Research On Computing Offloading Policy And Resource Allocation In Mobile Edge Computing System

With the rapid development of the Internet of Things,the number of mobile devices in the future wireless network will grow rapidly,which inevitably leads to massive data and computation and require high resource-consumption.Mobile edge computing(MEC)is widely regarded as the key technology to solve these problems.Mobile edge computing realizes real-time information transmission and calculation by transferring computationally intensive tasks from mobile devices to the edge of the network.In practical applications,there are many limitations such as limited wireless and computing resources.Therefore,reasonable computation offloading policy and resource allocation schemes for data transmission and processing in MEC system can effectively improve the performance of MEC system.It has become an especially important research topic.In this dissertation,by utilizing matching theory and Lagrangian duality,we study how to jointly optimize computation offloading policy and resource allocation in MEC system and propose new algorithms.Then we perform simulation experiments to verify its superiority and analyze the influence of parameters in the MEC system on performance.Our main work and contributions are detailed below:(1)The basic concepts,development overview and application scenarios of the two-sides matching theory are introduced.And then the Lagrangian multiplier method and the use of the KKT condition are explained in detail.On this basis,the concept and characteristics of the Lagrangian duality are introduced.Finally,knowledge of the ellipsoid method is is briefly mentioned.(2)A new distributed computing offloading policy based on the two-sides matching theory is proposed.Specifically,we consider a MEC system consisting of multiple users and multiple APs,and model the interactions between the users and APs as a two-sided match,where each user selects one best AP and determine the corresponding optimal amount of data for offloading to minimize its own latency according to the current status of each channel,while each AP makes decision whether to accept the offloaded computation tasks from the users according to its current remaining computation resources and the benefits that can be brought by receiving these tasks.(3)A resource allocation algorithm based on Lagrangian duality and ellipsoid method to jointly optimize power and bandwidth allocation is proposed.Specifically,we build a MEC system consisting of multiple users and multiple APs,and we assume that each user can simultaneously offload data to all APs without interfering with each other instead of selecting only one AP to offload tasks.We formulate the joint power and bandwidth resource allocation problem with the objective of maximizing the sum of the total number of computation bits offloaded from users,and then decompose the joint optimization problem into two subproblem,which are solved by Lagrangian duality and ellipsoid method.