Software trustworthiness evaluation model based on a behaviour trajectory matrix

Abstract

Software trustworthiness is a highly important research topic. Trustworthiness evaluation based on factors that affect software behaviour is conducted mainly according to the influence degrees of these factors on the software behaviour to evaluate trustworthiness. As a result, minimization of the interference of human factors is considered. In this study, to ensure the objectivity of evaluating the trustworthiness of software behaviour, a software trustworthiness evaluation model based on a behaviour trajectory matrix, namely, BTBM-TM was proposed. Checkpoints were set up in the trajectory of the software behaviour, and binary code was introduced to express the software behaviour trajectory tree. The scenario information of the checkpoints was acquired, and used to construct behaviour trajectory matrices, which were used to represent the behaviour trajectory. The behaviour trajectory matrices were transformed into grayscale images, which were used to train the deep residual network (ResNet) to classify the software behaviour. The trained deep residual network was used to categorize the current software behaviour, and the cosine similarity algorithm was used to calculate the deviation degree of the software behaviour trajectory; to perform a dual evaluation of the trustworthiness of software behaviour. The behaviour trajectory information of the Model class of 300 cycles was used to evaluate the trustworthiness of the mine-sweeping game. The trustworthiness evaluation results of the software behaviour of the scheme proposed in this paper (BTBM-TM) were compared with those of the schemes from references [6] and [10]. The accuracies of the schemes from [6] and [10] are lower than that of the BTBM-TM scheme. The trajectory of software behaviour is represented by a matrix and converted into a grayscale image, whose processing method is used to evaluate the trustworthiness of software behaviour more objectively and accurately.