Research Of Optimization Algorithms For Workflow Task Scheduling In Cloud Computing Environment

With the rapid development of information technology,as one of distributed computing technologies,cloud computing has become a research hotspot in the industry.It is a basic technical environment that supports the development of emerging technologies such as artificial intelligence and block chain.The cloud computing system consists of large pools of resources,so it should schedule tasks properly for the purpose of meeting specific service quality requests from users.At the present stage,the energy consumption of data centers are growing quickly with the high-speed expansion of data centers,and it will hinder the development of cloud computing.In the cloud computing environment,how to assign tasks to resources effectively is one of the main issues that service providers need to study.At present,task scheduling algorithms focus on studying single-objective optimization problem such as minimum time or cost,while less considers complex service quality requests from users.But service providers should research task scheduling from multi-objective so as to provide better services and get more profits.So this paper regards time,cost and energy consumption as targets of task scheduling.The main research work is summarized as follows:(1)Analyzing the architecture of the cloud computing and data centers for the complexity of task scheduling demands.In order to reduce the energy consumption of the data center,the dynamic voltage frequency adjustment technology is adopted.And the concept of multi-objective optimization is introduced.The system model of multi-objective task scheduling is established.(2)Multi-objective particle swarm optimization algorithm has the difficulty in balancing the exploration and exploitation,multi-objective workflow scheduling algorithm based on grid variance is presented.The algorithm improves grid coordinate system and comes up with grid variance.The Elite particle selection strategy is designed and applies to the bad particles self-learning strategy for enhancing the diversity of Pareto Front.Using grid variance to evaluate the diversity of current Pareto Front and dynamically adjust evolution strategies according to it,so the Pareto schedule set is both diversity and convergence.(3)For the sake of further optimizing the time of task scheduling,multi-objective task scheduling algorithm based on task clustering using NSGA-? algorithm is proposed.Considered the feature of paying on time in cloud computing,task clustering is designed so as to reduce the spending in communication.Coding the individual chromosome combined with the characteristic of tasks,analyzing and designing related genetic operators,and introducing dynamic crowding distance operator for improving the strategy of population diversity maintenance,so it can be better applied to the generation of schedule set.In order to verify the performance of the proposed algorithms,other multi-objective optimization algorithms are selected.According to WorkflowSim simulation platform,the experimental results show that multi-objective workflow scheduling algorithm based on grid variance has better performance on diversity and convergence indexes,and has a certain improvement in scheduling time,cost and energy consumption.Multi-objective task scheduling algorithm based on task clustering using NSGA-? algorithm has an obvious advantage in scheduling time,and the Pareto schedule set has better distribution.