Credibility Evaluation Of Software Behavior Based On Behavioral Attribute Distance

Software is the soul of information industry. As the demand for function increases, the size of software system is gradually huge, defects and vulnerabilities are usually unavoidable. Software credibility refers to the dynamic behavior and result of software system is always meet prospective target. It can adapt to the change of certain environment and requirement. It has the ability to resist abnormalities and does not affected by accidental or malicious reasons. Software system can still maintain safe and reliable operation and continuous information service when it is subject to outside interference. However, because software often happens to malfunction or becomes invalid, the inconvenience is involved in people's work and life, and even huge economic losses. Evaluating accurately whether software is credible is the prerequisite for application and dissemination of software system.Software credibility mainly reflects in the behavior credibility, namely running behavior can be monitored, behavior results can be evaluated and abnormal behavior can be controlled. At present, the evaluation of software credibility is only limited to the measurement on static flow, which cannot ensure that software has same credibility in run time. Because the credibility measurement and evaluation method have serious defect, software products contain some known or unknown defects. In order to evaluate the credibility of software accurately, we need to monitor and formally describe running behavior of software, and then judge whether software is credible through subtle change of behavior data.In order to evaluate the credibility of software behavior more accurately, a credibility evaluation of software behavior based on behavioral attribute distance is presented in this paper. First, we monitor the operation of software in real time with sensor, formalize the behavior information and obtain the expected behavior trajectory and actual behavior trajectory of software. Then software behavior trajectory is described by the concept of software attribute. Meanwhile, the attributes of node in the trajectory are made a quantization disposal. Through an evaluation process consists of fixed-point evaluation, vertical evaluation and horizontal evaluation based on behavioral attribute distance, we conducted a consistency evaluation of the actual behavior trajectory and expected behavior trajectory of software, and thus judged the credibility of software behavior. Finally, the experiment has validated the efficiency and accuracy of this evaluation method.