Research On Joint Resource Allocation In Mobile Edge Computing

With the rapid development of mobile devices and wireless networks,the types of mobile devices are constantly enriched and computing power is constantly enhanced which makes the corresponding applications more diverse.Applications such as augmented reality,face recognition,virtual reality and so on all need to run on mobile devices which leads to high demand for the computing power.Correspondingly,the computing,storage and power resources of mobile devices are often inadequate.Therefore,mobile cloud computing and mobile edge computing emerge as an effective solution.Tasks are offloaded to cloud data centers or edge servers which makes use of the powerful computing power of servers to make up for the shortage of mobile device resources.As an improvement of mobile cloud computing,mobile edge computing reduces the computing power of data centers from the center to the edge of the network.By shortening the physical distance,it overcomes the problems of high latency of mobile cloud computing and high pressure of data centers which has attracted wide attention from industry and academia.With the development of 5G,Internet of Things,artificial intelligence and big data,mobile edge computing has become an indispensable supporting technology.Different from resource management in mobile cloud computing,mobile edge computing needs to consider the characteristics of close integration of computing and communication resources,and jointly allocates computing and communication resources to achieve the optimal allocation scheme.Therefore,there are a lot of research work on joint resource allocation,which has greatly improved the power efficiency and shortened the task delay compared with the traditional allocation method.With the further development of mobile devices and applications,on the one hand,the demand for low energy consumption and low latency has been increasing,on the other hand,many new problems have arisen in the process of development,and the disadvantage of optimizing resource allocation alone is beginning to highlight.For this reason,this paper aims at minimizing power consumption on the premise of guaranteeing user's quality of service.Aiming at the problem of high transmission delay of mobile devices which are far away from the mobile edge computing scenario and even can't be directly connected to the edge network,short battery usage time of mobile devices and extra power and time overhead caused by task offloading process.This paper models the practical scene and jointly optimize several key parameters such as computation,communication,offloading decision and so on under the new system architectures combining Ad Hoc network,wireless power transmission technology and edge caching technology,respectively.Finally,the problem is formulated and solved by proposed advanced mathematical optimization algorithms based on convex optimization and block coordinate descent.Firstly,aiming at the problem that the offloading delay of mobile devices is too high to connect directly to the base station,this paper regards the C-RAN environment as an implementation form of mobile edge computing,and studies the solution of combining wireless self-organizing network and D2 D under the C-RAN architecture.The long-distance devices communicate with the base station through multi-hop self-organizing network,which improves network bandwidth and reduces transmission latency.The transmission delay is transmitted.Considering the power limitation of mobile devices and the mobility of devices,this paper proposes a joint optimization scheme for the stable path of data transmission and the virtual machine responsible for computing,and finds the optimal joint allocation scheme to minimize the total energy consumption through heuristic algorithm.The simulation results show that the proposed joint allocation scheme has obvious advantages over the separate allocation of computing and communication resources.Secondly,in order to solve the problem that mobile devices can't work for a long time due to power constraints,this paper we optimize the resource allocation under the wireless power transmission combined mobile edge computing architecture.By the time division multiplexing,mobile devices harvest power firstly and then use the harvested power to complete tasks where local computing is executing all the time slot.In order to achieve the goal of maximizing power harvesting while minimizing power consumption,a problem of maximizing residual power is modeled,and an optimization scheme of joint optimizing of unloading ratio,power harvesting time,communication offloading power and calculated CPU frequency is proposed.Finally,aiming at the formal optimization problem,a method combining convex optimization and sequence unconstrained minimization is proposed to solve the problem.The simulation results show that the proposed joint allocation scheme can significantly improve the power efficiency,and the algorithm has less time overhead.Finally,aiming at the problem of extra power and time overhead caused by task offloading in mobile edge computing.This paper redefines the concept of task caching and proposes a mobile edge computing system model that integrates edge caching technology to reduce unnecessary duplicate data transmission where caching the tasks frequently performed in the local area of the edge server.In order to further improve power efficiency,the joint optimization problem of computing,communication,cache resources and unloading ratio in the environment of combining mobile edge computing with task cache is modeled.Finally,a solution method based on block coordinate descent method and convex optimization is proposed.The original non-convex problem is split into two parts and the optimal allocation scheme is obtained by alternate iteration.Experiments show that the proposed joint optimization scheme can achieve lower energy consumption while guaranteeing the user's quality of service,and the proposed solution method has a lower computational cost.Generally speaking,each chapter of this paper contains two levels of innovation: architecture,model and optimization algorithm.At the model level,on the basis of the architecture that integrated with some new technologies,we formulate the joint optimization problems of computing,communication resources and other important parameters considering the Practical application scenarios.At the algorithm level,combined with convex optimization and other advanced mathematical optimization algorithms,the problem is strictly deduced,simplified and solved to ensure the accuracy and efficiency of the algorithm.