Software aging oriented trustworthiness measurement based on weighted Boltzmann entropy

With the rapid development of software, various software accidents emerge one after another. The catastrophic consequences caused by these accidents make people realize the importance of software trustworthiness. As an indispensable means to ensure software quality, traditional trustworthiness measurement evaluates the software trustworthiness by studying the trustworthy attributes in a static way. However, most of the factors considered in trustworthy attributes tend to be dynamic with time. The current research often ignores the changes in software after running for some time, and cannot reflect the changes in software trustworthiness at different running times. Our objective in this paper is to study the relationship between running time and software trustworthiness, and design a running time-related software trustworthiness measurement model from the untrustworthy evidence related to software aging. We first extract the untrustworthy evidence from the bugs related to software aging in 5 subsystems of 4 public defect databases and 18 well-known software accidents, establish a risk level model, and design metric elements of untrustworthy evidence based on software aging. Then we construct a software aging cause category trustworthiness measurement model based on Boltzmann entropy. Finally, we build a software trustworthiness measurement model based on weighted Boltzmann entropy. For the weight values used in the model, the Brassard Priority Synthesis Analysis method was used to determine them. Different from the common resource consumption parameter and performance parameter, a model based on weighted Boltzmann entropy can describe the influence of various parameters on the software’s trustworthiness through risk state. It can reflect the change of system state and describe the system state completely. The empirical study shows the effectiveness and practicality of our method for evaluating software dynamic trustworthiness. Meanwhile, it also indicates a promising avenue for future research and application in the field of software trustworthiness measurement.