From cell tower location to user location: Understanding the spatial uncertainty of mobile phone network data in human mobility research

Abstract

Mobile phone network data is a vital source for unveiling human mobility characteristics in accordance with its large-scale spatiotemporal trajectory information. However, mobile phone network data usually records location at the level of cell towers, lacking accurate individual locations. Therefore, the authenticity and credibility of the conclusions drawn from such data are often questioned due to the spatial uncertainty. In this paper, we evaluate the location differences between users and the cell towers during connection establishment. Furthermore, we delve into the representation and contributing factors of spatial uncertainty, including cell tower density, antenna status, and user mobility. Our analysis is based on one-month mobile signaling data and taxi GPS data collected in Foshan (a prefecture-level city in China), which represent two forms of data on the mobility of the same individual. We conclude that to estimate user positions, areas significantly larger than the nearest cell tower are necessary. The influence of tower density and antenna load on connection accuracy does not exhibit a straightforward linear dependency; instead, it fluctuates once a threshold is reached. Connection accuracy is typically higher when users are stationary than when they are in motion. Our findings together indicate that it should carefully assess the accuracy of position estimation when mapping from cell tower location to user location.