分布式问题求解中的妥协策略  

The Behaviors of Compromises in Distributed Problem Solving

作　　者：唐屹[1] 

机构地区：[1]广州大学理学院数学系,广州510405

出　　处：《计算机工程与应用》2004年第17期89-92,共4页Computer Engineering and Applications

基　　金：广州大学2003年科研项目资助

摘　　要：基于多agent系统的分布式约束满足(CSP)问题的求解进程依赖于agent间的有效交互。该文针对着色问题(GCP)的分布式求解,提出了agent妥协的概念。通过妥协,两个相邻agent改变了各自原有的局部目标,实现了相邻约束的满足。模拟实验表明,妥协策略有助于提高分布式GCP问题的求解性能。该文还讨论了不同的妥协实现方式对性能的影响。The performance of the multi-agent based solving system for distributed constraint satisfaction problems (CSP)depends on the interaction among various agents.This paper introduces the notion of agent compromise in distributed graph coloring problems (GCP).The compromise is constructed between two neighboring agents and these two agents will change their own local behaviors in order to make a common decision.This common behavior implicitly satisfies all the neighboring constraints between the two agents.We have conducted some experiments to evaluate the system performance with some GCP instances.The results demonstrate the improvement of the performance with agent compromises.

关 键 词：多AGENT系统 分布式问题求解 约束满足问题 agent妥协 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]