be sent among supporting platforms. A mobile agent [6] is a program that represents a user in a computer network, and is capable of migrating autonomously from node to node, to perform computation on behalf of the user. Its tasks can range from online shopping to real-time device control to distributed scientific computing. By using agent technology, we can move the code to the remote data. We can also send out the agent to access the remote resources without keeping the network connection alive all the time. Despite its many practical benefits, mobile agent technology results in significant new security threats due to the increased mobility. A malicious agent can corrupt information on its host and in other agents. It is even more difficult to prevent a host from changing an agent’s states or even killing it. Therefore, solving the security problems of multi-agent distributed systems is crucial. It would be especially ideal if we can have a method to evaluate how secure an agent or a host is. Currently there is not much work done in this area. [1] and [2] are the few attempts to quantitatively measure the security of the mobile agent distributed systems. In this paper, we develop the security risk graph to model the system’s vulnerabilities. Based on the analysis of the different security threat situations in a mobile agent system, we derive a mathematical security model for quantitatively evaluating the security of an agent-based distributed system. This chapter is organized as follows: Section 2 gives the taxonomy of security threats for an agent-based system and describes other related works. In section 3, we will present a mathematical model for security assessment of an agent-based distributed system, followed by an example to explain how this model works. The section 4 provides the conclusions and future directions.