Migration Strategies Of Agent Based On Support Vector Machines

Mobile Agent is a program, which can migrate from one machine to another in a heterogeneous network and keeps communion with other agents and resources of hosts in network. Besides autonomy, intelligence and adaptability, it has mobility. The migration mechanism and migration strategies of mobile agent are kernel technology of mobile agent research. Compared with the research of migration mechanism, the research of migration strategies is fewer.Support Vector Machines (SVM) is a machine-learning tool based on statistical learning theory (SLT). It is developed from the theory of the optimal separating hyperplane in condition of linearly separable. It is approximate realization of structural risk minimization theory too. Compared with traditional machine-learning theory, SVM has stricter theoretic and mathematic basis. It has nicer extension and will not arise the problem of local minimization.Numerous systems only consider the software resource but ignore the effect of network resource and hardware resource of hosts when they migration an agent. Moreover, they don't use the traveling experience of agent. Refer to current models and criterions, this paper proposed migration strategies based on support vector machines and design the model. In the process of creating the model, this paper mends the itinerary table of Concordia and design itinerary table with linked structure. Then, we import SVM into the intelligent model and mend traditional agent model. This paper lodges a method that can get more knowledge from itinerary experience than current model. By this method, agent can get preferable learning effect after lesser traveling times. So, this method improves the practicability of the intelligent migration model. At last, we simulate and validate this model.