Font Size: a A A

Automatic Fault Diagnosis And Isolation Method Of Safety Critical Software

Posted on:2008-08-28Degree:MasterType:Thesis
Country:ChinaCandidate:Q XuFull Text:PDF
GTID:2178360242998985Subject:Computer Science and Technology
Abstract/Summary:PDF Full Text Request
Safety critical software in the country's current national defense modernization shoulder the heavy responsibility. If they are fail, which will result in major losses of security, economic and even life. As the high complexity of safety critical software themselves and the environment, software faults are hard to avoid completely. Therefore, it is very important for us to make it possible that the critical task can be sustained before faults are repaired, with effective fault diagnosis and safe fault isolation.Based on the features of safety critical software, this paper introduces a framework for automatic fault diagnosis and fault isolation, which name is AFDIF. Through inserting binary code into the program, the method collects runtime data. Then it analyses data and locates fault by means of Boolean Learning mechanism. Finally, the method will automatically produce isolation codes and insert them into the program. It can effectively prevent the fault recurring, thereby increasing software reliability, safety and survivability. On this basis, my research mainly was in following respects:First, paper focus on the principles and the technical routes of software fault diagnosis and isolation methods. On this basis, paper introduce a framework of software fault automatic diagnosis and isolation that named AFDIF, including the realization of software run-time data acquisition method, method that data sampling and abstract in the course of test, fault diagnosis method based on Boolean Learning mechanism and fault isolation method based on Automatic Patch Instrumentation.Second, I research on the software's automatic fault diagnosis and positioning technology, including use of instrument technology in the procedures specified location insert instructions, which aim to realize software running data collection; at the same time, paper introduces fault diagnosis algorithm base on Boolean Learning mechanism, through analysis of data samples, and ultimately find "key" variables that bring on fail.Third, I research on fault isolation base on code instrumentation and realize the directional Jump procedures, thereby avoiding the implementation of code deficiencies.Finally, based on the above theory certification and technical support, I design and realize prototype tool of the framework. The results show that the method can accurately positioning the cause of the malfunction, and the timely and effective fault isolation.
Keywords/Search Tags:safety critical software, fault diagnosis, fault isolation, boolean learning, binary code instrumentation
PDF Full Text Request
Related items