CoChat
Sign Up
Cited by 1,645Privacy-Preserving Technologies in DataHuman Mobility and Location-Based AnalysisPrivacy, Security, and Data Protection

The DOI matches its Crossref record; 30 of the 30 references it cites were found in a registry or an index.

Unique in the Crowd: The privacy bounds of human mobility

Open access
de Montjoye, Yves-Alexandre, Hidalgo, César A., Verleysen, Michel, Blondel, Vincent D.·2013·Scientific Reports
DOI 10.1038/srep01376PMC PMC3607247
Open this paper in CoChat — free accountOpen PDF

Abstract

We study fifteen months of human mobility data for one and a half million individuals and find that human mobility traces are highly unique. In fact, in a dataset where the location of an individual is specified hourly, and with a spatial resolution equal to that given by the carrier's antennas, four spatio-temporal points are enough to uniquely identify 95% of the individuals. We coarsen the data spatially and temporally to find a formula for the uniqueness of human mobility traces given their resolution and the available outside information. This formula shows that the uniqueness of mobility traces decays approximately as the 1/10 power of their resolution. Hence, even coarse datasets provide little anonymity. These findings represent fundamental constraints to an individual's privacy and have important implications for the design of frameworks and institutions dedicated to protect the privacy of individuals.

Overview

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1038/srep01376

Cites

30 of 30 references verified0 citation errors · 0 not checkable

Statistics

Figures

Figure 1 (A) Trace of an anonymized mobile phone user during a day.The dots represent the times and locations where the user made or received a call. Every time the user has such an interaction, the closest antenna that routes the call is recorded. (B) The same user's trace as recorded in a mobility database. The Voronoi lattice, represented by the grey lines, are an approximation of the antennas reception areas, the most precise location information available to us. The user's interaction times are here recorded with a precision of one hour. (C) The same individual's trace when we lower the resolution of our dataset through spatial and temporal aggregation. Antennas are aggregated in clusters of size two and their associated regions are merged. The user's interaction are recorded with a precision of two hours. Such spatial and temporal aggregation render the 8:32 am and 9:15 am interactions indistinguishable.
Figure 1 (A) Trace of an anonymized mobile phone user during a day.The dots represent the times and locations where the user made or received a call. Every time the user has such an interaction, the closest antenna that routes the call is recorded. (B) The same user's trace as recorded in a mobility database. The Voronoi lattice, represented by the grey lines, are an approximation of the antennas reception areas, the most precise location information available to us. The user's interaction times are here recorded with a precision of one hour. (C) The same individual's trace when we lower the resolution of our dataset through spatial and temporal aggregation. Antennas are aggregated in clusters of size two and their associated regions are merged. The user's interaction are recorded with a precision of two hours. Such spatial and temporal aggregation render the 8:32 am and 9:15 am interactions indistinguishable.CC BY-NC-ND · PubMed Central
Figure 2 (A) Ip = 2 means that the information available to the attacker consist of two 7am-8am spatio-temporal points (I and II).In this case, the target was in zone I between 9am to 10am and in zone II between 12pm to 1pm. In this example, the traces of two anonymized users (red and green) are compatible with the constraints defined by Ip = 2. The subset S(Ip = 2) contains more than one trace and is therefore not unique. However, the green trace would be uniquely characterized if a third point, zone III between 3pm and 4pm, is added (Ip = 3). (B) The uniqueness of traces with respect to the number p of given spatio-temporal points (Ip). The green bars represent the fraction of unique traces, i.e. |S(Ip)| = 1. The blue bars represent the fraction of |S(Ip)| ≤ 2. Therefore knowing as few as four spatio-temporal points taken at random (Ip = 4) is enough to uniquely characterize 95% of the traces amongst 1.5 M users. (C) Box-plot of the minimum number of spatio-temporal points needed to uniquely characterize every trace on the non-aggregated database. At most eleven points are enough to uniquely characterize all considered traces.
Figure 2 (A) Ip = 2 means that the information available to the attacker consist of two 7am-8am spatio-temporal points (I and II).In this case, the target was in zone I between 9am to 10am and in zone II between 12pm to 1pm. In this example, the traces of two anonymized users (red and green) are compatible with the constraints defined by Ip = 2. The subset S(Ip = 2) contains more than one trace and is therefore not unique. However, the green trace would be uniquely characterized if a third point, zone III between 3pm and 4pm, is added (Ip = 3). (B) The uniqueness of traces with respect to the number p of given spatio-temporal points (Ip). The green bars represent the fraction of unique traces, i.e. |S(Ip)| = 1. The blue bars represent the fraction of |S(Ip)| ≤ 2. Therefore knowing as few as four spatio-temporal points taken at random (Ip = 4) is enough to uniquely characterize 95% of the traces amongst 1.5 M users. (C) Box-plot of the minimum number of spatio-temporal points needed to uniquely characterize every trace on the non-aggregated database. At most eleven points are enough to uniquely characterize all considered traces.CC BY-NC-ND · PubMed Central
Figure 3 (A) Probability density function of the amount of recorded spatio-temporal points per user during a month.(B) Probability density function of the median inter-interaction time with the service. (C) The number of antennas per region is correlated with its population (R2 = .6426). These plots strongly emphasize the discrete character of our dataset and its similarities with datasets such as the one collected by smartphone apps.
Figure 3 (A) Probability density function of the amount of recorded spatio-temporal points per user during a month.(B) Probability density function of the median inter-interaction time with the service. (C) The number of antennas per region is correlated with its population (R2 = .6426). These plots strongly emphasize the discrete character of our dataset and its similarities with datasets such as the one collected by smartphone apps.CC BY-NC-ND · PubMed Central
Figure 4 Uniqueness of traces [ε] when we lower the resolution of the dataset with (A) p = 4 and (D) p = 10 points.It is easier to attack a dataset that is coarse on one dimension and fine along another than a medium-grained dataset along both dimensions. Given four spatio-temporal points, more than 60% of the traces are uniquely characterized in a dataset with an h = 15-hours temporal resolution while less than 40% of the traces are uniquely characterized in a dataset with a temporal resolution of h = 7 hours and with clusters of v = 7 antennas. The region covered by an antenna ranges from 0.15 km2 in urban areas to 15 km2 in rural areas. (B–C) When lowering the temporal or the spatial resolution of the dataset, the uniqueness of traces decrease as a power function ε = α − xβ. (E) While ε decreases according to a power function, its exponent β decreases linearly with the number of points p. Accordingly, a few additional points might be all that is needed to identify an individual in a dataset with a lower resolution.
Figure 4 Uniqueness of traces [ε] when we lower the resolution of the dataset with (A) p = 4 and (D) p = 10 points.It is easier to attack a dataset that is coarse on one dimension and fine along another than a medium-grained dataset along both dimensions. Given four spatio-temporal points, more than 60% of the traces are uniquely characterized in a dataset with an h = 15-hours temporal resolution while less than 40% of the traces are uniquely characterized in a dataset with a temporal resolution of h = 7 hours and with clusters of v = 7 antennas. The region covered by an antenna ranges from 0.15 km2 in urban areas to 15 km2 in rural areas. (B–C) When lowering the temporal or the spatial resolution of the dataset, the uniqueness of traces decrease as a power function ε = α − xβ. (E) While ε decreases according to a power function, its exponent β decreases linearly with the number of points p. Accordingly, a few additional points might be all that is needed to identify an individual in a dataset with a lower resolution.CC BY-NC-ND · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 29 references

