摘要
联盟形成是多agent系统中的一个核心问题,由于系统的计算资源有限,因此设计一种能有效缓解计算瓶颈和信息拥塞的联盟形成策略是关键。基于蚁群算法的正反馈思想,以熟悉度表示agent之间的相互关系,提出了一种具有"正反馈"特性的联盟形成策略。仿真实验的测试及分析说明了该策略能在一定程度上降低整个系统的通信代价和资源开销,提高了系统的可靠程度。
Coalition formation is a key problem in multi-agent systems. Due to the constraint of system’s computation resource, a coalition formation strategy is needed to resolve the computational bottleneck and information congestion. Based on the positive feedback mechanism of Ant Colony Algorithm, this paper adopts familiarity to describe the interrelation between agents and proposes a new coalition formation strategy with positive feedback characteristic. The simulation experiment results show this proposed strategy can reduce the communication costs and resource consumption of the whole system and enhance its reliability.
基金
高等学校博士学科点专项科研基金项目(20060359004)
安徽省自然科学基金项目(070412035)
关键词
控制理论
联盟形成
蚁群算法
正反馈
熟悉度
control theory
coalition formation
ant colony algorithm
positive feedback
familiarity
