Research And Implementation Of Genetic Algorithm Cloud-based Task Scheduling Strategy

Cloud computing since 2007 put forward, with its simple, inexpensive features has been an unprecedented development in all walks of life there is a huge market demand, will inevitably have an impact on people’s future life and work.After the commercial cloud computing, is intended for a large user base, their resource needs are many and various, how to meet the needs of all users of the case for the massive task of reasonable scheduling becomes very crucial.Genetic algorithm is a global search algorithm, compared to the heuristic search algorithm, which has the adaptability, learning and parallelism, especially in solving the massive task of genetic algorithm parallelism has great advantages, can after the split the task assigned to multiple processors simultaneously for processing. At the same genetic algorithm also has a scalable and can easily be combined with other algorithms, absorbing the advantages of other algorithms. Already scholars genetic algorithm to grid computing task scheduling, scheduling scheme to obtain a shorter task completion time. Cloud computing is based on grid computing evolved, and its basic framework is similar to grid computing, this paper genetic algorithm is applied to task scheduling cloud computing.Cloud computing task scheduling problem is a classic combinatorial optimization problem.In the process of genetic algorithm for solving scheduling problems often ask questions to grasp the accuracy is not high, making it difficult to search for the optimal solution.The main reasons for these problems are:in the case of large number of iterations to produce local optima often overlooked insufficient global search capability; you can not get over the initial population and so on.Directed to solving algorithm converges too slow problem, this traditional serial binary coding and coding scheme has been improved, using a task-resource encoding, the chromosome expression more direct and more to effectively solve the problem of excessive chromosome.For the genetic algorithm in the process of solving the objective function considering the limitations of the problem, this fitness function made the following improvements:The QoS constraints of the total task completion time, bandwidth, cost of these three conditions as the fitness function. To evaluate a fitness function by setting different parameters.Finally, using cloudsim platform to achieve the proposed task scheduling policy based on improved genetic algorithm. By demonstrating relevant experimental results, this method is feasible.