Research On Computation Offloading Strategy In Fog Computing Distributed Wireless Network

With the large-scale commercialization of 5G,the era of "Internet of Everything" has arrived,but the current transmission latency of cloud computing is too long,which is not friendly to latency-sensitive applications.In order to solve this problem,Cisco has proposed the concept of fog computing,which has the advantages of flexible deployment,various access methods and support for inter-node communication compared with other edge computing implementations.First,three system implementations of edge computing: Cloudlet Computing,Multi-access Edge Computing and Fog Computing and their application scenarios and system architectures are introduced and compared in terms of both system architectures and request processing process,and the selection strategies for the three edge computing implementations are summarized.Then,the model of the fog computing distributed wireless network system architecture is developed to combine fog computing technology with 5G mobile networks.The architecture is divided into three layers,namely the local computing layer,the fog computing layer and the cloud computing layer,and the three computing layers correspond to three computation offloading strategies,where the fog computing layer is further divided into the fog computing service layer,the fog computing service scheduling layer and the infrastructure device layer.In addition,the request process of computation offloading is studied,a mathematical model of the architecture is established,and the optimization problem of the overall cost of computation offloading is proposed.After that,an optimization algorithm for joint computation offloading,data compression,energy harvesting and application scenarios(JCDEA algorithm)is proposed,based on a random coordinate shrinking classification algorithm,which transforms the problem of solving the computation offloading strategy into an optimization problem of solving the overall cost of computation offloading,and then,by controlling the variables,into a sub-optimal solution problem of solving the local computing cost,the fog computing cost and the cloud computing cost respectively,and finally,the sub-optimal solutions of the computation offloading policy are obtained.Several sets of related simulation experiments are also carried out to obtain the cost variation curves of user equipment in local computing,fog computing and cloud computing under the specific situation of real base station distribution,and to analyze the relationship between the amount of offloaded data,the number of users and the compression ratio with the overall cost of computation offloading and the utilization of computing resources.Finally,the relationship between data rate on overall cost and resource utilization in different application scenarios is investigated.The advantages of joint optimization algorithms in different application scenarios are explored,and the relationship between data rate and overall computation offloading cost in three application scenarios,as well as the relationship between data rate and fog-cloud computing resource utilization,is analyzed.A certain theoretical reference is provided for the future large-scale deployment of fog and cloud servers.