“Guess Who ?” Large-Scale Data-Centric Study
of the Adequacy of Browser Fingerprints for
Web Authentication
Nampoina Andriamilanto, Tristan Allard, and Ga¨
etan Le Guelvouit
Abstract Browser fingerprinting consists in collecting attributes from a web
browser to build a browser fingerprint. In this work, we assess the adequacy of
browser fingerprints as an authentication factor, on a dataset of 4,145,408 fin-
gerprints composed of 216 attributes. It was collected throughout 6 months from
a population of general browsers. We identify, formalize, and assess the proper-
ties for browser fingerprints to be usable and practical as an authentication factor.
We notably evaluate their distinctiveness, their stability through time, their col-
lection time, and their size in memory. We show that considering a large surface
of 216 fingerprinting attributes leads to an 81.8% unicity rate on a population of
1,989,365 browsers. Moreover, browser fingerprints are known to evolve, but we
observe that between consecutive fingerprints, more than 90% of attributes remains
unchanged after nearly 6 months. Fingerprints are also affordable. On average, they
weight a dozen of kilobytes, and are collected in a few seconds. We conclude that
browser fingerprints are a promising additional web authentication factor.
1 Introduction
Web authentication widely relies on identifier-password pairs. Passwords are easy
to use, but suffer from severe security flaws. Indeed, users use common passwords,
paving the way to brute-force or guessing attacks [1]. They also use similar pass-
Nampoina Andriamilanto
Institute of Research and Technology b<>com, Rennes, France e-mail: nampoina.
andriamilanto@b-com.com
Tristan Allard
Univ Rennes, CNRS, IRISA, Rennes, France e-mail: tristan.allard@irisa.fr
Ga¨
etan Le Guelvouit
Institute of Research and Technology b<>com, Rennes, France e-mail: gaetan.
leguelvouit@b-com.com
1
arXiv:2005.09353v2 [cs.CR] 10 Jun 2020
2 Nampoina Andriamilanto, Tristan Allard, and Ga¨
etan Le Guelvouit
words across websites [15], which increases the impact of attacks. Phishing attacks
are also a major threat to passwords. Over the course of a year, Thomas et al. [13]
achieved to retrieve 12.4 million credentials stolen by phishing kits. These flaws
gave rise to multi-factor authentication [2], such that each additional authentication
factor provides an additional security barrier. However, this usually comes at the
cost of usability (i.e., users have to remember, possess, or do something).
In the meantime, browser fingerprinting gains attention. The Panopticlick study [3]
highlights the possibility to build a browser fingerprint by collecting attributes from
a web browser. In addition to being widely used for web tracking purposes [4] (rais-
ing legal and ethical issues), browser fingerprints are used as an authentication factor
in real-life. Browser fingerprints are indeed good a candidate as an authentication
factor thanks to their distinctive power, their frictionless deployment (e.g., no addi-
tional software), and their usability (no secret to remember, no additional object to
possess, and no supplementary action to carry out). As a result, companies like Mi-
croFocus1or SecureAuth2include browser fingerprints within their authentication
mechanisms (see Figure 1 for an example of such mechanism).
Fig. 1 Simplified browser fingerprinting web authentication mechanism.
Related works. To the best of our knowledge, no large-scale study rigorously
evaluates the adequacy of browser fingerprinting as an authentication factor. Most
works about their use for authentication concentrate on the design of authentication
mechanism [14, 10, 6, 11], and empirical studies on browser fingerprints focus on
their efficacy as a web tracking tool [3, 7, 5]. Such a mismatch between the un-
derstanding of browser fingerprints for authentication – currently poor – and their
ongoing adoption in real-life is a serious harm to the security of web users. The lack
of documentation from the existing tools (e.g., about the used attributes, the distinc-
tiveness, and the stability of the resulting fingerprints) only adds up to the current
state of ignorance. All this whereas security-by-obscurity contradicts the most fun-
damental security principles.
