An Improved Simulated Annealing Genetic Algorithm for Workflow Scheduling in Cloud Platform

This paper proposed a multi-objective optimization algorithm combining simulated annealing algorithm with the genetic algorithm concerning makespan and reliability. Firstly, this algorithm makes tasks prioritization considering their influence degree tasks schedule policy to generates a suitable initial population. Secondly, for individuals which produced through crossover and mutation genetic operations to perform simulated annealing operation, in order to avoid the premature convergence problem caused by random algorithm. Then, it introduced failure rate in mutation phase to increase the reliability of the scheduling results. Experimental analysis indicates that the algorithm in solving the problem of premature convergence and result unreliable better than GA.