Computing Offloading And Migration Algorithm For Energy Consumption And Delay In Cloud-Edge Collaborative Networks

With the rapid development of smart city and 5G technology,mobile data traffic is growing exponentially.Mobile cloud computing(MCC)mode is difficult to continuously exchange and process information generated by hundreds of Internet of things(IOT)devices.In order to overcome the problems of high end-to-end delay and high channel transmission pressure caused by the centralized processing mode of cloud data center,the academia and industry consider the integration of mobile edge computing(MEC)and mobile cloud computing to form a cloud edge collaborative service mode,jointly support a variety of business needs in the era of Internet of things,and provide diversified services.However,there is still a lack of reasonable solutions in the specific operation mode and energy efficiency of the edge network.Therefore,this paper explores the cloud-edge-terminal network cooperation operation mode,and based on the cloud-edge-terminal collaborative network(CETCN),studies the offloading strategy among cloud platform,edge platform and terminal,jointly allocates the communicational and computational resources of each network segment,reduces the network delay and energy consumption,and realizes the offloading optimization through task migration on edge platform,so as to further improve the resource utilization of edge nodes.The contributions of this paper are as follows1.Aiming at the smart city scenario,taking the delay and energy consumption as the optimization objectives,this paper proposes an energy constrained computing offloading with sleep control for cloud-edge-terminal collaborative network algorithm.The algorithm combines the related services and design the model of computing offloading problem with server sleeping.Then,considering the terminal energy constraints in the actual network,the model is transformed into a cross slot energy consumption model based on energy consumption queue by Lyapunov optimization method.Finally,the offloading vector is solved by reinforcement learning method.The algorithm allocates CETCN network resources reasonably and dynamically for IOT services.Finally,in theory and simulation,compared with the edge-terminal offloading algorithm?cloud-termianl offloading algorithm and offloading algorithm without sleeping mechanism,the effectiveness of the algorithm is verified.2.This paper proposes a CETCN resource allocation algorithm based on task migration,which uses edge task migration to balance the computing pressure of edge nodes and further improve the resource utilization of edge nodes.The algorithm optimizes the resource allocation between multiple network segments and within the edge platform by designing a computing task offloading algorithm based on reinforcement learning and a computing task migration algorithm based on Lagrange.Simulation results show that the proposed algorithm reduces the long-term average energy consumption by 42%-74%compared with the offloading algorithm without task migration.