Our contributions. We conduct the first large-scale data-centric empirical study
of fundamental properties of browser fingerprints when used as an additional au-
thentication factor. We base our findings on an in-depth analysis of a real-life fin-
gerprint dataset collected over 6 months, that contains 4,145,408 fingerprints com-
posed of 216 attributes. We formalize, and assess on our dataset, the properties nec-
1https://www.microfocus.com/media/white-paper/device-
fingerprinting-for- low-friction- authentication-wp.pdf
2https://docs.secureauth.com/pages/viewpage.action?pageId=
33063454
Data-Centric Study of the Adequacy of Browser Fingerprints for Web Authentication 3
essary for paving the way to elaborate browser fingerprinting authentication mecha-
nisms. The selected properties are usually used to evaluate biometric characteristics
for authentication [8]. We stress that we do not make any assumption on the inner
working of the authentication mechanism, and consequently on the adversarial strat-
egy. Our properties aim at characterizing the adequacy and practicability of browser
fingerprints, independently of their use within future authentication mechanisms. In
particular, we measure the size of browser anonymity sets through time, and show
that 81.8% of our fingerprints are unique. Moreover, we measure the proportion of
identical attributes between two observations of the fingerprint of a browser, and
show that 90% of attributes remains unchanged after nearly 6 months. Finally, we
measure the collection time and the size of fingerprints. We show that on average,
they weight a dozen of kilobytes, and are collected in a few seconds.
The rest of the paper is organized as follows. Section 2 presents and formalizes
the properties evaluated in our analysis. Section 3 describes the dataset analyzed in
this study. Section 4 presents our experimental results. Finally, Section 5 synthesizes
the results and concludes.
2 Authentication Factor Properties
The “Handbook of Fingerprint Recognition” [8] summarizes properties that a bio-
metric characteristic requires to be usable3as an authentication factor, and addi-
tional properties required for a biometric authentication scheme to be practical. We
make the connection between fingerprints used to recognize persons, and browser
fingerprints used to recognize browsers. So we evaluate browser fingerprints ac-
cording to these properties, to assert their adequacy for web authentication. In this
section, we list these properties, formalize how to measure some properties, and
explain why the others are not addressed in this study.
The four properties needed for an anatomical or a behavioral characteristic to be
usable as a biometric authentication factor are described below.
Universality: the characteristic should be present in everyone.
Distinctiveness: two distinct persons should have different characteristics.
Permanence: the same person should have the same characteristic over time. We
rather use the term stability.
Collectibility: the characteristic should be collectible and measurable.
The three properties that a biometric authentication scheme requires to be prac-
tical are the following.
Performance: the scheme should consume few resources, and be robust against
environmental changes.
Acceptability: the users should accept to use the scheme in their daily lives.
3Here, usable refers to the adequacy of the characteristic to be used for authentication, rather than
the ease of use by users.
4 Nampoina Andriamilanto, Tristan Allard, and Ga¨
etan Le Guelvouit
Circumvention: it should be difficult for an attacker to deceive the scheme.
The properties that we study are the distinctiveness, the stability, and the per-
formance. We consider that the universality and the collectibility are satisfied, as
the HTTP headers that are automatically sent by browsers constitute a fingerprint.
However, we stress that a loss of distinctiveness occurs when no JavaScript attribute
is available. About the circumvention, we refer the reader to Laperdrix et al. [6] that
analyzed the security of an authentication mechanism based on browser fingerprints.
We let the evaluation of the acceptability as future works, but we stress that such
mechanisms are already used in a rudimentary form4.
2.1 Distinctiveness
To satisfy the distinctiveness property, browser fingerprints should enable two dif-
ferent browsers to be distinguishable. The distinctiveness depends on the used at-
tributes, and on the fingerprinted browser population. The two extreme cases are
every browser sharing the same fingerprint, which makes them indistinguishable
from each other, and no two browsers sharing the same fingerprint, making every
browser distinguishable.
