Research On Latency Optimization Scheme For Dynamic Resource Allocation In Edge Computing

The Internet of Things connects things through the Internet,and establishes the interconnection among users,devices,and servers.The devices upload real-time data to the data center which could meet the requests of users in communication.However,with the increasing growth of information data,it is difficult to meet users' latency-sensitive requests only by a powerful data processing center,and it brings a huge computing pressure to the data center.Edge computing has become a new computing paradigm to relieve the computing pressure of data centers,and bring safer and faster services to users.Edge computing helps local servers or cloud centers release workload during busy hours,greatly speeding up network transmission efficiency.With the computing capabilities of network edge and device terminal,edge computing realizes the function of data collection,processing,analysis and decision making.Edge computing avoids the delay disadvantage of core network congestion,which can greatly reduce the delay of data processing.Traditional communication networks mainly focus on communication capabilities,whose core function is information transmission.However,the resources of edge computing is limited,and the users location,node capacity,task size,equipment,and the server communication distance is dynamic.It is a noteworthy problem that how to mobilize the full coordination of limited resources and meet the mobility of users,devices,and edge servers as well as bring services with optimal delay.Aimed at the dynamic resource allocation delay optimization scheme in edge computing,the main contributions of this thesis are as follows:1.On the basis of the existing IoT-Cloud framework,an Edge-Cloud framework suitable for edge computing is proposed.The framework divides the communication situations related to edge computing into three categories according to the judgment distance.The total cost of each categories is analyzed to provide schemes in a communication scenario.2.According to the Edge-Cloud framework,the edge computing delay optimization scheme for dynamic resource allocation in the MEC field is discussed,and a resource allocation scheme based on artificial potential field in mobile edge computing(RAAPF)is proposed.RAAPF provides an adaptive resource allocation scheme under the condition of trade-off between distance and node capacity.The problem is transformed into a knapsack problem to optimize the waiting latency,and a simulated annealing algorithm is applied to solve the optimal solution.3.According to the Edge-Cloud framework,practical edge computing service architecture applicable to IoT-based applications with dynamic cost evaluation schemes(PECSA)is proposed.In this framework,edge network devices and edge platforms cooperate with each other.The user service with the minimum delay,the user service with the minimum price,and the user service with the minimum price with given time are realized respectively,to improve trust effectiveness of management methods.By evaluating the computing cost and runtime resource capacity of the edge network,all edge computing devices and cloud resources to achieve faster response to edge network demands are reasonably coordinated.