Indoor Positioning Based Consecutive Pattern Mining for Pedestrian Flow Analysis

In recent years, pedestrian flow analysis has gained popularity in public area such as shopping malls, hospitals or public facilities. Also, as the location-based service (LBS) become prevalent, more indoor environments have provided wireless positioning system which can record user's location and generate user's trajectory database. In this paper, a pedestrian flow analysis scheme is proposed on the basis of recorded location sequences provided by indoor wireless positioning system. To consider different application scenarios for pedestrian flow and with the existence of positioning errors, we proposed a trajectory regularization method to normalize the location sequences in a suitable format. Furthermore, to analysis the pedestrian flow, a trajectory consecutive pattern mining method which considers the sequential continuity of the trajectories is proposed based on the properties and proofs of consecutiveness of frequent patterns. Simulation results show that our proposed scheme can provide effective pedestrian flow analysis for both route and hotspot scenarios with lowered computational complexity.