Finding Frequently Visited Indoor POIs Using Symbolic Indoor Tracking Data

Indoor tracking data is being amassed due to the deploymentof indoor positioning technologies. Analysing such data dis-closes useful insights that are otherwise hard to obtain. Forexample, by studying tracking data from an airport, we canidentify the shops and restaurants that are most popularamong passengers. In this paper, we study two query typesfor finding frequently visited Points of Interest (POIs) fromsymbolic indoor tracking data. The snapshot query findsthose POIs that were most frequently visited at a giventime point, whereas the interval query finds such POIs fora given time interval. A typical example of symbolic track-ing is RFID-based tracking, where an object with an RFIDtag is detected by an RFID reader when the object is in thereader’s detection range. A symbolic indoor tracking systemdeploys a limited number of proximity detection devices, likeRFID readers, at preselected locations, covering only part ofthe host indoor space. Consequently, symbolic tracking datais inherently uncertain and only enables the discrete cap-ture of the trajectories of indoor moving objects in terms ofcoarse regions. We provide uncertainty analyses of the datain relation to the two kinds of queries. The outcomes of theanalyses enable us to design processing algorithms for bothquery types. An experimental evaluation with both real andsynthetic data suggests that the framework and algorithmsenable efficient and scalable query processing.