Research On Application Mapping Under Edge Cloud Collaboration Based On Particle Swarm Optimization Algorithm

In the era of “Internet of Everything,Comprehensive perception”,diversified and complicated terminal applications occupy the mainstream market.Due to the rapid development of the Internet of Things technology and the large-scale application of terminal application services,the traditional Internet architecture model is facing severe challenges of big data and overloaded computing.In this case,edge computing came into being.Although cloud computing has powerful computing performance,sometimes it cannot finish terminal application services in time.As a supplement to cloud computing,edge computing can handle tasks with high delay sensitivity close to the terminal device that generates data,reduce the transmission time of application data,and effectively reduce the high load of cloud computing.Besides,data that doesn't need a long-term backup is stored in the edge node,which can solve the problem of massive redundant data storage in the cloud and protect the security and privacy of the tasks.However,the current edge computing network architecture still has problems which cannot be ignored.(1)With the continuous increase of terminal applications,each terminal application will face the problem of resource device selection,that is,how to balance the relationship between link transmission time and application calculation time.(2)Due to the load of the resource equipment and the communication time between the resource equipment greatly affect the overall architecture power consumption and application delay time,a reasonable application module deployment become the key to determine the optimization of network architecture energy consumption and terminal application delay in a fixed network architecture.Faced with the end-users' requests for high-quality application services and low power consumption of the network architecture,this paper optimizes the network architecture combining cloud computing and edge computing first.By analyzing the network architecture in academic research,it finds advantages and disadvantages of the original architecture,and combines a variety of network structures to create a new edge cloud collaborative network architecture.Then,in a new type of "edge-cloud collaboration" heterogeneous network environment,faced with the application of stepby-step and sequential execution,the use of dynamic task scheduling algorithm reduces the processing time of periodic tasks by sharing resources with neighboring node devices,thereby further reducing the overall application delay.Before the terminal device sends a service request,this paper proposes a mapping scheme between the application module and the basic resource device based on the improved particle swarm optimization algorithm,which can efficiently complete the global search in the solution space and quickly converge to obtain the global optimal solution.and fully consider the two important factor of task delay tolerance and system power consumption.The improved particle swarm optimization algorithm considers the two important factors of task delay tolerance and system power consumption.The obtained optimal application placement strategy can achieve the trade-off between power consumption and delay.This paper uses i Fog Sim simulator to carry out simulation experiments.First,build a new type of edge cloud collaborative network architecture.Secondly,the application mapping strategy and dynamic task scheduling strategy proposed in this paper are added.Compared to fixed application module placement scheme and static task scheduling strategy,application module placement strategy based on improved particle swarm optimization algorithm and dynamic resource allocation and task scheduling algorithm can improve the quality of application services and reduce the power consumption of network architecture significantly.Experimental results show that the average total delay of terminal applications is reduced by 14.6% when using dynamic task processing strategies.After using the improved particle swarm optimization algorithm to obtain the optimal application module placement solution,the average total delay of the terminal application is reduced by 25.3%,and the overall architecture power consumption is reduced by 8.4%.This shows that the application placement strategy and dynamic task processing strategy proposed in this paper can optimize delay and power consumption at the same time.