Energy Efficiency Optimization Task Scheduling In Collaborative IoT Environmen

In the collaborative Internet-of-Things(C-IoT)environment,which is built upon edge computing technology,the computation tasks of a terminal device can be distributed over severs or other terminal devices with available resources for execution or be processed locally,for the purpose of addressing the issue of limited computing resource on terminal devices.As a critical procedure to ensure the quality-of-service(QoS)of C-IoT,task scheduling seeks for the optimal assignment of limited computing resources in the C-IoT system that can reduce the processing delay and energy consumption to the most extent and eventually deliver high-quality computing services.Considering the increasing number of terminal devices,the design of C-IoT task scheduling strategies are facing more challenges.Aiming at the characteristics of the C-IoT environment,this thesis takes into account device-to-device(D2D)-enabled task offloading mechanism as well as the limited battery capacity of terminal devices,and formulates the energy-efficient task scheduling model for minimizing the flowtime of all computation tasks under the terminal device’s energy constraint.Considering the complex scenario with the increasing number of terminal devices and computation tasks,as well as the limitations of traditional task scheduling algorithms such as the simplicity of heuristic rules and the difficulty in obtaining high-quality feasible solutions,this thesis designs a series of metaheuristic algorithms based on the optimization principle of bat algorithm to solve the above-mentioned task scheduling problem.Through the enhancements of the main operators in bat algorithm,the algorithms in this work can produce high-quality energy-efficient scheduling solutions.According to the formulated energy-efficient task scheduling model in the C-IoT environment,this thesis proposes metaheuristic scheduling methods based on the improved bat algorithm.First,this thesis designs a population updating method that employs a velocity control strategy to update the positions of bat individuals.In addition,a position-based mapping operator is used to efficiently convert individuals into task sequences,based on which highquality feasible solutions can be obtained while satisfying the energy constraint of terminal devices.Next,a greedy search-based task allocation strategy is developed for calculating the flowtime obtained by the task sequence,which is regarded as the fitness value of the corresponding individual.In order to further improve the solution space exploration capability of the scheduling algorithm,this thesis further proposes a new scheduling algorithm that incorporates local search and Levy flight strategies.On the one hand,considering the fact that the mapped solution is highly dependent on the initial best solution in the position-based mapping operator,local search strategy is introduced to produce a high-quality initial best solution.On the other hand,to address the limitation of easily falling into local optima in metaheuristic algorithms,this thesis introduces the Levy flight strategy that involves more randomness to benefit the metaheuristic algorithm from escaping local optima.This thesis analyzes the global convergence of the improved algorithm theoretically.The above-mentioned two improvements substantially enhance the quality of energy-efficient scheduling solutions and further reduces the flowtime of the C-IoT system.This thesis designs exhaustive simulation experiments to fully evaluate the scheduling performance of the proposed scheduling algorithms.Experimental results demonstrate that the methods proposed in this work can effectively reduce system flowtime under the energy constraint.Based upon these contributions,this thesis designs and implements a visualized task scheduling systems oriented toward C-IoT environments.This system is capable of providing users with the flowtime-minimization scheduling solution according to user-specified task characteristics and mobile device’s energy limit.The series of algorithms and system implemented in this work can be applied to practical scenarios,such as industrial Internet of Things and smart cities,and are of great significance to improving system energy efficiency and reducing operating costs.