嵌入式软件状态监测与自恢复技术研究

发布时间：2018-04-04 05:01

本文选题：嵌入式软件　切入点：VxWorks　出处：《哈尔滨工程大学》2013年硕士论文

【摘要】：随着计算机技术与微电子技术的高速发展，人们对嵌入式系统的应用已经普及到社会的各个角落。伴随着嵌入式软件复杂性的不断提高，嵌入式系统自身的隐患也越来越多，嵌入式软件在运行的过程中一旦出现系统故障或软件失效，将很难完成使命，甚至有可能对用户造成巨大的财产损失。因此，对嵌入式软件可靠性的研究，已经成为当下的热门研究方向。本文以VxWorks下的嵌入式软件为研究对象，对嵌入式软件状态监测与自恢复技术进行研究。论文的主要工作如下：首先，阐述了嵌入式软件状态监测与自恢复技术的国内外研究现状，，简要的介绍了本课题的研究背景及意义，给出了论文的主要研究内容；主要介绍了嵌入式软件自恢复相关技术，包括介绍了嵌入式软件自恢复的必要性和研究过程中存在的主要问题；分别讲述了两种常用的嵌入式软件自恢复策略以及各自的优缺点；详细介绍了软件自恢复策略中最常用的修复技术——微重启技术。其次，分别从嵌入式系统性能和嵌入式软件行为两个方面对嵌入式的软件状态监测方法进行了研究。在系统性能监测方法研究方面，分别对系统内存可用量、任务自身的运行状态、CPU和任务堆栈的占用情况四个指标的监测方法进行了深入研究；同时提出了一种嵌入式软件衰老趋势分析的方法；并利用最小二乘法进行曲线拟合，采用了一种基于阈值的嵌入式软件失效时间估计方法。在软件行为监测方法研究方面，运用了一种构建软件行为树的方法来进行软件行为监测，并介绍了软件行为树中节点的分类、构建软件行为树的方法和软件行为树中行为的匹配规则。再次，结合VxWorks操作系统自身的特点，提出了一种适合嵌入式软件的嵌套式自恢复策略；分别介绍了基于系统性能监测的嵌入式软件自恢复策略和基于软件行为监测的嵌入式软件自恢复策略的详细流程。在第三章嵌入式软件状态监测结果的基础上，用所选择的监测指标构建状态空间；根据所选择指标的异常状态所对应的修复行为，构建动作空间，给出了修复动作评价模型；采用SARSA(λ)算法进行自恢复决策规则强化学习，在此基础上采用了一种基于规则的嵌入式软件自恢复决策方法。最后，对前面提出的嵌入式软件状态监测方法和嵌入式软件自恢复策略进行实验验证，证实了方法的有效性和可行性；最后通过比较，证明了本文采用的嵌套式自恢复策略的优越性。
[Abstract]:With the rapid development of computer technology and microelectronics technology, the application of embedded system has been popularized to every corner of the society.With the increasing complexity of embedded software, there are more and more hidden dangers of embedded system itself. Once there is a system failure or software failure in the running process of embedded software, it will be very difficult to complete the mission.It is even possible to cause huge property losses to users.Therefore, the research of embedded software reliability has become a hot research direction.In this paper, the state monitoring and self-recovery technology of embedded software based on VxWorks is studied.The main work of the thesis is as follows:First of all, the status quo of embedded software state monitoring and self-recovery technology at home and abroad is described, the research background and significance of this topic are briefly introduced, and the main research content of this paper is given.This paper mainly introduces the related technologies of embedded software self-recovery, including the necessity of embedded software self-recovery and the main problems in the research process.In this paper, two commonly used self-recovery strategies of embedded software and their respective advantages and disadvantages are described respectively, and the most commonly used repair technology of software self-recovery strategy, micro-restart technology, is introduced in detail.Secondly, the embedded software state monitoring method is studied from two aspects: embedded system performance and embedded software behavior.In the aspect of system performance monitoring methods, the monitoring methods of the system memory consumption, the running status of the task itself, the CPU and the occupation of the task stack are studied.At the same time, an embedded software aging trend analysis method is proposed, and the least square method is used for curve fitting, and an embedded software failure time estimation method based on threshold is proposed.In the research of software behavior monitoring method, a method of constructing software behavior tree is used to monitor software behavior, and the classification of nodes in software behavior tree is introduced.The method of constructing software behavior tree and the matching rules of behavior in software behavior tree.Thirdly, according to the characteristics of VxWorks operating system, a nested self-recovery strategy for embedded software is proposed.The detailed flow chart of embedded software self-recovery strategy based on system performance monitoring and embedded software self-recovery strategy based on software behavior monitoring is introduced respectively.Based on the results of embedded software state monitoring in the third chapter, the state space is constructed with the selected monitoring index, and the action space is constructed according to the repair behavior corresponding to the abnormal state of the selected index, and the evaluation model of the repair action is given.SARSAA (位) algorithm is used for reinforcement learning of self-recovery decision rules, and a rule-based self-recovery decision method for embedded software is proposed.Finally, the experimental verification of the state monitoring method and the self-recovery strategy of embedded software is carried out, which proves the effectiveness and feasibility of the method.The superiority of the nested self-recovery strategy is proved.
【学位授予单位】：哈尔滨工程大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP368.1

【参考文献】