Software reliability accelerated testing based on the combined testing method

Abstract

For increasing the effectiveness and efficiency of the software reliability testing, a two-stage combined software reliability accelerated testing is proposed. The combined testing method combines representative testing and directed testing, and it is able to test some operations or program modules with low operation frequency to satisfy the coverage criteria if the number of the test cases is the same as the conventional reliability testing. Transition rule and stopping rule which employed in the testing method are also discussed. Important sampling is used to get the accelerated factor. A novel approach to software reliability prediction is presented, which gives more accurate estimation than the classic method when the accelerated testing is applied during the software reliability growth testing. The combined testing method has certain advantages over ordinary software reliability testing verified through the simulation experiment. Meanwhile the comprehensive performance of the approach to software reliability prediction is evaluated.

Shortening the Cycle of Software Reliability Testing with Fuzzy Discrimination

“…Many accelerated testing methods have been proposed to improve the efficiency of SRT, for example, accelerated testing methods based on the Markov chain usage model [2], program transformation [2,3], the Musa operational profile [4,5], mixed testing [6,7], and other accelerated testing methods [8,9]. Most of these methods focus on modifying the software reliability models or changing the operational profiles that detract from the already precarious confidence that industries currently have toward the quantitative evaluation of software reliability.…”