Enhancing Performance of Random Testing through Markov Chain Monte Carlo Methods

The drawback of classical software random testing is low efficiency to find failure-causing inputs, because it requires a large number of test cases compared to a family of partition testing. This paper proposes a software random testing scheme based on Markov chain Monte Carlo (MCMC) method. In this paper, we propose a probability model to represent the activities for finding failures in software testing. In experiments, we compare effectiveness of MCMC random testing with both ordinary random testing and adaptive random testing in real program sources. These results provide that MCMC random testing can drastically improve the effectiveness of software testing.