Research On Software Self-recovery Technology For Real-time Embedded System

As the complexity of the Real-time Embedded System increases,many software bugs cannot be avoid, it induce software aging and has been a critical problem that have to be accepted. Understand the causes of it, take steps to limits its effects, temporarily reverse some of the software failure it has caused.Therefore, research on software self-recovery technology for Real-time Embedded System has an important significance to improve the reliability of the system.Taking the Real-time Embedded System VxWorks as the research object.Firstly,this paper present the design and implementation of an detection based monitoring tool used to collect operation system resource usage and the state of software operation at regular intervals, from VxWorks system.. After statistical information was collected, it propose a methodology for detection and estimation of aging and failure time in the VxWorks operating system.The slope estimation techniques are applied to these data to detect/validate the existence of aging.Then this method determine the time to exhaustion with least square method. In order to describe the implementation process of the self-recovery strategy more clearly, the background of software self-recovery and the characteristics of real-time embedded systems should be researched in this paper.The multi-level nested software self-recovery policy is modeled and analyzed,and it is described by Non-deterministic Finite Automata.At the end of this paper,the software recovery granularity is the task of VxWorks system.According to the four principles of communication between tasks,the restart chain is formed.In the experiment,software architecture and data access relationship between tasks are analyzed. The paper present here is a recursive approach, in which some tasks at the beginning of the restart chain is recovered first.Do not progressively the tasks in the next node are recovered, until it does not work. These cheap methods of self-recovery in the article are new approach to high availability and reliability, extend the time to failure and shorten mean time to repair.