Finding frequently visited paths: Dealing with the uncertainty of spatio-temporal mobility data

With the ever-increasing advancements in sensor technology and localization systems, large amounts of spatio-temporal data can be collected from moving objects equipped with wireless sensor nodes. Analysis of such data provides the opportunity of extracting useful information about movement behaviour and interaction between moving objects. Inherent characteristics of wireless sensor nodes cause the data collected by them to have low or irregular frequency and often be erroneous. Existence of different levels of uncertainty in these data makes the procedure of finding movement patterns difficult and ambiguous. In this paper, we propose a hierarchical approach to find the frequently visited paths using location data of people carrying a custom designed mobile wireless sensor node. We hierarchically cluster trajectories and find their resemblance at the finest level while dealing with the uncertainties. The performance evaluation results show that compared with previous schemes, our method performs better in presence of ambiguity and sources of data uncertainty.