| In recent years,mobile terminal devices such as smartphones and tablets have gained tremendous popularity.Although the related technologies of mobile terminal devices have been greatly improved in the past few years,the use of mobile terminal devices is still restricted due to the low processing capacity,limited storage capacity and relatively short battery life,which afects the quality of experience(QoE).Ad Hoc cloudlet consists of a number of mobile terminal devices with different performances,and each mobile terminal device has the capability to serve both as a server and,at the same time,a client.This allows different mobile terminal devices in the cloudlet to share computing resource,which can reduce the impact of time delay,energy consumption and other factors on the user's QoE.This thesis focuses on assigning tasks in Ad Hoc cloudlet efficiently,studying the task assignment algorithm,developing the experimental verification system in Ad Hoc cloudlet,and analyzing the proposed algorithms by comparing with other task assignment algorithms on the experimental verification systemThe main work of this thesis is as follows:(1)The applications of Ad Hoc cloudlet are introduced and its unique advantages over traditional cloud are illustrated.A task assignment model in Ad Hoc cloudlet is proposed.Meanwhile,both advantages and disadvantages of task assignment in Ad Hoc cloudlet are analyzed.(2)A task assignment algorithm based on particle swarm optimization and simulated annealing(PSO-SA)in Ad Hoc cloudlet is proposed to minimize the energy consumption and time delay of all mobile terminals as well as balancing the fairness of resource utilization of all mobile terminals.Firstly,particle swarm optimization(PSO)is used to update the position and velocity of all the particles in the population,and then the simulated annealing(SA)optimization is used to avoid falling into the local optimal solution.Aiming at the existence of the dependencies between tasks,a model of dependency based on directed acycline graph(DAG)is proposed and a task assignment algorithm based on directed acyclic graph and PSO-SA(PSO-SA-DAG)is proposed to solve the problem of optimal assignment when there are dependencies between tasks.The proposed PSO-SA and PSO-SA-DAG are validated by computer simulation.(3)The experimental platform of task assignment in Ad Hoc cloudlet is built.The event management part allows the devices to communicate with each other and initiate a task assignment request.The cloud management layer is used to manage the joining,exiting of nodes in the cloudlet.In the decision distribution layer,the proposed PSO-SA is implemented.The device management layer can obtain the information of the node by interacting with the operating system.(4)Based on the platform of task assignment experiment in Ad Hoc cloudlet,the proposed algorithm is validated and compared with other heuristic task assignment algorithms based on computer simulation results.The results show that compared with other heuristic algorithms,PSO-SA algorithm can reduce more delay and energy consumption in Ad Hoc cloudlet when the amount of tasks increases or the number of mobile terminals increases. |