Our dataset entries are composed of a fingerprint, the source browser, and the
time of collection in the form of a Unix timestamp in milliseconds. We denote Bthe
domain of the unique identifiers (UIDs), and Tthe timestamp domain. The finger-
print dataset is denoted Dand is formalized as:
D={(f,b,t)|fF,bB,tT}(1)
We use the size of browser anonymity sets to quantify the distinctiveness,
as browsers belonging to the same anonymity set are indistinguishable. We de-
note S(f,D)a function returning the set of browsers that provide the fingerprint f
in the dataset D. It is formalized as:
S(f,D) = {bB| ∀(g,b,t)D,f=g}(2)
We denote A(ε,D)a function providing the set of fingerprints having an anonymity
set size of ε(i.e., being shared by εbrowsers) in the dataset D. It is formalized as:
A(ε,D) = {fF|card(S(f,D)) = ε}(3)
We measure the anonymity set sizes on the fingerprints currently in use by each
browser, and not on their whole history. It is performed by simulating datasets com-
posed of the last fingerprint seen for each browser at a given time. Let Eτ(D)be
the simulated dataset originating from Dthat represents the state of the fingerprints
after τdays. With tτthe last timestamp of this day, we have:
4https://support.google.com/accounts/answer/1144110
Data-Centric Study of the Adequacy of Browser Fingerprints for Web Authentication 5
Eτ(D) = {(fi,bj,tk)D| ∀(fp,bq,tr)D,trtktτ}(4)
2.2 Stability
Browser fingerprints have the particularity of evolving through time, due to changes
in the web environment like a software update or a user configuration. We measure
the stability by the mean similarity between two consecutive fingerprints coming
from a browser, given the elapsed time between them. The two extreme cases are
every browser holding the same fingerprint through its life, and the fingerprint of a
browser changing completely at each observation.
We denote C(,D)a function providing the set of consecutive fingerprints in D
that are separated by a time difference comprised in the time range. It is formal-
ized as:
C(,D) = {(fi,fp)|∀(( fi,bj,tk),(fp,bq,tr)) D2,
bj=bq,tk<tr,(trtk)}(5)
We consider the Kronecker delta δ(x,y), being 1 if xequals y, and 0 otherwise.
We denote f[ω]the value taken by the attribute ωfor the fingerprint f. Let sim(f,g)
be a simple similarity function between fingerprints, formalized as:
sim(f,g) = 1
n
n
ω=1
δ(f[ω],g[ω]) (6)
We define the function meansim(,D)providing the mean similarity of the con-
secutive fingerprints, for a given time range and a dataset D, as:
meansim(,D) = (f,g)C(,D)sim(f,g)
card(C(,D)) (7)
2.3 Performance
To evaluate the performance of browser fingerprints used in an authentication con-
text, we consider three aspects. The first two are the consumption of time and mem-
ory resources. The third is the loss of efficacy (i.e., distinctiveness and stability)
among device types.
The collection time of fingerprints only depend on JavaScript attributes, as HTTP
headers are transmitted anyway. So we measure the collection time of our finger-
prints composed of 200 JavaScript attributes.
The size of fingerprints depends on their storage format. For example, a can-
vas [9] image can be encoded as a base64 string or as a hash. We stress that com-
pressing the complete fingerprint to a single hash is unpractical due to the evolution
6 Nampoina Andriamilanto, Tristan Allard, and Ga¨
etan Le Guelvouit
of fingerprints. The size of attributes is not specified, hence we measure the size of
the fingerprints of our dataset.
Previous works showed that mobile and desktop devices present differences in the
properties of their browser fingerprints [12, 7, 5]. Mobile browsers usually have less
distinctive fingerprints. Following these findings, we assess that the distinctiveness
and the stability of the fingerprints of these two groups are similar.
3 Fingerprint Dataset
To study the properties of browser fingerprints on a real-world browser population,
we launched a fingerprint collection experiment. It was performed in collabora-
tion with the authors of [5], and an industrial partner that controls one of the top
15 French websites according to Alexa5. The authors of [5] held the 17 attributes of
their previous work [7] and focused on web tracking, whereas we held 216 attributes
and focused on web authentication.
