Cooperative behavior in decentralized mobile agents requires communication among them. However, such communication can also have negative consequences for a variety of reasons, and the benefits of communication can vary over the course of the agent team's task or mission. In this paper, we compare the benefits of communication over time for a team of mobile agents performing a general search-and-engage mission. Monte Carlo simulations show that communication provides variable benefit at

more »

... t times and under different circumstances. In particular, communication can sometimes lead to poorer performance if agents have imperfect sensors. I. INTRODUCTION The interaction between decision making and communication strategy is critical in a decentralized multi-agent system [1], [2], [3] . An approach used to study this is information value theory [4] used in single-agent contexts with two myopic assumptions [5] . Several attempts have been made to extend this theory to multi-agent systems [6], [7] , but the value of information measured by the difference between the improvement in the agents' performance and the costs associated with communication is usually intractable in decentralized problems. The role -and, therefore, the benefit -of communication in a multi-agent system depends strongly on the system's internal parameters, protocols and processes, as well as on the functional and environmental context of the agent team. In order for the agent team to make consistent decisions, the agents have to work from a shared base of information, which can only arise through communication. This may suggest that more communication -leading to more consistent shared knowledge -is always better. However, in a system where agents make decisions autonomously by applying relatively simple decision rules to available information, this is not necessarily true. Since communication always has a cost, significant savings may be realized by using communication only when it is likely to be most useful. We address this issue systematically for a simple decentralized multi-agent model. A decentralized Multi-Agent System (MAS) model based on perfect communication is established and analyzed [8], [9]. When, how and what to communicate in order to Y. Liao is with the