Quantification of Software Code Coverage Using Artificial Bee Colony Optimization Based on Markov Approach

Abstract

Software test suite generation and development of techniques to optimize the test suite are vital parts of the software development life cycle. In this paper, a combination of Markov chain and artificial bee colony (ABC) optimization techniques are adopted to attain the software code coverage. Initially, dd-graph is captured from the control flow graph of the source code and is represented as a Markov chain. The number of paths is obtained based on linear code sequence and jump (LCSAJ) coverage. LCSAJ is used to reduce the number of independent paths as compared to paths obtained by basis path testing. Automatic test cases are generated and based on the operation profile of the test suite; transition probabilities are obtained using gcov analysis tool. Further, ABC optimization is adopted to ensure software code coverage. The initial population is randomly selected from the test suite and populated for subsequent generations using the ABC algorithm. The test cases are generated for three mixed data type variables, namely integer, float and Boolean. The quality of the test cases is improved during every iteration of ABC optimization and traversed through number of LCSAJ-based independent paths thereby ensuring software code coverage. Finally, software code coverage is quantified using the fitness/happiness value computed as a product of node coverage and the corresponding transition probability values based on the path covered.