Particle Swarm Optimization with Adaptive Mutation Algorithm for data flow testing
Abstract
The standard Particle swarm optimization (PSO) was motivated by the social and cognitive performance of swarm. PSO is a population-based stochastic search algorithm, which has shown a high-quality performance over many benchmark and real-world optimization problem. PSO also simply falls into local optima in solving multifaceted multimodal problems. This paper presents a developed hybrid algorithms for test case generation by introducing adaptive mutation in PSO. We have given name to our proposed algorithm as particle swarm optimization with adaptive mutation (PSOAM) to solve this problem. PSOAM take the instrumented version of program to be tested, def-use path list to be covered, swarm size as input and gives optimized number of test cases with better def-use coverage as output. However, PSOAM still suffer from premature convergence on some example programs. It suggests that only pure mutation techniques cannot avoid local optima. The proposed PSOAM is more successful than other a fore mentioned algorithms. Simulation results express that the proposed adaptive mutation strategy can efficiently improve the performance of PSO.