Preferential Choice Scheme Of Service Provider Based On Genetic Algorithm In Migrating Workflow System

The migrating workflow is a new direction within the workflow management area. In migrating workflow system based on mobile computing paradigm, the three elements are migrating workflow management engine, workstation and migrating instance. Management engine is in charge of the definition of workflow procedure, the generation of migrating instance and the coordination of migrating instances. Migrating instance executes the tasks. Workstation which refers to the anchorage server and working-machine network supports resource and place for migrating instance. The workflow is completed by one or more migrating instance, and migrating instance will execute task according to its itinerary graph between work stations which we called anchorage.As the operating platform for migrating instance, anchorage should have a good preferential choice scheme of service provider to improve system performance and resource utilization of computer group.According to conceptual model and reference architecture of the migrating workflow, this paper presents an algorithm to make the resource of computer group get load balancing by studying on preferential choice scheme of service provider based on genetic algorithm and Pareto optimal, introduces the Pareto genetic algorithm to optimize the multi-objective problems.Firstly, this paper builds a mathematic model to describe the preferential choice of service providers, and gets multi-object function. Codes the running status of migrating instance in migrating domain; gives out the concept of a fixed point in order to avoid the migrating instance gets out from its domain, which caused by the genetic algorithm; and we present the initial stocks, options, cross mutation operator which meet the migration flow characteristics to get an optimal solution.However, because of turning multi-objective problems to be a simple one, the result will cause many migrating instance to migrate again, that will increase the system's load. We use Pareto optimal algorithms for optimal the algorithm: first, use the Pareto optimal to instead the fitness function, and choose the choice operator to suit it, the operation of other operators remain unchanged. Then add a Pareto optimal solution sets and strategy to retain the elite, and propose the arithmetic for removing illegal solutions and selecting the solution among optimal set based on concepts of hamming distance and matching to ensure optimum results.Comparing with the experimental result of the Pareto genetic algorithm and the fuzzy overall decision-making algorithm, we can get that this algorithm could realize multi-resources get load balancing, improve performance and resource utilization...