Research On Task Offloading And Load Balancing Schemes In Distributed Fog Computing

In order to meet the requirements of mobile communications for the development of new services such as the Internet of Things(Io T),virtual augmented reality(VR \ AR),and driverless vehicles,5G will use various new technologies to meet these business types and application scenarios.The cloud computing paradigm provides end users with a variety of high-quality services with its powerful computing and storage capabilities.However,with the increasing amount of terminal equipments,the core network traffic is increasing exponentially.In 2020,mobile traffic will increase to 291.8 exabytes.Such a large amount of traffic will bring huge challenges to the core network.To solve this problem,edge computing that extends computing and storage from the network center to the edge of the network becomes an effective solution.As an implementation of edge computing,fog computing aims to connect all devices(including terminal devices,Io T devices,sensors,routers,gateways,etc.)through heterogeneous networks,so as to provide a new solution for the Internet of Things.How to achieve efficient collaborative computing of various nodes in fog computing is an important research direction in fog computing.This thesis studies the collaborative computing methods of various fog nodes in distributed fog computing,and conducts in-depth research on the tasks offloading and load balancing.The main work is as follows:Aiming at the distributed collaborative computing problem in the fog computing network,this thesis proposes a solution based on dynamic programming.When allocating computing resources for each user,the communication capabilities and computing capabilities of each fog node are considered comprehensively.Impact of other users,minimizing total service latency for all users.Aiming at the task offloading in the fog computing network,this thesis considers the wireless transmission delay between fog nodes in the wireless fog access network,the task calculation delay of the fog nodes,and the wireless transmission delay between the fog node and the remote cloud data center.The optimal offloading scheme is calculated within the acceptable delay threshold.Simulation results show that the multi-tasking offloading scheme based on dynamic programming proposed in this thesis can effectively reduce service delay compared with single fog system and resource pre-allocation system.The resource utilization rate is much higher than the resource pre-allocation scheme.Aiming at the load balancing problem in distributed fog computing networks,this thesis proposes a dynamic load balancing scheme based on resource prediction.When performing load balancing for each task,the task execution time,task execution cost,and available memory of fog nodes are considered.factor.Fully improve the effective utilization of distributed fog clusters and reduce response delays.The simulation results show that,compared with the static load balancing scheme in the scenario where a large number of tasks are processed and the system network is in poor condition,the proposed scheme has a great advantage in response time.