论文部分内容阅读
本文探讨了在Internet环境下,知识型MAS系统的任务负载分配问题,指出了它与网格计算的区别,本文的方法与传统的计算资源分配方法相反,将任务负载看作资源,而将Agent作为消费者,每处理一个任务,就获得一定数量的边际效用,利用市场机制,使Agent自主进行任务负载的分配。设计了基于均衡市场和连续CDA市场的两种调度算法,对算法的效果和特点进行了分析,证明市场机制的调度算法可以有效地实现MAS系统任务负载的平衡。此外,本文还给出了一个简单的确定均衡价格的方法。
This paper discusses the problem of task load distribution in knowledge-based MAS system and points out the difference between it and grid computing. In contrast with the traditional method of computing resource allocation, this paper regards task load as resource, As a consumer, every task is handled, a certain amount of marginal utility is obtained, and the market mechanism is used to make Agent allocate task load autonomously. Two scheduling algorithms based on equilibrium market and continuous CDA market are designed and their effects and characteristics are analyzed. It is proved that the market mechanism scheduling algorithm can effectively balance the task load of MAS system. In addition, this article also gives a simple way to determine the equilibrium price.