PTC [3] AIU [7] HITC [5] This study
Collection period 3 weeks 3-4 months* 6 months 6 months
Number of attributes 8 17 17 216
Number of browsers - - - 1,989,365
Number of fingerprints 470,161 118,934 2,067,942 4,145,408
Number of distinct fingerprints 409,296 142,0236-3,578,196
Proportion of desktop fingerprints - 0.890* 0.879 0.805
Proportion of mobile fingerprints - 0.110* 0.121 0.134
Unicity of global fingerprints 0.836 0.894 0.336 0.818
Unicity of mobile fingerprints - 0.810 0.185 0.399
Unicity of desktop fingerprints - 0.900 0.357 0.884
Table 1 Dataset comparison between Panopticlick, AmIUnique, Hiding in the Crowd, and this
study. -denotes missing information. *denotes deduced information. The attributes only comprises
original ones, and fingerprints are counted after data preprocessing.
3.1 Fingerprint Collection
We compiled 200 JavaScript properties and 16 HTTP header fields, and designed
a fingerprinting probe that collects these attributes. We integrated the probe to two
general audience web pages of our industrial partner, which subjects are political
5https://www.alexa.com/topsites/countries/FR
6This number is provided in Figure 11 as the distinct fingerprints, but also corresponds to the raw
fingerprints. Every fingerprint would be unique if the number of distinct and collected fingerprints
are equal, hence we are not confident in this number, but it is the one provided by the authors.
Data-Centric Study of the Adequacy of Browser Fingerprints for Web Authentication 7
news and weather. The probe collected fingerprints from December 7, 2016, to June
7, 2017. Only the visitors that consented to cookies were fingerprinted, in compli-
ance with the European directives 2002/58/CE and 2009/136/CE in effect at the
time. To differentiate browsers, we assigned them a unique identifier (UID) as a
6-months cookie. Similarly to [3, 7], we coped with cookie deletion by storing a
one-way hash of the IP address, computed by a secure cryptographic hash function.
Previous datasets were collected through dedicated websites, and are biased to-
wards privacy-aware and technically-skilled persons [3, 7]. Our population is more
general audience oriented, but the website audience is mainly French-speaking
users. This leads to a bias towards this population. The timezone is set to 1 for
98.48% of browsers, 98.59% of them have daylight saving time enabled, and fr is
present in 98.15% of the Accept-Language HTTP header value.
3.2 Dataset Filtering and Preprocessing
Given the experimental aspect of fingerprints and the scale of our collection, the raw
dataset contained erroneous or irrelevant samples. We remove 70,460 entries entries
that have a wrong format (e.g., empty or truncated data), that are duplicated, or that
come from a robot.
Cookies are an unreliable identification method, hence we perform a resynchro-
nization similar to [3]. We consider the entries that have the same (fingerprint, IP
address hash) pair to come from the same browser, and assign them the same UID.
Similarly to [3], we do not synchronize the interleaved UIDs, being the pairs that
have UID values b1,b2, then b1again. We replace 181,676 UIDs with 116,708
replacement UIDs using this method.
To avoid counting multiple entries of identical fingerprints coming from the same
browser, the usual way is to ignore them during collection [3, 7]. Our probe collects
fingerprint on each visit, and to stay consistent with common methodologies we
deduplicate the fingerprints afterward. For each browser, we hold the first entry hav-
ing a given fingerprint, and ignore the following entries if they have this fingerprint.
For example, if a browser bhas the entries {(f1,b,t1),(f2,b,t2),(f2,b,t3),(f1,b,t4)},
we only hold the entries {(f1,b,t1),(f2,b,t2),(f1,b,t4)}. The deduplication consti-
tutes the biggest cut in our dataset, with 2,420,217 entries filtered out.
We extract 46 additional attributes from 9 original attributes, which are of two
types. The first type consists in extracted attributes composed of parts of original
attributes, like the screen resolution that is split into the values of width and height.
