基于熟人的Agent联盟策略

在多ａｇｅｎｔ系统中，ａｇｅｎｔ可以通过协商形成联盟来完成任务求解。本文基于ａｇｅｎｔ联盟的统计规律提出了ａｇｅｎｔ熟人的概念，以熟人为基础的联盟策略可以有效减少联盟形成过程中的通讯开销和计算量。随后给出了一种基于该策略的联盟竞争任务的算法，该算法适用于动态，开放的环境、无中心控制，不存在通讯和计算瓶颈。

STRATEGY OF AGENT COALITION BASED ON ACQUAINTANCE

Task execution in multi agent environment may require agents forming agent coalition. This paper discusses the concept of acquaintance based on the regulation of forming agent coalition. The strategy of agent coalition based on acquaintance can reduce cost of communication and computing works effectively during forming coalition. Then we present the algorithm of coalition striving for task based on the strategy, The algorithm can be used for task allocation in dynamic, open system. There is no central node and bottleneck in communication and computation.