A Genetic Algorithm with Improved Convergence Capability for QoS-Aware Web Service Selection

Genetic algorithm (GA) is a good service selection algorithm to select an optimal composite plan from many composite plans. Since the execution of GA relies on a randomly search procedure to seek possible solutions, bad convergence of GA is produced by random sequences generation. To improve the convergence of GA for Web service selection with global quality-of-service (QoS) constraints, chaos theory is introduced into GA with the relation matrix coding scheme. The chaotic law is based on the relation matrix coding scheme. During crossover process phase, chaotic time series are adopted instead of random ones. The effect of chaotic sequences and random ones is compared during several numerical tests. The performance of GA using chaotic time series and random ones is investigated. The simulation results on Web service selection with global QoS constraints have shown that the proposed strategy based on chaotic sequences can enhance GA´s convergence capability.