基于策略的任务流控制模型的研究

发布时间：2018-06-12 00:32

本文选题：策略 + 任务流　；参考：《沈阳师范大学》2016年硕士论文

【摘要】：随着IT技术的迅猛发展,信息处理的相关产业逐步得到发展,由此带来的计算机处理必成大势所趋。我们可以看到周围每天都会出现新的产品和服务,为了新产品的研发和更好的客户体验,任务流技术应用而生。系统将若干任务流按需求分离,同时将任务流分解成若干任务,通过调度算法并按照一定的策略规则来执行这些任务流及任务。为此,本文首先对任务流的理解加以总结,并对本文涉及到的策略知识和控制模型做了总结,在此基础上对基于策略的任务流进行了分析,接着,对本文提出的调度算法对所用的算法分析不足,最终提出了适合本文研究方向的调度算法和策略并设计出了SBAC模型。具体研究结果如下:1,SBAC以TBAC和PBAC为基础,从模型的对比研究,制定策略规则及设计调度算法三个方面进行了具体阐述,并将策略模型应用到任务流中,从而实现了基于任务和基于策略控制模型的有效结合,为本文的模型研究做好铺垫。2,通过对多种优先级算法和RTOS算法的对比研究、实时性/非实时性以及静态/动态调度的分析,本文在EDF算法基础上,分析和证明了EDF调度算法中最大可挪用时间的性质,结合EDF的最优调度过程,提出一种延长可利用时间的临近值NVEDF算法。3,调度策略规则分为静态调度策略和动态调度策略。大多数的任务调度都使用静态任务调度,也就是所有任务需求设定都是提前设置好的,只要等待系统开始运行就可以了。但是我们不得不考虑到一些复杂情况,因此本文提出了动态调度策略。本文的策略规则包括两种,一是系统内部的不同任务流执行策略;二是单条任务流内部的不同任务执行策略。当系统任务开始时,根据预先定义的策略选择调度算法并执行任务流间或单条任务流内的任务。
[Abstract]:With the rapid development of IT technology, the related industry of information processing has been developed step by step. We can see new products and services around us every day and task-flow technology applications for new product development and better customer experience. The system separates several task flows according to their requirements and decomposes them into several tasks. These tasks are executed by scheduling algorithms and according to certain policy rules. Therefore, this paper first summarizes the understanding of task flow, summarizes the strategy knowledge and control model involved in this paper, and then analyzes the policy-based task flow. The scheduling algorithm proposed in this paper is insufficient to analyze the algorithms used in this paper. Finally, the scheduling algorithm and strategy suitable for the research direction of this paper are put forward and the SBAC model is designed. The specific research results are as follows: 1: 1 SBAC is based on TBAC and PBAC. It is expounded in detail from three aspects: model comparison, formulation of policy rules and design of scheduling algorithm, and the policy model is applied to the task flow. Thus, the effective combination of task-based and policy-based control models is realized, which lays the groundwork for the research of the model in this paper. Through the comparative study of various priority algorithms and RTOS algorithms, Based on the analysis of real-time / non-real-time and static / dynamic scheduling, this paper analyzes and proves the property of maximum misappropriation time in EDF scheduling algorithm, combined with the optimal scheduling process of EDF. An approach NVEDF algorithm. 3, which extends the available time, is proposed. The scheduling policy rules are divided into static scheduling policy and dynamic scheduling policy. Most task scheduling uses static task scheduling, that is, all task requirements are set in advance, just wait for the system to start running. However, we have to consider some complex situations, so this paper proposes a dynamic scheduling policy. There are two kinds of policy rules in this paper, one is the different task flow execution strategy within the system, the other is the different task execution strategy within the single task flow. When the system task begins, the scheduling algorithm is selected according to the predefined policy and the tasks in the intertask flow or the single task flow are executed.
【学位授予单位】：沈阳师范大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP301.6

【参考文献】