Master Thesis - 2003

Recently, mobile agent technology has received much attention. Mobile agent completes its task instead of the user in large scale and complex networks. Especially, in a multi-agent system, some clone agents work in parallel on network nodes.

In this paper, we propose an effective mobile agent execution control method based on processing status of each clone agents. The proposed method achieves the reduction of the amount of the response time with allocating CPU time to each clone agent in ascending order of the number of its executing nodes. Moreover, to assure fairness among agents, the proposed method bounds the amount of CPU time per unit time allocated to each agent over the network.