The second type consists of information sourced from an original attribute, like the
number of plugins extracted from the list of plugins.
8 Nampoina Andriamilanto, Tristan Allard, and Ga¨
etan Le Guelvouit
3.3 Work Dataset
The work dataset obtained after the preprocessing step contains 5,714,738 en-
tries (comprising identical fingerprints for a given browser if interleaved), with
4,145,408 fingerprints (no identical fingerprint counted for the same browser), com-
posed of 3,578,196 distinct fingerprints. The fingerprints are composed of 216 orig-
inal attributes and 46 extracted ones, for a total of 262 attributes. They come from
1,989,365 browsers, 27.53% of which have multiple fingerprints. Table 1 presents
a comparison between the dataset of Panopticlick [3], AmIUnique [7], Hiding in the
Crowd [5], and this study.
4 Empirical Evaluation of Browser Fingerprints Properties
4.1 Distinctiveness
Figure 2 presents the size of the anonymity sets (AS) alongside the frequency of
browser arrival for the daily-partitioned datasets. We call unicity rate the proportion
of fingerprints that belong to an AS of size one. Our fingerprints have a stable unicity
rate of approximately 81.3% on the long run, and at least 94.7% of fingerprints are
shared by 8 browsers or less. However, the fingerprints of the mobile group are more
uniform than that of the desktop group, with a unicity rate of approximately 42% on
the long run.
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 20 40 60 80 100 120 140 160 180 200
0
50000
100000
150000
200000
Proportion of fingerprints
New browsers
Days since collection beginning
Anonymity set size == 1
Anonymity set size <= 2
Anonymity set size <= 8
New browsers
Fig. 2 Anonymity set sizes and frequency of browser arrivals through partitioned datasets obtained
after each day.
New browsers are encountered continually, but starting from the 60th day, the
arrival frequency stabilizes around 5,000 new browsers per day. Before this stabi-
Data-Centric Study of the Adequacy of Browser Fingerprints for Web Authentication 9
lization, we have a variable arrival frequency with some major spikes. They seem to
correspond to events having happened in France that lead to more visits. For exam-
ple, the spike on the 38th day corresponds to a live political debate on TV, and the
spike on the 43rd day correlates with the announcement of a cold snap.
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 20 40 60 80 100 120 140 160 180 200
Proportion of unique fingerprints
Days since collection beginning
All
Desktops
Mobiles
Fig. 3 Proportion of unique fingerprints for overall, mobile, and desktop groups, through parti-
tioned datasets obtained after each day.
Figure 3 presents the proportion of unique fingerprints through partitioned datasets
for overall, mobile, and desktop groups. The proportion of unique fingerprints is
stable for the desktop browsers, with a slight increase of 1.04 points from the 60th
day to the 183th, from 84.99% to 86.03%. The unicity rate of the fingerprints of
the mobile group is lower than that of the desktop group, and has a little decrease
of 0.29 points on the same period, from 42.42% to 42.13%.
4.2 Stability
Two fingerprints of a given browser can be linked as only a small portion of the at-
tributes is expected to change, even after several months. Figure 4 displays the mean
similarity between consecutive fingerprints in function of the time difference. The
ranges are expressed in days, so that day don the x-axis represents the fingerprints
separated by = [d;d+1[days. We ignore the comparisons of time ranges having
less than 10 pairs, or with a time difference higher than the limit of our experiment
(182 days), which account for less than 0.03% of each category. Our stability re-
sults are a lower bound, as consecutive fingerprints are necessarily different (i.e.,
their similarity is strictly lower than 1).
We have a total of 3,725,373 compared pairs for the overall group, 2,912,860 pairs
for the desktop group, and 594,591 pairs for the mobile group. A fingerprint is ex-
pected to have at least 90% of its attributes having an identical value after 170 days.
