中央研究院 資訊科學研究所

Fingerprint lookup and proximity detection are positioning technologies commonly used by indoor positioning and indoor object tracking systems (IOTS) to locate objects and track their movements. Ideally, every fingerprint captured by every device is mapped to the cell containing its current location, and the UUID of every tag is detected by the proximity detector closest to its location. This enables the IOTS to return the correct location of the device/tag. As long as the device/tag does not move, a graphical user interface (GUI) of the IOTS shows it at the location. When the device/tag moves, the GUI shows its new correct location after a short delay. In real-life operating environments, many factors, including the presence of objects in signal paths, interferences from emitters nearby, natural fluctuations in transmitted power, and so on, can cause unpredictable fluctuations in received signal strengths. Consequently, the GUI may show the object moving sporadically even when the object is standing still or moving haphazardly when it actually moves along a well defined trajectory. Such apparent changes in object positions returned by an IOTS due to unpredictable and sporadic changes in signal strengths is referred to as noisy movements or false movements. Users are likely to find noisy/false movements annoying and the location information confusing. The techniques described in this report aim to eliminate or reduce noisy/false movements and improve the accuracy of locations returned by IOTS.