A Research On Edge Computing Resource Scheduling Based On Swarm Intelligence Algorithms

With the rapid development of 5G technology and the Internet of Things industry,smart mobile devices and emerging technology applications(such as augmented reality,intelligent transportation,etc.)are in a booming state.All of these emerging applications place higher demands on latency,computing resources,and energy consumption.However,mobile device has to balance its mobility and portability,its hardware configuration is very limited to meet the computing resource requirements of the application.The emergence of edge computing technology makes up for the lack of computation resources of smart mobile devices.At the same time,edge servers are deployed close to users and can provide edge computing services with lower latency.Therefore,it has become the focus of research in recent years.At present,the main research work of edge computing is how to formulate a more scientific and reasonable calculation offloading decision and resource allocation scheme to reduce the energy consumption and delay.However,transferring computing tasks from mobile devices to edge servers for processing can reduce device latency and energy consumption,but it will bring additional network transmission energy consumption and server processing energy consumption.Few studies use these two additional energy consumption as an optimization target.For enterprises that provide edge computing services,they need to increase the service life of the servers and extend their usage time.Based on the above situation,this thesis mainly does the following work:(1)This thesis constructs a resource scheduling system model of multi-user devices single edge server.The model takes reducing server energy consumption as one of the optimization goals,and takes computing resources such as storage,maximum energy consumption,and CPU cycles of edge servers as constraints.And the total energy consumption related to energy consumption of mobile devices,data transmission energy consumption and edge server computing energy consumption is formulated.Then,this thesis propose a Level-Based Learning Swarm Optimizer is proposed to solve this problem.Using the global search ability and convergence speed of the particle swarm algorithm,the optimal resource allocation scheme can be found more quickly and stably,and the total energy consumption of the system model can be effectively reduced.Through comparative experiments with other swarm intelligence algorithms,it is found that the proposed algorithm can more quickly obtain the optimal solution of resource scheduling that satisfies the constraints and has lower energy consumption.(2)On the basis of the previous work,This thesis constructs a resource scheduling system model of multi-user devices multi-edge servers.the smart mobile devices can select the most suitable server to help itself complete the computing task according to the requirements of its own computing tasks and the computing resources of the server.Aiming at the problem of uneven distribution in solving the multi-server system model using the Level-Based Learning Swarm Optimizer,an improved Level-Based Learning Swarm Optimizer algorithm is proposed.Combined with the crossover operation of genetic algorithm,it is more suitable for solving multi-server system energy consumption optimization problem.Finally the superiority of the algorithm is proved by experiments.This thesis takes the energy consumption generated by the processing data of edge server as one of the optimization objectives and proposes edge computing system models to jointly optimize the total energy consumption of smart mobile devices and edge servers and proposes a Level-Based Learning Swarm Optimizer algorithm to optimize the problem.Aiming at the problems of this algorithm in optimizing the energy consumption of multi server system model,this thesis combines particle swarm optimization algorithm,genetic algorithm and hierarchical learning strategy and proposes an improved Level-Based Learning Swarm Optimizer algorithm.The experimental results show that the two algorithms proposed in this thesis can get the lowest energy consumption under two different models.