Cooperation among agents is critical in the field of agents' artificial intelligence (AT) and multi-agent systems (MAS). Many works have been done for cooperative target observation [1], foraging [2], and peer-to-peer systems [3]. Coordination is important to improve the performance of the entire system; and cooperation is the first step of coordination [4]. In such system, each agent can make the decision whether to cooperate (C) or defect (D) with other agents based on their past interactions and its decision making system. Evolutionary games, such as prisoner's dilemma (PD) [5], have been used widely to model a solution for the question of why and how intelligent agents successfully evolve cooperation instead of defection within their systems. Many researches [6, 7] have discussed network reciprocity, which is one of the five fundamental mechanisms that Nowak has classified [7] in 2006. Network reciprocity based on graph theory provides a very natural and convenient model to describe the spatial structure of cooperation evolutionary groups. Based on network reciprocity, individuals play games with their neighbors on an underlying network and copy a successful strategy among neighbors. The mechanism is simple but it still seems plausible for explaining why cooperation survives in many real complex systems [6].
展开▼
