The Design And Implementation Of A Multi-Tenant Remote Virtual Smartphone Platform

With the rapid development of mobile internet,people tend to use smart mobile devices for work instead of using traditional personal computers or laptops.However,performance of smartphones varies.Some applications may not work well on low-end devices due to hardware capability,which can greatly hurt the user experiences.Meanwhile,the virtualization technology becomes more mature and has been applied by cloud service providers to increase hardware utilization and reduce maintenance costs.Desktop virtualization,which combines virtualization and remote desktop,has attracted more and more attention these years.By deploying virtual machines in cloud,users can connect to their virtual desktop as long as they have internet access,which is very flexible.However,this architecture is usually used for desktop computers.Therefore,it is of great significance to research how to apply this architecture to the mobile platform.The proposed system creates virtual smartphones in cloud and transfers the screen of virtual smartphones to end users' physical devices.And users' inputs on their physical devices are transferred back to the virtual smartphones.As a result,the virtual smartphones can be used on remote devices.The features of the system are as follows: 1.Low-end devices can run applications which require more resources by using remote virtual smartphones.2.The system intercepts screen update at the virtual machine monitor,transparent to the virtual smartphones.3.An ant colony algorithm is proposed to solve the virtual machine placement problem to improve hardware utilization.The evaluation shows that the virtual smartphone outperforms Google Nexus 5 by 2.17 times.The CPU utilization on the server is only 12.07% when running 4 virtual smartphones.And the CPU utilization increases linearly as the number of virtual smartphones grows,which shows the good scalability of the system.By encoding the screen using lossless compression,the proposed system costs no more than 700 kbps network bandwidth.At the same time,an ant colony algorithm is proposed to solve the virtual machine placement problem on this multi-tenant platform,in order to improve the resource utilization.The experiment indicates that the proposed algorithm can figure out a solution to place 1000 virtual machines within 9 minutes,which is suitable to use in practice.