Computing Task Scheduling Method In Cloud-edge Fusion Environmen

By deploying computing nodes at the edge of the network which is close to mobile devices,edge computing can offload computation tasks from mobile devices to edge servers,in order to avoid the transmission of data from the edge of the network to the cloud.In this manner,the heavy burden on cloud datacenters can be reduced.However,due to limited resources,it is difficult for edge servers to meet the deadline requirements when processing computationintensive tasks.Therefore,the cloud-edge computing paradigm has drawn extensive attention.In the situation that the edge server does not have enough resources to process the computation tasks,the tasks will be further uploaded to the cloud datacenter for execution.Therefore,designing high-quality task scheduling algorithms oriented toward cloud-edge environments is of great importance to improving the energy efficiency and quality-of-service of cloud-edge systems.Focusing on a cloud-edge system consisting of mobile devices,edge servers,and the cloud datacenter,this thesis establishes a task scheduling optimization model under a deadline constraint.The scheduling model targets for minimizing the total energy consumption of the cloud-edge system,by using a task sequence to represent the scheduling solution.Considering that the formulated optimization model is an integer program,we design an effective metaheuristic scheduling algorithm to solve it and obtain an energy-efficient task scheduling solution while satisfying the deadline constraint.In addition,we further improve the fundamental operators in the metaheuristic search procedure to enhance the computational efficiency and solution exploration ability of the scheduling algorithm.Furthermore,we provide a theoretical analysis to justify the convergence of the algorithm and the reasonability of parameter tuning.To be specific,we propose a metaheuristic scheduling algorithm based on the optimization theory of firefly algorithm.The proposed algorithm uses a position-based mapping scheme to convert each individual firefly to a valid scheduling solution.Then,we design a heuristic rule for the converted scheduling solution to calculate the energy consumption required for task offloading,such that the best solution can be identified and updated iteratively according to the population updating strategy.Next,we improve the scheduling algorithm from the perspectives of computational efficiency and solution quality,respectively.On the one hand,we introduce a linear movement strategy to reduce the computational burden during the position updating procedure.On the other hand,we incorporate the idea of simulated annealing,in which a solution of relatively lower quality also can be accepted as the current best solution with a certain probability,to help the metaheuristic algorithm avoid falling into the local optima and consequently to improve the algorithm’s solution exploration ability.For the purpose of verifying the reasonability and effectiveness of the above-mentioned strategies,we further provide a convergence proof to demonstrate in theory that the proposed algorithm is capable of converging to the global optimal solution.Moreover,we provide a movement trajectory analysis to determine the reasonable interval values for associated parameters,ensuring that the algorithm would not get stuck in local optima due to boundary traps.We perform extensive simulation experiments to evaluate the performance of the proposed algorithm when solving task scheduling problem in cloud-edge environments.Experimental results show that our algorithm can produce more energy-efficient scheduling solutions while significantly reducing the computation time.Based on the research contributions of this work,we develop a visualized task scheduling system oriented toward the cloud-edge environment.This system is responsible for obtaining the most appropriate assignment of tasks onto the edge server or cloud datacenter according to task characteristics and user-specified deadline constraints,minimizing the total energy consumption and guaranteeing the quality-of-service of the cloud-edge system.