Towards gateway to Cloud offloading in IoT publish/subscribe systems

It is not uncommon today that sensor devices connected to the Internet solely send their data to Cloud-based servers for storage and processing. This does not only mean clients requesting data have to contact the Cloud-based service, even if the data is available in the local network, but also that data is sent to external services with unknown or ambiguous privacy policies. The great potential in using closer to the edge fog computing instead of Cloud computing to both enable faster and more privacy-aware processing locally has been recognized in the research community. In particular, on premise smart gateways can provide local low-latency storage and processing capabilities that are controlled locally and can be trusted. In this work, we outline a combined fog Cloud system that automatically selects a suitable execution location for processing tasks. We emphasize on the design challenges of such a system and further demonstrate a solution for the interplay between Fog and Cloud by showing how processing task can be migrated from one system to another on the fly without service interruption.