Critical Term for Performances of Push Model and Pull Model in Agent Communication and Implementation of Push Model

In various situations, mobile agents at different hosts must cooperate with one another by exchanging information and making decisions collectively. Communication efficiency is one of the most important factors affecting the performance of the highly dynamic and large-scale mobile agents system. To improve the communication efficiency and find a reliable location-transparency communication infrastructure for mobile agents, push model and pull model are two basic models used in communication algorithms. A detailed analysis of the two models is given and a theoretical formula of their performance comparison is derived. The formula indicates that only when the information traffic between mobile agents is extremely high, the performance of the pull model outperforms the push model, otherwise, the performance of the push model outperforms the pull model. In addition, an efficient mailbox-based communication algorithm using the push model by means of network compartmentalization is implemented. Besides the great reduction of the communication delay, the algorithm alleviates the load of the network and can be used to communicate efficiently and transparently between mobile agents.