A Dynamic Planning Method for Mobile Agent Implementation based on Service Recommendation

Mobile agent is a task proxy of the social member that is able to migrate continuously among hosts and use the local services to complete tasks. The hosts, known as workplaces, can be regarded as service proxies of other social members. The sequence of workplaces on which mobile agent completes its tasks is called path. The main contribution of this paper is proposed a dynamic planning method for mobile agent path based on the variable decision space. The path planning model is described with MDP, and the variable decision space is generated through social acquaintance recommendation, assuming that every social member can provide workplace and service recommendation for mobile agent in cooperative work manner. The experimentation results show that by utilizing social acquaintance recommendation, mobile agent is able to dynamically construct its path based on variable recommended space.