On time delay in distributed cooperation f self-interested multi-agent systems

摘要

The distributed cooperation is one of the most important problems in a multi-agent research. The problem to uniformly arrange the self-interested agent group which performs autonomous movement is the most basic problem in the distributed cooperation research. However, there is no research with consideration to the influence of the time delay by movement until now. In this paper, we consider an efficiency fall of the distributed cooperation by time delay, and its measure. First, when an agent's degree of activity is a degree in the middle, time delay has the largest influence, and it is shown that uniform arrangement achievement of an agent is greatly late then. We consider some acceleration technology, in order to prevent an efficiency fall. So we considered techniques as following: When an agent determines a move target, it is important to consider the neighborhood, And distribution of the agent group of the target neighborhood is considered. Furthermore, we are introduced the mechanism that is re-chosen destination during movement. Since the good acceleration effect was acquired by some evaluation experiments, reports it.