Hybrid Task Scheduling for Mobile Devices in Edge and Cloud Environments

Mobile cloud computing enables mobile devices to augment their computational capabilities with centralized resources. However, due to latency issues, mobile cloud computing is unsuitable in many situations. Lately, mobile edge computing appeared as a more decentralized paradigm, which utilizes nearby resources. It benefits from the continuously increasing amount and the enhancing performance of end-user devices. Compared to mobile cloud computing, nearby resources are limited, but reachable with a significantly shorter latency. We argue that a combination of both approaches can drastically improve the performance of mobile applications. In this paper, we introduce a hybrid scheduling approach. Beyond centralized scheduling on remote edge and cloud resources, we introduce an ad-hoc scheduling mechanism on nearby edge devices. Our context-aware scheduler considers both options along with the characteristics of the task to decide between an execution in the cloud or in the edge. We use mechanisms for edge device discovery, connection management, and task allocation as well as a context-aware utility function. We integrate our approach into an existing distributed computing system and evaluate the hybrid scheduler within a real-world testbed.