Research On Task Offloading And Virtual Machine Scheduling Strat Egies In Mobile Cloud Computing

Mobile Cloud Computing(MCC)is a new mode that integrates mobile computing,cloud computing and Internet technology.MCC aims at providing mobile devices with storage service and computing service.MCC is an ideal approach to overcoming the inherent defects of mobile devices.On one hand,MCC breaks down terminal hardware limitations.On the other hand,MCC augments the functions of mobile devices in computing,storage,energy saving and context awareness.Task offloading can reduce the burden of mobile devices and improve user experience.Virtual machine scheduling can reasonably allocate resources in cloud data center.Therefore,researches on task offloading and virtual machine scheduling are essential.Existing researches on task offloading strategies are mainly focused on reducing power consumption,and there are only a few studies on minimizing task execution time.Designing an effective task offloading strategy aimed at minimizing execution time has important theoretical significance.Deployment of virtual machines influences on many aspects,such as resource cost utilization,physical servers load balancing and overhead reducing.Designing a reasonable virtual machine scheduling strategy has important application value for resources allocation in cloud data center.By converting mobile applications to task graphs,a Time-Efficient Task Offloading Algorithm(TETOA)is proposed.TETOA is adopted by three steps.Firstly,convert an application to a task graph,which stands for the order and data transmission relationship of tasks.Secondly,by introducing “Time Gain”,a dynamic programming is adopted on conventional task graphs to determine the offloading queue.Thirdly,computation-intensive tasks on complex task graphs are given offloading priorities for further decision.Simulation results show that TETOA is effective in minimizing task execution time,reducing load of mobile devices and improving task execution efficiency.For virtual machine scheduling,a Multi-objective Particle Swarm Optimization algorithm with Adaptive Inertia weight and Acceleration coefficient based on Feedback information(MPSO-AIAF)is proposed.MPSO-AIAF modifies the standard PSO from four aspects:(1)A feedback mechanism is introduced,so that particles can adaptively adjust their speeds according to historical feedbacks of the fitness function information.(2)An inertia weight adaptive adjustment factor is introduced to balance the global search and local search.(3)An adaptive learning selection parameter is introduced to dominate the processes of self-learning and social-learning of particles.(4)The ratio allocation adjustment factor is introduced,so as to find the optimal solution of the overall goal.MPSO-AIAF is adopted to resource allocation scenario to achieve the aim of minimizing the number of migrated virtual machines and balancing the load of physical servers in the cloud data center.Simulation results show that MPSO-AIAF can significantly improve the speed and accuracy of finding the solution to virtual machine scheduling.The research achievements of this paper can provide references for task offloading and virtual machine scheduling in mobile cloud computing.