An Efficient Agent Control Method for Time-constrained Applications with Heterogeneous Work Demand

Abstract

Mobile agent is an efficient technology that makes it much easier to handle large scale and complex network systems. Especially in information retrieval, agents can search effectively with the ability of migrating autonomously through the networks. To find better results for such applications, it is important for agents to complete their tasks on as many nodes as possible by their deadlines. However, most existing agent systems using processor sharing as scheduling disciplines do not take such time constraints into account. Therefore, agents are likely to miss their deadlines on many nodes. In this paper, we propose an efficient and robust control method of agents with heterogeneous work demand under time constraint. When there are multiple agents running on a node, the mutual influence among them may cause to make the time for each agent to complete its task longer. When each agent has different work demand, the mutual influence among them is also different. The proposed method takes such different mutual influences into account to calculate the estimated value of the number of agents that can complete their tasks before their deadlines, and uses this value to control the dispatching and execution of agents. By using simulation experiments, we have proved that the proposed method can keep the fairness among all the agents and improve the number of nodes where agents can complete their tasks within their deadlines.