On the anonymizability of mobile traffic datasets

TL;DR

This paper investigates the challenges of anonymizing large-scale mobile traffic datasets, revealing inherent difficulties due to trajectory uniqueness and proposing insights for more robust anonymization methods.

Contribution

It introduces a novel measure for $k$-anonymizability and provides new insights into the reasons behind poor anonymizability of mobile traffic data.

Findings

Mobile traffic datasets exhibit high trajectory uniqueness.

Existing anonymization methods are often ineffective.

The study offers new insights for improving dataset anonymization.

Abstract

Preserving user privacy is paramount when it comes to publicly disclosed datasets that contain fine-grained data about large populations. The problem is especially critical in the case of mobile traffic datasets collected by cellular operators, as they feature elevate subscriber trajectory uniqueness and they are resistant to anonymization through spatiotemporal generalization. In this work, we investigate the $k$-anonymizability of trajectories in two large-scale mobile traffic datasets, by means of a novel dedicated measure. Our results are in agreement with those of previous analyses, however they also provide additional insights on the reasons behind the poor anonimizability of mobile traffic datasets. As such, our study is a step forward in the direction of a more robust dataset anonymization.