| With the explosion of software size and complexity, the traditional methods to enhance the software dependability are not sufficient. The researchers pay more attention on the health management of runtime software, the aim of which is to assure the state of software runtime complaint with the anticipation of users. Fault detection and recover is an efficient way in the scope of software health management.This dissertation surveys the research about fault detection and fault recover and discusses the relative merits of every method. A new framework of fault detection and recover is proposed in this dissertation. The main idea of the framework is that we first choose a specification language to define the properties of runtime software, second generate the fault detection code based on the specification, then intrude the detection code in the original program to monitor the software execution, and call the recover module to recover the faults being detected.The program invariants which are a typical specification have been successfully used in static checking, program verification, software testing, assertion checking et al. Because of the development of program analysis, some tools have been invented to generate program invariants automatically. So this dissertation researches the fault detection and recover based on invariants.This dissertation discusses the techniques about the generation of invariants, the generation of fault detection code, and the fault recover method. Based on these techniques, the tool of fault detection and recover is realized. The experiments show the effectiveness of the tool.Researchers keep on pursuing the reducing of software failure, and fault detection and recover is an effective way to reduce the software failure. The tools of fault detection and recover based on invariants will make positive influence on enhancing the software dependability. |
