Location Prediction for Indoor Spaces based on Trajectory Similarity

Abstract

Predicting the further whereabouts of a large scale of users in indoor spaces has just been a hot and challenging issue in recent years due to the ubiquitous emerging of mobile devices and free Wi-Fi services, e.g. in shopping centers. While the existing prediction algorithms focusing on modeling the movement of users via mathematical models are useful in the outdoor environment, they fall short in predicting the position of indoor moving objects in a constraint but full of spatial-semantic information environment. To tackle this problem, we present a similarity based model by incorporating the spatial and the location contexts into a unified framework. We first present a novel trajectory similarity method which considers the spatial and contextual information in the indoor settings, then based on the similarities we present a clustering algorithm to group the trajectories, finally the most similar trajectory is returned for the prediction. In order to evaluate the precision of our proposed method, we designed a comprehensive performance evaluation using a large-scale dataset collected between September 2012 and October 2013 from over 120,000 anonymized, opt-in consumers in a large supermarket. Results show that our approach achieves a much better trajectory prediction performance against the baseline methods.