Quantum-Inspired Hyper-heuristic Energy Consumption Management On Cloud Computation System

Cloud computing,which profit is the main purpose of the cloud service provider is a new business computing model,and the energy consumption is the main cost of cloud data center.Reducing energy consumption under the different service requirements of customer as far as possible not only could improve profits for the providers but also reduce the pollution to the environment.As a result,the research of the energy consumption management on the cloud computing will be a very meaningful topic.Dynamic Voltage Scaling is an effective energy saving technology.There are some studies which combine dynamic voltage regulation and task scheduling to improve the energy efficiency of cloud computing system.However,most of researchs can't meet the customer's different demands.In this paper,a task scheduling model that combine cloud characteristics(heterogeneous,on-demand service characteristics)and dynamic voltage adjustable factors based on the existing task scheduling model for distributed system is built.Task scheduling problem on cloud is a NP-hard combinatorial optimization problems,the random search technique based on evolutionary optimization get a better solution than traditional to solve the task scheduling problem based on DAG task graph.However,because of the diversity and complexity of optimization problem domain making these algorithms universality is poor and computational overhead is too high to meet the real-time demand calculation.Under the background that traditional swarm intelligence algorithm adopts design complex search operation to DAG scheduling problem,but not obtained the expected effect.We try to discuss the problem from a new angle,the reduced search strategy set based on domain knowledge is built,and hyper-heuristic framework is introduced to manager them to improve the convergence and universal of algorithm.In order to improve the hyper-heuristic algorithm,we put forward a quantum-inspired hyper-heuristic algorithm(QHA)that inspired by quantum computation theory.In QHA,the performance of the heuristic strategy is associated with quantum state,and the method of individual and collective learning is adopted to update the quantum state.Besides,a rapid individual assessment method is proposed to reduce the computational overhead.Simulation experiments show that QHA can manage the energy under the different needs of customer,and the QHA has less computational overhead,high convergence and universal compare with other classical algorithm.