Research Of Adaptive Software Testing Strategy Based On Cybernetics

With the development of computer technology, computer application with software system as the core has penetrated into all walks of life and plays an increasingly important role in the industry, national defense, as well as people’s daily lives, followed by people’s increasingly attention on the software quality and the low productivity of software products. In the field of software engineering, many techniques and methods to improve software quality have been proposed and used. However, software testing has always been the main way to ensure the quality of the software.Currently, the test cases selecting technology has become an important research direction of software testing. A good test suite can greatly reduce the testing effort thus to reduce testing costs and improve the efficiency of software development. Therefore, it is of great significance to in depth study software testing strategy which is used to guide the test cases selection. While there is a lot of research in this area, there are still two issues worthy of further study:one is how to design an optimal test strategy with a given test target, another is how to use feedback mechanism for software testing systematically. The software cybernetics is proposed to solve these two problems.Using feedback control thinking to solve the problem of software testing, software testing method based on software cybernetics regard the software under test as a controlled object and test strategy as related controller, which constitute a closed loop feedback system. On the basis of the existing research, this paper carried out an in-depth research on cybernetic methods of software testing, and mainly includes the following aspects:(1) Markov decision model of software testing. Aim at the defects of existing Markov decision model, this paper, by making a series of new conversion of conditionality, builds a new Markov decision model with demand coverage test as stopping criteria. This model is more similar to t the actual software testing process and has higher practicality. (2) The feedback control law of software testing. Through researching and analyzing in the relationship of software testing effect and some feedback information produced in software testing process e, this paper designs a controller (or test strategy) with the concern requirements coverage, defect detection capabilities, critical degree and return of test cost of test cases as its four restraining factors.(3) Parameter estimation algorithms and corresponding adaptive testing strategy. Based on optimal test objective and the designed controller, this paper designs an algorithm used to estimating test case selecting probability based on concern requirements coverage and an algorithm used to estimating the return of test cost based on the test results. At the same time, this paper proposed a test case selection methods based on the greedy algorithm. During the testing, it using the greedy algorithm to adjust the testing strategy online according to the parameter estimation results. And thus to achieve the software adaptive testing.Finally, this paper does a series of simulation experiments to verify the contribution made by each restraining factor of controller in the process of the achieving the test objective and to compare the adaptive testing strategy with the random testing strategy. The experimental results show that the adaptive testing strategy proposed in this paper is effective and is able to achieve the testing goal of detecting more serious software defects as early as possible.