Research works printed but not linked: 2

  1. Clanchy, M. T. From Memory to Written Records England 1066–1307 (Harvard University Press, Cambridge, 1979).
  2. Newman, M. E. J. Networks: An Introduction (Oxford University Press, New York, 2010).

Cited by

Sorted by citations

Related

See all

Metadata

Authors
de Montjoye, Y.-A., Hidalgo, C. A., Verleysen, M., & Blondel, V. D.
Affiliations
  • Harvard University (US)
  • UCLouvain (BE)
  • Massachusetts Institute of Technology (US)
ORCID
Venue
Scientific Reports
Publisher
Springer Science and Business Media LLC
Published
2013-03-25
Volume
3
Issue
1
Pages
1376-1376
ISSN
2045-2322 (ISSN-L)
DOI
10.1038/srep01376
PMC
PMC3607247
OpenAlex
W2115240023
Semantic Scholar
7fe16a270f403048f78fa89a64410e570ae5c16f
Cited by
1,645
References
29
Citations by year
2026: 96 · 2025: 86 · 2024: 92 · 2023: 96 · 2022: 133 · 2021: 156 · 2020: 191 · 2019: 153 · 2018: 152 · 2017: 124 · 2016: 124 · 2015: 119 · 2014: 89 · 2013: 33
FWCI
134.68
Type
journal article
Language
en
Open access
Licence
Locations
  • Scientific Reports (publishedVersion, cc-by-nc-nd)
  • Europe PMC (PubMed Central) (publishedVersion, cc-by-nc-nd)
  • PubMed Central (publishedVersion, cc-by-nc-nd)
  • PubMed (publishedVersion)
  • DSpace@MIT (Massachusetts Institute of Technology) (submittedVersion, cc-by-nc-nd)
  • Digital Access to Libraries (Université catholique de Louvain (UCL), l'Université de Namur (UNamur) and the Université Saint-Louis (USL-B)) (submittedVersion, other-oa)
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › Privacy-Preserving Technologies in Data
  • Social Sciences › Transportation › Human Mobility and Location-Based Analysis
  • Social Sciences › Sociology and Political Science › Privacy, Security, and Data Protection
Keywords
human mobility, privacy, anonymity, spatiotemporal resolution, location data, privacy preservation
MeSH terms
Geography, Humans, Privacy, Databases as Topic, Cell Phone

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review