Activity recognition and step detection with smartphones: Towards terminal based indoor positioning system

For years now, many indoor positioning techniques have been researched and some approaches have even been developed into consumer products. Solutions range from using existing WiFi structures to higher accuracy proprietary solutions, e.g. products using UWB (Ultra Wide Band). When public institutions such as libraries or shopping malls offer applications for users (e.g. shop finders), a problem of user privacy can arise. That is, depending on the type of positioning topology implemented, the user and his habits might become traceable. Since modern devices are increasingly equipped with better sensors, they could be used for independent terminal based positioning systems based on step detection. Hence, we decided to explore the capabilities of using the sensors of the HTC hero smartphone for detection of steps and of the movement/activity type. We implemented and examined three state of the art algorithms for these types of detection. The experimental results of detecting activities, e.g. “walking” and “jogging,” will be presented along with the different influencing factors such as types of shoes and position of the device. For future work we are working on an automatic and autonomous step length calculation system in order to come one step closer to a real terminal based positioning system.