During the last 40 years, many software reliability growth models (SRGM) have been proposed for estimating the reliability of software. Most of them are nonlinear functions, which makes it hard to estimate the parameters. A parameter estimation method based on the Ant Colony Algorithm is proposed in this paper. Numerical examples based on three sets of real failure data are presented and discussed in detail. It is shown that (1) the traditional method can’t get possible solutions for some models and datasets, while the proposed method always can get one; (2) comparing with PSO algorithm, our method's results have nearly a 10 times higher accuracy for most of the models.
20 December 2023 in Journal of Software: Evolution and Process
Current testing methods for autonomous driving systems primarily focus on simple traffic scenarios, generating test cases based on traffic accidents, while research on generating edge test cases for complex driving environments by traffic regulations is not adequately comprehensive. Therefore, we propose a method for scenario modeling and violation testing using an autonomous driving system based on traffic regulations named TraModeAVTest. Initially, TraModeAVTest constructs a Petri net model for complex scenarios based on the combination relationships of basic traffic regulation scenarios and verifies the consistency of the model’s design with traffic regulation requirements using formal methods, to provide a representation of traffic regulation scenario models for the violation testing of autonomous driving systems. Subsequently, based on the coverage criteria of the Petri net model, it utilizes a search strategy to generate model paths that represent traffic regulations, and employs a parameter combination method to generate test cases that cover the model paths, to test the violation behaviors of autonomous driving systems. Finally, simulation experiment results on the Baidu Apollo demonstrate that the test cases representing traffic regulations generated by TraModeAVTest can effectively identify the behaviors of autonomous vehicles violating traffic regulations, and TraModeAVTest can effectively improve the efficiency of generating different types of violation scenarios.
Workflow systems are widely used in our daily life so that the validity, dependability and security with which they need to be assured are important. However, existing researches mainly focus on correctness validation, performance analysis and assignment scheduling, but the testing methods have been seldom suited. In this paper a formalized definition of workflows constrained by an input and output is presented, and based on that, a Petri Net-based model (I/O_WF_Net) is proposed. In I/O_WF_Net, the activities of the workflow can be modeled as transitions of a Petri Net, and the inputs and outputs of an activity can be modeled as places. After the modeling method for I/O constrained workflow net into the I/O_WF_Net model is described, the corresponding transforming algorithm and its simplifying method are given.
1 May 2021 in IEICE Transactions on Information and Systems
During the recent several years, deep learning has achieved excellent results in image recognition, voice processing, and other research areas, which has set off a new upsurge of research and application. Internal defects and external malicious attacks may threaten the safe and reliable operation of a deep learning system and even cause unbearable consequences. The technology of testing deep learning systems is still in its infancy. Traditional software testing technology is not applicable to test deep learning systems. In addition, the characteristics of deep learning such as complex application scenarios, the high dimensionality of input data, and poor interpretability of operation logic bring new challenges to the testing work. This paper focuses on the problem of test case generation and points out that adversarial examples can be used as test cases. Then the paper proposes MTGAN which is a framework to generate test cases for deep learning image classifiers based on Generative Adversarial Network. Finally, this paper evaluates the effectiveness of MTGAN.