An efficient information sharing approach for large scale multi-agent team

Effective communication among agents in large teams is crucial because the members share a common goal but only have a partial views of the environment. Information sharing is difficult in a large team because, a team member may have a piece of valuable information but not know who needs the information, since it is infeasible to know what each other agent is doing. Although much related work has been done on efficient delivery of information, most work is based on assumptions which are not suited to large scale multiagent teams. In this paper, we made two contributions. Firstly, we present a solution to sharing information that is applicable to large teams based on previous research. A key to the solution is imposing a static network topology on the members of the team where each agent requiring communication to be only along very few links in that network. The key observation underlying this solution is that each piece of information is interrelated and the sender of a piece of information can ldquoguessrdquo who might need some information based on previously sent messages. Thus, when an agent has a piece of information, it can determine which of its neighbors in the network is most likely to either need the information or know who does, based on related messages previously received. Secondly, we investigate the influence of different types of team network topology on the efficiency of information sharing. Our results show that our algorithm works with various topologies but gets the best performance on a scale free network.