Based Embedded Software Fault-tolerant Data Flow Anomaly Detection Research

Nowadays, electro-Mechanical-Products (EMPs) are getting more and more complex, this brings the growing demands on reliability and safety for EMPs. As an important part of EMPs, the safety and reliability of software determine the quality of EMPs directly. This paper takes research on improving reliability and safety of software for EMPs as the main clue, from the viewpoint of date flow analysis, techniques of software fault tolerance and fault rectification are studied deeply and systemly based on date flow abnomity inspection in this paper.Aiming at possible failure in software system, a novel software fault tolerance and fault rectification approach is proposed based on data flow abnomity inspection, software watch dog technique and rollback recover technique. This approach backups the object program regularly, establishes data flow analysis model via extract a group of related variables in object program and filter'outliers'from the data flow using data flow abnormity inspection method. When faults are found during the object program running, the watch dog force the object program jump to the backup point, rollback the object program to a backuped normal status, fault tolerance and fault rectification for instantaneous software failure can be implemented. Implementation framework, operation procedure for above software fault tolerance and fault rectification strategy are presented, binary regression model based on least square support vector machine and'outliers'inspection algorithm are also studied in this paper. Taking binary function as an example, simulation researches for above binary regression model and'outliers'inspection algorithm are conducted, both the correctness of the binary regression model and the validity of the'outliers'inspection algorithm are proven by simulation results.A recent bias distance based dynamic data flow abnomity inspection algorithm is proposed in this paper, it is a clustering concept based abnomity inspection algorithm, abnormities are detected by calculating the distance from cluster center to the test data, this algorithm is improved according to the characteristics of data stream, it can implement the data flow abnomity inspection effectively. This algorithm is simple and fast, it can choose the threshold adaptively, this improves the detection accuracy. Results of simulation experiments demonstrate the effectivity of this algorithm.Experimental scheme for software fault tolerance and fault rectification is designed based on data flow abnomity inspection, experimental platforms are constructed, software fault tolerance and fault rectification experiments are conducted. Experimental results show that the proposed strategies can implement software fault tolerance and fault rectification effectively, above research works provide some new solutions for software fault tolerance and fault rectification.