Abstract
Use cases in the Internet of Things (IoT) and in mobile clouds often require the interaction of one or more mobile devices with their infrastructure to provide users with services. Ideally, this interaction is based on a reliable connection between the communicating devices, which is often not the case. Since most use cases do not adequately address this issue, service quality is often compromised. Aimed to address this issue, this paper proposes a novel approach to forecast the connectivity and bandwidth of mobile devices by applying machine learning to the context data recorded by the various sensors of the mobile device. This concept, designed as a microservice, has been implemented in the mobile middleware CloudAware, a system software infrastructure for mobile cloud computing that integrates easily with mobile operating systems, such as Android. We evaluated our approach with real sensor data and showed how to enable mobile devices in the IoT to make assumptions about their future connectivity, allowing for intelligent and distributed decision making on the mobile edge of the network.
Highlights
Mobile devices such as smartphones, wearables and sensor nodes have become more powerful every year
Afterwards, we extend the evaluation and focus on the amount and type of data that is required to achieve good forecasting accuracy
To evaluate the performance we simulate the usage of a mobile device, as described in (Orsini et al 2018b)
Summary
Mobile devices such as smartphones, wearables and sensor nodes have become more powerful every year They often rely on the resource augmentation through centralized resources, enabling a multitude of cloud-augmented mobile applications. Examples are location-based advertising, real-time sensor networks, the Nvidia Shield videogaming console (Nvidia 2019), which computes parts of the gameplay on remote resources, or the voice recognition assistant Siri (Apple 2019). Common to these use cases is the fact, that they rely on a preferably fast and stable connection to centralized or edge clouds (Abbas et al 2018).
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
