Research On Pricing-based Collaborative Computing Offloading And Resource Allocation Scheme For Edge Computing

The rapid evolution of the mobile Internet and the Internet of Things has spawned an enormous amount of new applications in the past few years.The generation of massive data requires mobile communication systems to provide higher data processing capability,stronger access capability and computing capability.Mobile edge computing(MEC)technology provides users with computing resource in the neighborhood and improves quality of user experience by deploying MEC server with specific computing capability at the edge of the network.Users migrate computingintensive tasks to the MEC server through computing offloading,which not only expands the computing capability of terminal devices,but also meets the diverse needs of different applications,reducing service delivery delay and energy consumption.However,compared with traditional cloud servers,MEC servers have relatively few computing resources.When too many users offload tasks to the MEC server for processing,it will cause excessive load,congestion,and affect quality of user experience.Therefore,this dissertation studies the efficient computing offloading and resource allocation scheme under the limited computing resource of MEC server.The main research contents are summarized as follows:1.To deal with the fairness caused by limited edge cloud computing ability during computation-intensive task offloading,a dynamic resource pricing strategy based on Stackelberg game model is proposed.Firstly,the real-time relationship between the residual amount of edge cloud system resource and user demand is analyzed,and the edge cloud network revenue and user cost function are designed.Then,the optimal offloading strategy based on the lowest user cost is obtained by verifying the existence of non-cooperative game Nash equilibrium point between users.The bidirectional search iterative algorithm is used to solve the edge benefit problem,and the optimal pricing strategy of edge cloud network adjusted by price adjustment factor is obtained.The simulation results show that the proposed algorithm can guarantee the edge cloud revenue when the computing resource is poor,and improve the user service quality under the premise of ensuring user fairness.2.Aiming at the problem of network congestion caused by increased load in hotspots,the dissertation considers the integrated network architecture of MEC and D2D(Device-to-Device)communication,and uses the computing resource of idle users to assist MEC server to complete task request users' data processing.Based on the social attributes between users,the appropriate D2 D offloading assistant is selected to share the load pressure of the MEC server.We optimize offloading decision,wireless resource allocation and computing resource allocation strategies in order to minimize the energy consumption and total computing resource overhead of all task request users.The cost minimization problem is modeled as a mixed integer nonlinear programming problem.In order to solve this problem,we first propose the D2 D assistant user selection algorithm to determine the assistant selection strategy of the task request user.Secondly,for the given assistant selection strategy,the original problem is further transformed into a convex problem and decomposed.Finally,a method of distributed algorithm based on alternating direction method of multipliers is adopted to solve the optimization problem.The simulation results prove the effectiveness of the proposed scheme in reducing user overhead.