Exploiting Joint Wifi/Bluetooth Trace to Predict People Movement

It is well known that the daily movement of people exhibits a high degree of repetition in which people usually stay at regular places for their daily activities. This paper presents a novel framework to construct a predictive model by exploiting the regularity of people movement found in the collected joint Wifi/Bluetooth trace. Our obtained predictive model is able to answer three fundamental questions: (1) where the person will stay at a future time, (2) how long she will stay at the location, and (3) who she will meet at a future time. In order to construct the predictive model, we first propose an efficient clustering algorithm to cluster Wifi access points in the Wifi trace into clusters and use these clusters to represent locations. Then, we construct a Naive Bayesian classifier to assign these locations to records in Bluetooth trace. The combined Wifi/Bluetooth trace with locations is used to construct the location predictor, stay duration predictor, and people predictor. Finally, we evaluate three predictors over the real Wifi/Bluetooth traces collected by 50 experiment participants in University of Illinois campus from March to August 2010. The results confirm that our predictors provide highly accurate predictions of location, stay duration, and people.