Research On Ant Colony Optimization Algorithm-based Grid Services Composition

With the further development of Internet technology and Grid, support forcross-boundary, cross-industry, cross-platform business collaboration made it possible.The Grid service, a new network application model, can obtain a rapid developmentbecause of many extraordinary properties such as highly interoperability, inter-platform,loose coupling and so on. How to dynamically combine the existing various grid serviceto a new more complicated value-added services that can satisfy the needs of users, hasbecome a new hot spot research and the front edge. And the grid service compositionalgorithm is the key to determine the whole grid system performance. Hence, how todesign a high efficient and simple service composition algorithm is one of the mostimportant issues in the grid service composition technology. In this paper, a new servicecomposition method based on ant colony algorithm was proposed. Ant colony algorithmis analyzed and optimized and be applied to the grid service composition method. Afterthe improvement not only the time complexity is decreased but also the effect of thealgorithm is improved. The two major innovative works of this thesis are as follow:(1) The dissertation proposes a new ant colony optimization algorithm. Thealgorithm improves the searching for optimal solution and speeds up convergence rateby introducing a LionKing competition parameter. In the meantime, it can efficaciouslyavoid stagnation of the search through the range of possible pheromone trails on eachsolution component is limited to a maximum-minimum interval base on MMAS(Max-Min Ant System).(2) This paper puts forward a method of grid services combination base on antcolony optimization algorithm. It connected ant colony optimization algorithm withLionKing competition parameter with service composition technology and it hasanalyzed, designed and studied on abstract model of ant, service quality evaluationmodel, definition of pheromones and initialization, combining algorithms and so on.With the application of obvious advantage that ant colony algorithm is easy to solve theproblem of generating alliance, the design efficiency can be improved.