Research Of Hybrid Cloud Selection Strategy Based On Minimum Offloading Cost In Mobile Cloud Computing

Along with the continuous development of mobile communication technologies,the mobile devices represented by smart phones,tablet PCs,and vehicle terminals have been widely used.Users are constantly satisfied with various needs,such as face recognition,interactive games,image/video applications,through running rich mobile applications on mobile devices.Although mobile devices have made great progress in hardware technology,their resources,including computing resources(for example,CPU,memory,etc.)and energy supplies(battery power)are still limited compared to traditional computers.Therefore,running a variety of compute-intensive applications on resourceconstrained mobile devices remains a challenge.Mobile cloud computing(MCC)as an emerging and prospective computing paradigm,can be regarded as a solution to solve this challenge.Mobile cloud computing allows mobile devices to take advantage of elastic resources provided by the cloud,as well as offload the compute-intensive application to resource-rich cloud after partitioning,according to reducing application completion time and saving mobile device's energy consumption.However,there may be heterogeneous cloud resources around the mobile device,such as central cloud,cloudlets,edge cloud,and vehicle cloud.How to choose a suitable cloud for unloading,which can lead to the minimum execution cost including application completion time and mobile device's energy consumption is still the current research difficulties and hotspots.Aiming at the current hybrid cloud computing problem,this thesis puts forward how to partition application and computational migration,and proposes cloud selection algorithm and offload strategies based on the existing task offloading strategies.At the same time,this thesis designs a system framework to provide runtime support for computation partitioning and offloading,as well as cloud selecting,and the motivation is to minimize the execution cost of the application.The cloud selection algorithm firstly determines whether the application needs to be divided and migrated according to the migration cost of the application.Then,it can make decision which cloud will be selected to offload the task by the execution cost of a task on each type of cloud.The purpose is that the task can choose a kind of cloud to execute for getting the minimum execution cost among the variety of cloud resources around the mobile device.Furthermore,in this paper,the system framework is deployed on mobile terminal and cloud server by using clone mechanism,and consists of three parts: profiler,solver,and communication module.The profiler collects and analyzes the resource usage of the mobile device and the clouds,and sends the analyzed results to the solver.The solver makes decision of partitioning based on the results sent by the profiler,and determines whether to perform task migrated according to the cloud selection algorithm.At last,the communication module is responsible for the communication and data transmission between the mobile terminal and the clouds.Besides,this thesis studies the mobility problem for mobile users in mobile edge computing.We apply our framework to the mobile edge computing scene.And we solve the delay problem caused by the user moving in the mobile edge cloud through the virtual machine migration technology and improving the user's service experience.Finally,the experimental simulation results show that the proposed cloud selection algorithm can select the best cloud more accurately with the least execution cost for task offloading compared with other existing task migration algorithms,make the applications on the mobile device have faster completion time as well as lower energy consumption,and improve user's quality of service.Meanwhile,our proposed system framework is highly modular and easier to extend than previous system frameworks.