A New Computing Offload Algorithm For Maximizing User Revenue Under Cloud-Edge-End Architecture

The rapid development of 5G network technology and smart devices has led to the emergence of more and more advanced applications.These emerging application services are often computation-intensive and delay-sensitive,which brings great challenges to user equipments with limited computing power.Computing offloading is one of the effective ways to solve this challenge.The traditional mobile cloud computing mode is a centralized processing mode in which computing tasks are offloaded to the MCC cloud server located in the center of the network for execution.Its disadvantage is that the MCC cloud server is usually far away from users and has a large network transmission delay,which cannot meet the requirements of low latency of application services.The mobile edge computing mode allows user equipments to offload computing tasks to MEC servers deployed on the edge of the network in order to meet latency requirements and complete task execution.However,the computing resources of MEC servers are limited after all,and it is especially difficult to cope with large-scale computation-intensive business.The cloudedge collaborative computing model combines the great computing power of MCC with the low latency of MEC,making it suitable for handling various types of application services.No matter what kind of computing mode,computing offloading has always been one of the key research issues.A good computing offload strategy can offload tasks to the most appropriate computing nodes to meet the various requirements of application services.This paper research the computing offloading problem under the cloud-edge-end architecture.The specific research work can be summarized as follows:(1)First,the optimization goals of the existing work mostly focus on the minimization of time delay and energy consumption,instead of focusing on the user revenue in the process of computing offloading.User revenue to guide user equipments how to reasonable offload tasks has great significance: if a computing action taken in the present system environment got bigger revenue,this shows that the computing action is matching the current system environment,which will encourage the user equipment take the same action in the future similar system environment.Therefore,this paper puts forward the concept of the Total User Revenue(TUR)(the total revenue includes time revenue and energy consumption revenue,which can be represented by the weighted sum of the two),and introduces TUR into the definition of the objective optimization problem in this paper.(2)Second,this paper proposes an adaptive intelligent offloading algorithm BDTUR based on DQN to solve the computing offloading problem in the multi-user single cell?one cloud and one edge vertical collaborative network architecture.The algorithm can make the most appropriate offloading decision according to the current system environment,and can continuously self-learn according to the feedback of the environment,and constantly improve the accuracy of the decision.The simulation results show that the intelligent offloading algorithm proposed in this paper has a higher total user revenue than the comparison algorithm.(3)Finally,inspired by the multi-cell MEC network architecture in 5G network environment,this paper takes the horizontal collaboration between MEC servers into consideration,and promotes the BDTUR to obtain the MBDTUR algorithm for the multiuser,multi-cell,one-cloud,multi-edge vertical and horizontal collaboration network architecture.In a dual collaborative network architecture,there are more choices of offloading points for computing tasks than in a vertical collaborative architecture.According to the simulation results,the performance of MBDTUR algorithm is not only better than the comparison algorithm,but also better than the performance of BDTUR algorithm in the vertical architecture.