Research On Particle Swarm Algorithms For The Web Service Selection Problem

With the proliferation of the cloud computing and the Software as a Service(SaaS) concept, the main form, operation mode, production mode and using mode of software system in the internet environment is undergoing tremendous changes. Loosely coupled distribution application system constructed by service reuse and dynamic aggregation has become the important trend of network software development. The process of service aggregation is implemented by binding between the service ontology and a concrete service, and service selection is directly related with the overall quality of service composition and that whether the binding relationship is needed to dynamic adjust. So the research on this problem has always been the focus of attention.More and more web service with the same functionality but different non-functionality attributes has been offered as the number of service on the web increased exponentially. How to select a better qulity serve, which can run stably and meet the needs of user, from large size service sets has become an urgent problem to be solved. In addition, there are multiple service levels in many service systems, and the existing research mostly aims at a single service level, and there are few reasearches considering multiple service levels simultaneously, so how to select a service composition, which can meet the multiple SLA constraints at the same time make sure its overall quality optimal, is needed to make further research. For the above problems, this paper will make some researches from the flowing aspect:the Business oriented service selection, the function oriented service selection, non-sharing resource oriented multiple SLA service selection and sharing resource oriented multiple SLA service selction. In recent years, it has become evident that the concentration on a sole metaheuristic is rather restrictive. A skilled combination of concepts of different metaheuristics, a so called hybrid metaheuristic, can provide a more efficient behavior and a higher flexibility when dealing with real-world. As a effective metaheuristic, algorithm PSO has been successfully applied to solve many problem in various areas. Consequently, according to the different situations of the service selection problem optimization model above mentioned, this paper tends to solve these problem with hybrid PSO algorithms and experimently verify the effecitiviness of these algorithms, including:(1) Through the research on business-oriented service selection problem, a single-objective particle optimization model for this problem is built, and a hybrid HEU-PSO algorithm is also proposed by combining the HEU's local search with algorithm PSO in this paper. In this algorithm, it combines the PSO's global searching capacity with HEU' local search capacity, and some proming areas are found firstly by algorithm PSO, and then the deep search will be conducted based on these areas; so algorithm HEU-PSO will make a global and deep search of problem's solution space. The experiment reveals very encouraging results in terms of the solution quality and the processing time required.(2) Through the research on large scaled function oriented service selection problem and after a single-objective particle optimization model for this problem is been rebuilt, a novel efficient hybrid ACO-PSO algorithm is proposed by combining algorithm ACO and algorithm PSO in this paper. In this algorithm, a a-domination service skyline query process is used to filter the candidates related with each service class, which can greatly shrink the search space. Then, varying dynamic construct graph is designed to guide the ant search directions based on a clustering process. By the combination of the ACO's flexible search property and the HEU-PSO's deep search property, it will make a fast and effective search on solution space. The experiement results show that algorithm ACO-PSO has excellent performance on problem solving.(3) Through the research on SLA-aware service composition problem (SSC) from non-resource sharing angle, an multi-objective discrete optimization model for this problem is built, and a hybrid Multi-objective Discrete Particle Swarm Optimization algorithm (HMDPSO) is also proposed by incorporating mutation operator into algorithm PSO in this paper. According to the characteristic of this problem, a particle updating strategy is redesigned and a particle mutation strategy is proposed, based on the introduction of the swarm diversity indicator, to increase the swarm diversity. In addition, an improved algorithm HMDPSO+is also proposed by incorporating a local search strategy based on constraint domination into algorithm HMDPSO, to further improve the algorithm performance. The experiment results show that algorithm HMDPSO+has powerful problem solving capacity.(4) Through the research on sharing resource SLA-aware service composition problem (SSC) from resource sharing angle, this problem been modeld as a resource sharing optimization model, and a resource sharing Multi-objective Particle Swarm Optimization algorithm is also proposed in this paper. According to the characteristic of this problem, the particle position and particle deployment strategy are defined to reflect the sharing relationship of the same concrete service in different SLA service composition; and continue to use the traditional particle update strategy to make global search; a local search strategy is proposed to improve search accuracy; a particle mutation strategy is proposed to restrain particle swarm's premature convergence. The experiment results indicate that algorithm SMOPSO can solve this problem effectively, and it has powerful search ability and excellent convergence property.