Efficient management of mobile cloud computing services

Mobile devices such as smartphones and tablet PCs are becoming ever more popular. Mobile users appreciate small and light devices. Due to these design constraints, the computing power, storage and autonomy of these devices is limited. Consequently, not all applications can be executed on mobile devices. In this dissertation, the use of thin client computing is put forward as an enabler for accessing any application from any (mobile) device. In the thin client computing paradigm, applications are not executed on the user device but on powerful servers in the network. User input (e.g. keystrokes) and audiovisual output of the application is exchanged between the client device and the remote server executing the application. To offer an efficient, high-quality thin client service to a large group of potential mobile users, important challenges have to be overcome. In this dissertation, a system architecture is presented to offer a thin client service in the cloud. The servers executing the applications of the users have to be distributed over the network to limit the impact of network delay on the interactivity between the user and his/her application. Therefore, a dimensioning algorithm is presented to decide where to install servers and how many servers should be installed. An important part of the system architecture is the distributed management framework. The management framework implements intelligent algorithms for allocating resources for the applications of the users taking into account the mobility of the users and exploiting the shared resource platform of the servers. Furthermore, the management framework enables further optimization of the service from the provider’s point-of-view, for example, shutting down superfluous servers during calm periods to limit the energy consumption of the infrastructure. In this dissertation, several optimization algorithms are presented and evaluated by means of simulations. A prototype of the management framework has been implemented to validate the results from the simulations