Joint resource allocation and offloading strategies in cloud enabled cellular networks

The numerous features installed in recent mobile phones opened the door to a wide range of applications involving localization, storage, photo and video taking and communication. A significant number of applications involve user generated content and require intensive processing which limits dramatically the battery lifetime of featured mobile terminals. Mobile cloud computing has been recently proposed as a promising solution allowing the mobile users to run computing-intensive and energy parsimonious applications. This new feature requires new functionalities inside the cellular network architecture and needs appropriate resource allocation strategies which account for computation and communication in the same time. In this paper we present promising options to upgrade 4G architecture to support these new features. We also present two resource allocation strategies accounting for both computation and radio resources. These strategies are devised so that to minimize the energy consumption of the mobile terminals while satisfying predefined delay constraints. We compare online learning based solutions where the network adapts dynamically to the application that is run on mobile terminals, and pre-calculated offline solutions which are employed when a certain level of knowledge about the application and the channel conditions is available at the network side. We show, that even with imperfect knowledge about the application, pre-calculated offline strategies offer better performance in terms of energy consumption of mobile terminals.