Mobile Cloud Service Architecture Design And Research On Computation Offloading Strategy

With the rapid development of the Internet and explosive growth of mobile applications based on "cloud",the data traffic has risen sharply.Researchers need to find new development field of mobile communication network to meet the growing communication demand,but no matter how the future network evolves,the feature of“cloud” is essential.It will be conducive to a trinity network form of the formation of communication,computing and storage through the combination of mobile cloud computing and communication network,using the communication network to carry computing tasks,using the computing technology to solve communication problems.So the new computing and communication integration of cloud service network architecture need to be proposed and has a far-reaching significance.Computation offloading is the basic principle and the main service delivery mode of mobile cloud computing which is beneficial to users to break through the limitations of the lack of physical resources,to achieve complex cloud services and better user experience.The main features of computation offloading are interactive and replaceable.Users exchange information with the cloud and consume part of the communication energy and bandwidth resources to replace the cloud strong storage / computing resources.Whether to offload the tasks need to consider the cost of time,the cost of energy and other factors.Therefore,a reasonable calculation of the computation offloading strategy is conducive to the minimization of user spending and network utilization optimization.In order to solve these two problems,in this paper we first reconstruct and innovate the overall network architecture that supports mobile cloud services,and then consider the task information,cloud information and communication network information to put forward the multi-site opportunistic computation offloading strategy.The main research work and innovation of this paper are as follows:(1)Based on the idea of integration of computing and communication,This paperdesigns a hierarchical distributed mobile cloud service architecture,based on the traditional core cloud services,into the emerging edge cloud services,and innovative proposed the device cloud services.The edge cloud extends the cloud service to the access side,and reduce communication latency and mitigate trunk link pressure;We use the device cloud to make the cloud services integrated into the client,to achieve the terminal collaborative computing.Also,this paper proposes a collaborative control cloud service system based on Software-Defined Networking(SDN)technology,which can effectively alleviate the management problems caused by the scattered cloud resources.(2)Based on the hierarchical distributed cloud service architecture,this paper describes the communication scenario of future multi-site mobile cloud computing,and puts forward the computation offloading delivery process based on SDN local cooperative controller proposed in this paper,and analyzes the opportunity computing characteristics.Then,the computational model of the "device-cloud" effective connectivity probability in mobile environment is proposed.Finally,the simulation model is simulated and validated by Matlab,and the validity of the model is proved.(3)Based on the multi-site opportunity computation offloading framework,this paper designs the multi-site computation offloading strategy by considering the task information,the cloud state information and the network channel information.In this paper,the task state is defined by the amount of computation data and the amount of traffic data,the cloud state is measured with the effective connectivity probability,and the wireless channel state is fitted with the GE channel model.And in order to minimize system cost(including time cost,energy consumption cost and penalty cost)The computation offloading problem is transformed into the optimization problem,and the mathematic model of the Markov decision with constraint is established.The optimal strategy for computation offloading is solved by using the backward iterative algorithm with Q learning.Finally,the results show that the superiority of the algorithm can reduce the cost of the user effectively by comparing the other three strategies with Matlab simulation.