SenseWaves: Radiowaves for context recognition

We are surrounded by a sensor-enriched environment which is able to provide a broad spectrum of features from various sensor classes that can be utilised for situation and activity awareness. However, the probably most common sensor, shipped with nearly every electronic device nowadays such as mobile phones, notebooks, printers as well as keyboards, mouses, watches, shoes, media players – rumour has spread about even media cups – is mostly not utilised for activity recognition: The RF-sensor. Due to its high penetration, the additional cost for utilising this sensor in an application is considerably low. The application must simply utilise the information available but discarded unused in these devices. Although the wireless channel is frequently utilised for location detection of other active RF devices, it is seldom used to detect other contexts than location from entities that are not actively transmitting. We demonstrate a system for activity recognition based on features extracted from the RF channel. In particular, we show how static changes in the environment such as moved furniture, activity of a person and an ongoing phone call are detected based on RF channel measurements.