A Variable Probability Adaptive Random Testing Approach Based On Restricted Selection And Program Information

Nowadays,with the intensification of information technology,the scale of software is also increasing.However,the quality of software has always been the focus of attention.With more and more investment in software testing,the cost of software testing is getting higher and higher.How to do test more automatically is a problem that software testers must consider.One common method that used in automated testing is random testing.But the effectiveness of random test is not very satisfactory,so in recent years some scholars studied how to improve Random testing.T.Y.Chen proposed an adaptive random testing method and achieved better results than random testing.The main idea of the adaptive random testing method is to distribute the generated test cases more evenly in the input space.This paper analyzes an adaptive random testing based on the probability density function.This type of adaptive random testing is currently one of the most effective adaptive random testing method.The adaptive random testing based on the probability density function firstly generates probability density function based on already executed test cases and then use this probability density function to generate next test case.This paper analyzes two kinds of adaptive random testing methods based on probability density function.One is the Restricted Random Testing?RRT?algorithm and the other is the Adaptive Random Testing through Test Profile(ART_TP)algorithm.After analyzing two adaptive random testing based on probability density function,this paper improves the shortcomings of these two algorithms.At the same time,we also did some experiments on the proposed algorithms.A test prototype system was also designed and implemented.The main work completed in this article is described below:1.Aiming at the problem that RRT algorithm incurs a high time cost,an active Restricted Random Testing by Largest Available Zone(RRT_LAZ)algorithm is proposed.The main idea of RRT_LAZ is that it firstly generates exclusion zones and available zones around each executed test cases,and then calculate all the available zones by algorithm,and last find the largest available zone to generate test cases.At the same time,the algorithm considers the situation of multidimensional input domain,and gives two specific solutions.Finally,the simulation program and the real program are tested.The experiment results show that the RRT_LAZ algorithm can effectively reduce the time cost of RRT and can use less test cases to find more errors.2.Analyzing the probability density function of ART_TP algorithm and a new probability density function is proposed.With the new probability density function,the proposed algorithm is more efficient than the ART_TP algorithm.At the same time,aiming at the shortcomings of adaptive random testing that does not incorporate the dynamic information of the program,probability ART method based on coverage and path-based is proposed?Probability Adaptive Random Testing with Dynamic Program Information,PART-DPI?.Through the integration of coverage and path information,ART with variable probability is more targeted in selecting test cases and can meet higher coverage or paths.Experimental results also show that variable probability ART based on program dynamic information can be more effective than the original variable probability ART algorithm.3.Designed and implemented a test prototype system?Probability ART Prototype Testing System,PART_PTS?.The prototype system includes parametric analysis module,mock data module,program instrumentation module,mutation module,execution module,and result analysis module.Through the collaboration of these modules,the entire testing process of the paper was completed.Through experiments,we can find that the system has good feasibility and high scalability.