Joint Optimization Of Computing Task Offloading And Communication Resource Allocation In Edge Networks

With the development of mobile Internet technology,face recognition,virtual reality,holographic projection and other emerging applications emerge in endlessly,these emerging applications put forward higher requirements for communication and computing under delay constraints.In order to reduce the transmission delay,mobile edge computing(MEC)deploys the computing server to the edge of the network,which makes it possible to unload the low latency tasks.Although many researches on energy optimization in mobile edge computing have been published,there are still some problems: firstly,computing offload is a process of energy transfer,many literatures only optimize the energy consumption of user equipment,but ignore the energy consumption of base station and MEC server.Secondly,for simplifying the modeling and solving process,few literatures optimize the unloading decision,computing and communication resource allocation at the same time.Thirdly,for simplifying the allocation of computing resources in MEC server,most literatures use parallel computing model,but some assumptions of the model do not conform to the mobile edge computing scenario.To make up for the lack of existing research,the main work of this thesis is as follows:1)When parallel computing model is used in MEC server,we model and solve the problem of multi-user computing offload.In order to minimize the weighted energy consumption of the system,we make joint optimization from the three dimensions of unloading decision,communication resource allocation and computing resource allocation,under the constraints of task completion time,limited communication and computing resources,and get the unloading decision,transmission time and computing resource allocation strategy of each task.Because of the integer constraints,the optimization problem is a mixed integer nonlinear programming(MINLP),we proved that it is an NP-hard problem,and the global optimal solution cannot be obtained within the polynomial time complexity,so we use penalty function method and successive convex approximation to design an approximation algorithm for finding the local optimal solution.Simulation results show that the approximate algorithm has fast convergence speed,better performance and lower system weighted energy consumption compared with other baseline algorithms.2)This thesis extends the sequential computing model,proposes a multi-core sequential computing model which is more in line with mobile edge computing scenario,and applies it to MEC server.The new optimization problem consists of three sub-problems:task unloading decision,task scheduling,computing and communication resource allocation,the first two sub-problems are proved to be NP-hard problems.Inspired by the flexible job shop scheduling problem(FJSP),we designed an improved genetic algorithm to solve task offloading and scheduling.On the basis of solving the first two sub-problems,the computing and communication resource allocation problem is a convex optimization problem,which is easy to solve.In the end,simulation results show that the improved genetic algorithm has faster convergence speed and better performance than other baseline algorithms.In this thesis,we propose a multi-core sequential computing model which is more in line with mobile edge computing.For multi-user scenarios,we solve the problems of offload decision,computing and communication resource allocation in parallel computing model and multi-core sequential computing model respectively,and reduce the weighted energy consumption of the whole system.