Software Reliability Test Based on Markov Usage Model

摘要

The reliability test for embeddedsoftware system is very difficult due to its complexstructure and large scale. Markov usage modelwhich describes the software usage with Markovprocess is widely used for statistical test. Softwarestatistical test based on Markov usage model is aneffective approach to the generation of test caseswith high efficiency and the evaluation of softwarereliability in a quantitative way. This paper mainlyfocuses on the generation of Markov usage model ofsoftware system and the method of software reliabilitytest based on it. Firstly, a method to build Markovusage model based on improved state transitionmatrix (STM), which is a table-based modelinglanguage, is proposed. Then a software reliability testmethod including test case generation and testadequacy determination based on Markov usagemodel is presented. An improved Kullbackdiscriminant is selected as the judgment criteria ofconvergence from the test chain to the usage chain inorder to measure whether the testing process issufficient. Finally, a Markov test cases generation toolMTCG is developed which implements the methodsput forward in this paper. Experimental verificationof test adequacy and efficiency is made through acomparison between the Markov usage model-basedmethod and a completely random test method, theresults of which show that software reliability testbased on Markov usage model is approving,high-efficient and promising.