一个移动Agent算法性能评价的仿真环境

移动Agent是一种新的分布计算技术，自90年代被提出以来受到广泛关注。研究人员提出了很多基于移动Agent的算法，以解决传统分布式计算问题和移动计算中的问题。然而由于移动Agent所具有的反应性、自治性、可移动性等的特性使得对移动Agent的算法的性能评价成为一个相当复杂的问题。本文使用直接执行仿真(Direct Execution Simulation)的方法建立了一个移动Agent系统的仿真环境MADESE，通过执行移动Agent代码来获取其基本数据，并通过仿真计算出移动Agent算法在各种环境中的性能。文章讨论了MADESE的事件和同步机制以及其系统环境的设计结构，并介绍了其实现原型。

A Simulation Environment for Performance Evaluation of Mobile Agents

Mobile agent has attracted researchers in distributed computing for its features such as reactivity, autonomy and mobility. Many mobile agent-based algorithms are proposed to solve different kinds of distributed problems. However, it is those features that result in the problem to evaluate the performance of these algorithms, because the agent's behavior can be dependent with the environment and can migrate freely. In this paper, we propose a direct execution simulation environment called MADESE which consists of a set of simulators. MADESE can monitor and control the process of the agents, simulate execution environment, and generate the events for performance evaluation. The event and synchronization mechanism of MADESE is described in detail, and a prototype of MADESE is also introduced.