Research On Multi-path And Multi-hop Computing Offloading In Ad Hoc Networks

In the related research of cloud computing,edge computing is still a challenging problem.And providing high-quality edge computing services to users in remote areas is a more difficult challenge.Due to the lack of infrastructure in the remote areas,edge servers cannot be deployed and users lack communication measures,which makes it difficult to provide edge computing services for users.To solve the problem,the current researches consider to provide edge computing services for users by carrying servers on drones.However,the researches mainly focus on how to provide high-quality computing services for independent users.Few people have considered the edge computing problem of associated users.At the same time,due to the mobility of drones,the research on providing computing services for associated users is very challenging.And the research on computing services of associated users can provide better quality of service when users interact through edge computing.Therefore,this paper proposes an edge computing architecture based on the cooperation between drones with edge servers and drones without edge servers.Specifically,the computing task is transmitted by multi-path and multi-hop.This paper studies the architecture from the following aspects.(1)In this paper,we study a multi-path and multi-hop offloading method in static network firstly.Because the problem of computing offloading in dynamic network is complex,the problem of associated users in static network is considered.The computing offloading problem of associated users in static network is considered.The aim is to minimize the total delay of all computing tasks and the energy consumption of drones by optimizing the transmission and computing decision of each computing task.Firstly,it is modeled as a non convex nonlinear mixed integer programming problem.Then it is approximately transformed into a non convex multi-path and multi-hop task transmission problem,and is equivalent to a linear programming problem by the traversal algorithm proposed in the paper,so that the optimal solution of the approximate problem can be obtained.And the performance of the algorithm is proved.(2)In this paper,based on the method of offloading in static network,we study a multi-path and multi-hop offloading method in dynamic network.The research considers the problem of computing offloading of associated users in dynamic networks.Firstly,considering the movement of the network,the task transmission model of static network is modified,and a non convex nonlinear mixed integer programming problem related to time period is obtained.Then,the traversal algorithm in static network is used to solve the problem in each time period,so that the approximate solution of the problem can be obtained.And the performance of the algorithm is analyzed.(3)In this paper,we test and analyze the proposed architecture.The experimental results of dynamic network show that compared with the single-path and multi-hop offloading scheme,the energy consumption of all tasks is increased,but the total delay required is greatly reduced,and the overall performance of multi-path and multi-hop offloading scheme is greatly improved.Compared with multi-path and multi-hop random transmission scheme,the energy consumption and total delay are significantly reduced.The multi-path and multi-hop offloading scheme proposed in the paper can provide high-quality edge computing services for users in remote areas more efficiently,and prolong the network lifetime as much as possible.