A Trade-Off Approach to Optimal Resource Allocation Algorithm with Cache Technology in Ubiquitous Computing Environment
A Trade-Off Approach to Optimal Resource Allocation Algorithm with Cache Technology in Ubiquitous Computing Environment
复制标题
DOI：
10.1109/cse.2009.472
复制
发表时间：
2009-08
期刊：
影响因子：
--
通讯作者：
M. Dong;Long Zheng;K. Ota;Song Guo;M. Guo;Li Li-Li
中科院分区：
文献类型：
--
作者：
M. Dong;Long Zheng;K. Ota;Song Guo;M. Guo;Li Li-Li
无处不在的技术是人类日常生活越来越现代化不可或缺的。然而，这些技术并不容易通过基础设施和其他相关技术在我们的世界各地普及。我们致力于一个项目，以满足这些挑战，目标是为即将到来的无处不在的社会构建一个框架。在我们以前的工作中，我们已经提出了UMP-PerComp，一个普遍存在的基于多处理器的流水线处理架构，以支持强大的和普遍的应用程序的开发。本文提出了一种能够提高系统整体性能的资源分配策略，特别是对动态分配算法进行了深入的研究，探讨了各种缓存策略，包括禁用缓存、非限制缓存和限制缓存。基于改进的原型系统和仿真环境，对这些算法在大规模应用场景下的性能进行了评估，实验结果表明，该算法在负载均衡、执行时间、等待时间和执行效率等方面都有明显的改善.
Ubiquitous technologies are indispensable formodernizing human daily life more and more. However,the technologies are not easily widespread everywhere inour world through infrastructures and other relatedtechniques. We have worked on a project to meet thesechallenges with a goal to construct a framework for thecoming ubiquitous society. In our previous works, we haveproposed UMP-PerComp, a Ubiquitous Multiprocessorbasedpipeline Processing architecture, to supportdevelopment of powerful and pervasive applications. Inthis paper, we proposed a new resource allocation policythat can improve the overall performance of the system.In particular, we extensively studied the dynamicallocation algorithms by exploring various cache policieswhich include disabled cache, unrestricted cache andrestricted cache. Performance of these algorithms in largescale application scenario is also evaluated based on boththe improved prototype and a simulation environment.The experimental results show a significant performanceimprovement achieved by the proposed algorithm interms of load balance, execution time, waiting time andexecution efficiency.