【作者】 冯雪丽；

【导师】 刘素芹；

【作者基本信息】 中国石油大学 ， 计算机应用技术， 2008， 硕士

【摘要】 随着网格技术的高速发展,网格资源管理已成为实现高性能计算的关键。如何高效、准确、科学地发现网格资源是网格资源管理的一个重要问题。因为整个网格的计算资源、连同网格本身都是动态的,应用开发者更加迫切需要移动计算技术的支持。移动Agent可以携带可执行代码、数据和运行状态在网格各主机之间自主移动,能够较好的适用于网格的动态环境。在对国内外已有的网格资源发现机制研究的基础上,发现目前的分布式与集中式发现相结合的技术还不成熟、存在网格资源发现效率不高、不能适应网格动态性等问题。本文基于Agent技术将分布式与集中式网格资源发现机制相结合,建立了一种新的网格资源发现模型。在虚拟组织内提出了从本地存储到虚拟组织管理节点的快速资源发现方法,并改进了模型中移动Agent在动态网格环境下的路径优化算法。针对移动Agent在网格环境中的路径优化问题,研究了蚁群算法及其已有技术在网格环境中发现资源效率和动态适应性的不足,提出了相应的可行性的改进算法,其核心思想是用遗传算法对蚁群算法初始化规则进行改进,提高算法收敛速度,并采用节点更新规则来反映网格中不同节点符合用户的满意程度和它们的变化,将算法更好的应用于网格资源发现问题中。为了验证改进的蚁群算法在移动Agent网格路径优化方面的性能,本文设计并实现了几组实验,编程模拟了网格环境,从算法有效性、网格动态适应性以及算法的普适性方面对改进的蚁群算法与基本蚁群算法、遗传蚁群算法进行比较。实验结果证明,本文改进的算法是有效的,该算法解决了移动Agent在网格环境中的动态路径优化问题,从而,提高了网格资源发现的效率。更多还原

【Abstract】 With the rapid development of grid, grid resources management is the key to achieving high-performance computing. How to discovery grid resources efficiently, accurately and scientifically is an important issue to grid resources management.Because the whole grid computing resources, together with the grid itself is dynamic, so application developers more urgent need the support of mobile computing technology. Mobile agent technology used in grid can move automatically between the hosts of the grid carrying executable code, data and the running state, so it can be better used in dynamic grid environment.On the basis of in-depth study on the domestic and abroad grid resources discovery mechanism, the current technology of distributed finding-mechanism integrating with centralized finding-mechanism was discovered still not be mature.There are still some problems existing,such as low-efficiency finding grid resource, not adjustting to dynamic grid resources.This paper based on agent technology,combined with distributed and centralized grid resources discovery mechanism, and established a new grid resources discovery model. Designed rapid resources discovery method from the local storage node to the virtual organization manage node in the virtual organization, and improved model of mobile agent path optimization algorithm in the dynamic grid environment.Against to the path optimization problem of mobile agent in the grid environment, studied ant colony algorithm in the grid environment, found the ant conlony algorithm inadequate of efficiency and not adjustting to dynamic of the grid environment, the core idea of improving algorithm is to use genetic algorithms to improve ant algorithm initialization rules ,enhance algorithm convergence speed. And , use the rules of updating nodes to reflect the user’s satisfaction to different grid nodes and their changes,the algorithm will be applied to better grid resource discovery issues and make the algorithm applied to grid resources discovery issues better. In order to validate the performance of improved ant colony algorithm in grid path optimization of mobile agent, this paper designed a few experiments, simulate grid environment by programming, from algorithm effectiveness, grid dynamic environment as well as the universal application, compared the basic ant colony algorithm, genetic ant colony algorithm with improving ant colony algorithm. The experimental results show that this improved algorithm is effective, the algorithm solved the dynamic path optimization problem of mobile agent in grid environment, thus, improved the grid resources discovery efficiency.更多还原