Genetic Algorithm And Ant Colony Algorithm Based Energy-Efficient Task Scheduling

With the development of computer science and technology and computer network, cloud computing, as a new computing mode, with very high scalability and availability, quickly becomes the research hotspot of academia and industry. Due to the large amount of data and calculating that concentrated in the cloud, computing platform and data center provide high performance.Meanwhile, the energy consumption becomes larger and larger. High energy consumption in data center has brought many problems, including not only the waste of electrical energy, instability of the system, but also a bad influence on our living environment. Therefore, the high energy consumption in calculation center has become an urgent problem to be solved.This paper first reviews three main means of cloud computing to save energy: dynamic voltage adjustment technology, virtualization technology and closing-sleep technology. These three kinds of energy consumption optimization management technology have different application scenarios with different problems. Fundamentally speaking, high energy consumption is caused by unreasonable task scheduling. Therefore, research and design high efficient task scheduling algorithm is the best way to achieve the optimization of energy consumption and system performance. The main scheduling methods currently exist are the task scheduling based on the Agent, task scheduling based on cost, task scheduling based on Petri-net, heuristic algorithm for task scheduling and other algorithms.This paper starts from the heuristic algorithm, firstly we summarize the basic theory of genetic algorithm and ant colony algorithm, then on the basis of these theories we propose an energy-efficient scheduling algorithm based on genetic algorithm and ant colony algorithm. The main idea of the algorithm is by copying the tasks which are not assigned to the same node with their successors. By this way we can shorten the task execution time and reduce the communication energy consumption. The evolutionary rate of Genetic algorithm is decreased with the passage of time, but the ant colony algorithm is just the opposite. The lack of pheromone in early stage leads to a slower evolution rate, but latter on the speed becomes quicker and quicker. Another innovation of this paper is to design a dynamic fusion strategy. The genetic algorithm suspends and the ant colony algorithm starts at the optimal time. Compared with simple genetic algorithm or ant colony algorithm, this strategy can improve the performance. Finally, we do research and simulation to check the effectiveness, advantages and disadvantages of the new algorithm.The proposed dynamic fusion of genetic algorithm and ant colony algorithm shows good performance in the experiment and the simulation. Compared with simple genetic algorithm, this algorithm can improve the evolutionary time by10%. Once the scheduling algorithm is embedded into the cloud computing platform and determine the parameters needed for the actual problems,it can be applied in broader field.