The Improvement of GK-tail Algorithm of Software Behavior Modeling

Abstract

Software behavior modeling is very important to analyze software system security. The goal of software behavior analysis is to obtain the semantic description and understanding of the software behavior. Software behavior modeling and testing are the key steps to achieve the goal. Therefore, this paper studies the software behavior model algorithm, GK-tail algorithm. GK-tail algorithm uses interaction trace to describe the software behavior, which is concerned interaction between the software components. However, GK-tail algorithm is not accurate. Because GK-tail algorithm uses Daikon analysis tool to obtain the predicates, which contain some errors. In order to improve the accuracy of the GK-tail algorithm in software behavior modeling, this paper improves the traditional GK-tail algorithm based on the extended finite automaton. The improved GK-tail algorithm utilizes the combination of Daikon and static (ESC/Java) tools to filter out some errors of the generated predicates. In addition, this paper designs and implements a prototype system of the improved GK-tail algorithm. The prototype system proves that the improved GK-tail algorithm can filter out errors in the second step of GK-tail algorithm effectively.