Sensor Signatures - AI Assisted Co-Location Detection of IoT Devices.

This paper proposes a system utilizing a variety of simple, cheap sensors in order to determine the co-location or relative position of IoT devices. The system relies on machine learning to evaluate a set of co-location clues - so called signatures and based on those signatures determines whether devices are likely to share the same pallet, vehicle or storage room. We present real-life experiments showing that our system can detect co-location even if the sensory data are noisy and common signatures are hard to detect for the human eye. The presented system is aimed for industrial applications primarily, such as logistics.