A Multi-objective Optimization Scheduling Method Based on the Improved Differential Evolution Algorithm in Cloud Computing

Cloud computing provides a large number of opportunities to solve large scale scientific problems. Task scheduling is important in cloud computing and attract a lot of attentions in recent years. To more efficiently scheduling the resources in cloud systems, this paper studies a novel multi-objective task scheduling problem which aims to Minimize the task’s Completion Time as well as to Minimize the Resource Payment (termed as MCT-MRP problem). However, the multi-objective optimization problem for task scheduling is generally an NP-hard problem. To efficiently solve the problem, this paper proposes an improved differential evolution algorithm. With adaptive parameter setting (control parameter F and the crossover factor CR) and an novel crossover operation and selection strategy, our improved differential evolution algorithm can solve the problems faced in traditional differential evolution algorithm such as premature convergence, slow convergence rate and difficult parameter setting. We have done extensive simulations. The simulation results demonstrate the efficiency and affectivity of our proposed algorithm.