Computing at the Mobile Edge: Designing Elastic Android Applications for Computation Offloading

Current centralized cloud datacenters provide scalable computation- and storage resources in a virtualized infrastructure and employ a use-based "pay-as-you-go" model. But current mobile devices and their resource-hungry applications (e.g., Speech-or face recognition) demand for these resources on the spot, though a mobile device's intrinsic characteristic is its limited availability of resources (e.g., CPU, storage, bandwidth, energy). Thus, mobile cloud computing (MCC) was introduced to overcome these limitations by transparently making accessible the apparently infinite cloud resources to the mobile devices and by allowing mobile applications to (elastically) expand into the cloud. However, MCC often relies on a stable and fast connection to the mobile devices' surrogate in the cloud, which is a rare case in mobile scenarios. Moreover, the increased latency and the limited bandwidth prevent the use of real-time applications like, e.g. Cloud gaming. Instead, mobile edge computing (MEC) or fog computing tries to provide the necessary resources at the logical edge of the network by including infrastructure components to create ad-hoc mobile clouds. However, this approach requires the replication and management of the applications' business logic in an untrusted, unreliable and constantly changing environment. Consequently, this paper presents a novel approach to allow mobile app developers to easily benefit from the features of MEC. In particular, we present a programming model and framework that directly fit the common app developers' mindset to design elastic and scalable edge-based mobile applications.