Evolutionary Approaches For Software Testing Resource Allocation

Software testing is an important phase in software development process.In recent years, as modern software systems have been expanded continuously, the problem of how to design reliable software systems attracted lots of attention. The software testing process is the most important process to improve the system reliability. As an optimal allocation of testing resource may lead to significant improvement of system reliability, the Optimal Testing Resource Allocation Problems (OTRAPs) is well worth studying. There are two major objectives in the OTRAPs:reliability and cost. In the past decades, researchers mainly studied OTRAPs wth the former objective. However, the software production time and the testing resource are limited. Hence designers pay more and more attention to reducing the cost.In this paper, we have studied OTRAPs with the goal to minimize the testing cost. In previous work, the best way to solve this problem is genetic algorithms. We adopted genetic algorithm and simulated annealing algorithm to solve OTRAPs, but experimental results showed that the performance of these two algorithms on the selected model has yet to be improved. Hence, in this paper we proposed a new algorithm based on genetic algorithm and local search strategy. After that we apply the negative correlation search algorithm on the issue, by considering the exchange of information between individuals, increasing the diversity of the population, and thus a good solution to the multimodal problem. Experimental results show the software testing resource problem can be solved better by using the algorithms we propose.