Characterizing and Removing Oscillations in Mobile Phone Location Data

Abstract

Human mobility analysis is a multidisciplinary research subject that has attracted a growing interest over the last decade. A substantial amount of such recent studies is driven by the availability of original sources of real-world information about individual movement patterns. An important task in the analysis of mobility data is reliably distinguishing between the stop locations and movement phases that compose the trajectories of the monitored subjects. The problem is especially challenging when mobility is inferred from mobile phone location data: here, oscillations in the association of mobile devices to base stations lead to apparent user mobility even in absence of actual movement. In this paper, we leverage a unique dataset of spatiotemporal individual trajectories that allows capturing both the user and network operator perspectives in mobile phone location data, and investigate the oscillation phenomenon. We present probabilistic and machine learning approaches for detecting oscillations in mobile phone location data, and a filtering technique for removing those. Our analyses and comparison with state-of-the-art approaches demonstrate the superiority of our solution, both in terms of removed oscillations and of error with respect to ground-truth trajectories.

Complete trajectory reconstruction from sparse mobile phone data

“…This phenomenon consists in the mobile device association bouncing across nearby antennas, so that the location proxy for a same user changes even in absence of physical movement. The effect occurs at fast timescales of seconds [21,22], hence oscillations are highly likely to occur within a single time slot i, which spans tens of minutes or more in our case. In each slot, oscillations are then canceled by the antenna selection process above.…”

TRANSIT: Fine-grained human mobility trajectory inference at scale with mobile network signaling data

“…Since user association in mobile networks follows operator-specific schemes based on dynamic metrics such as received signal power or base station load, oscillations can easily take place between two or more antennas, even in absence of an actual mobility of the user. These characteristics add noise to the localization information that can be inferred from CDR data and make extremely hard the task of reliably discriminating between static and mobile sessions with CDR [23].…”

“…We thus tag as static antennas for user i those antennas with a daily cumulated time above a threshold T w . In our experiments, we set T w to 20 minutes, which falls within the range of commonly accepted values for the typical minimum duration of a significant activity carried out by an individual at a same location [44,18], and is employed also with high-frequency longitudinal (e.g., GPS) data [23]. An example is provided in Figure 6a.…”

Spatial Accuracy Evaluation for Mobile Phone Location Data With Consideration of Geographical Context

“…The geographical coordinates of the cell tower serving the mobile phone at a given time are used as an approximation of the subscriber position [7]. Therefore, the spatial accuracy of MPL data is closely related to the coverage (i.e., service area) of the cell towers.…”

“…A key characteristic of MPL data is that the locations are documented at the level of cell towers. These locations, which are usually represented as geographical coordinates of the cell towers, do not necessarily reflect the actual locations of the phone users [7], [8]. Obviously, the spatial accuracy of the MPL data directly affects the validity of human mobility research.…”