Using acceleration signatures from everyday activities for on-body device location

This paper is part of an effort to facilitate wearable activity recognition using dynamically changing sets of sensors integrated in everyday appliances such as phones, PDAs, watches, headsets etc. A key issue that such systems have to address is the position of the devices on the body. In general each devices can be in a number of different locations (e.g. headset on the head or in on of many pockets). At the same time most activity recognition algorithms require fixed, known sensor positions. Previously we have shown on a small data set how to recognize a set of on-body locations during a walking motion using an accelerometer signal. We now extend the method to work during arbitrary activity. We verify it on a much larger data set with a total 9 hours from real life activity by three divers users ranging from a 70 year old housewife to a 28 year male student.