10 Nampoina Andriamilanto, Tristan Allard, and Ga¨
etan Le Guelvouit
0.86
0.88
0.9
0.92
0.94
0.96
0.98
1
0 20 40 60 80 100 120 140 160 180
10
100
1000
10000
100000
1x106
1x107
Mean similarity of fingerprints
Number of compared pairs
Days between two consecutive fingerprints
All
Mobiles
Desktops
Number of comparisons
Fig. 4 Mean similarity between consecutive fingerprints in function of the time difference, with
the number of compared pairs.
Few attributes among those included in our script are highly unstable. Getting rid of
these attributes could reduce the distinctiveness of fingerprints, but would improve
their stability.
The fingerprints of the mobile group are generally more stable than these of the
desktop group, as suggests their respective similarity curve. However, it seems to
be the case only for high time differences, as in the range of [1;2[days difference,
the fingerprints of the mobile group are slightly less stable than these of the desktop
group with a mean similarity of 97.87% against 98.11%.
4.3 Performance
4.3.1 Time Resource Consumption
Our script takes several seconds to collect the attributes composing the fingerprints.
Figure 5 displays the cumulative distribution of the collection time of fingerprints in
seconds, with the outliers removed. We measure it by the time difference between
the starting of the script and the fingerprint sending. Some values take from several
hours to days, that can come from a web page put in background or accessed after a
long time. We limit our population to the fingerprints that take less than 30 seconds
to collect, and consider higher values as outliers. Outliers account for less than 1% of
each group.
We present the collection time of fingerprints in seconds for the (5th percentile,
median, 95th percentile). Our script collects most fingerprints within a few sec-
onds, with values (0.66, 2.92, 10.42). A difference occurs between the desktop
browsers (0.61, 2.64, 10.45) and the mobile browsers (2.06, 4.44, 10.16). The me-
dian collection time is less than the estimated median time taken by web pages to
Data-Centric Study of the Adequacy of Browser Fingerprints for Web Authentication 11
0
0.2
0.4
0.6
0.8
1
0 5 10 15 20 25 30
Cumulative proportion of fingerprints
Collection time in seconds
All
Mobiles
Desktops
Fig. 5 Cumulative distribution of the collection time of fingerprints in seconds.
load completely7, being at 6.6 seconds for the desktop browsers and 19.5 seconds
for the mobile browsers, at the date of February 1, 2020.
4.3.2 Memory Resource Consumption
Our script consumes a dozen of kilobytes per fingerprint, a size easily handled by
the current storage and bandwidth capacities. Figure 6 displays the cumulative dis-
tribution of fingerprint size in bytes, with the outliers removed, and canvases stored
as sha256 hashes. The mean fingerprint size is µ=7,692 bytes, and the standard
deviation is σ=2,294. We remove 1 fingerprint from a desktop browser considered
an outlier because of its size being greater than µ+15 ·σ.
Half of our fingerprints take less than 7,550 bytes, and 99% less than 14 kilo-
bytes. It is negligible given the current storage and bandwidth capacities. We observe
a difference between the fingerprints of mobile and desktop browsers, with 95% of
fingerprints weighing respectively less than 8,020 bytes and 12,082 bytes. This is
due to heavy attributes being lighter on mobiles, like the plugins or mime types lists
that are most of the time empty.
5 Synthesis of Results and Conclusion
In this study, we evaluate the properties offered by browser fingerprints as an ad-
ditional web authentication factor, through the analysis of a large-scale real-life
fingerprint dataset. We show that browser fingerprints offer a satisfying distinc-
7https://httparchive.org/reports/loading-speed#ol
12 Nampoina Andriamilanto, Tristan Allard, and Ga¨
etan Le Guelvouit
0
0.2
0.4
0.6
0.8
1
0 5000 10000 15000 20000 25000
Cumulative proportion of fingerprints
Fingerprint size in bytes
All
Mobiles
Desktops
Fig. 6 Cumulative distribution of the size of fingerprints in bytes.
tiveness, as 81.8% of our fingerprints are only shared by one browser. Moreover,
fingerprints are stable. At least 90% of the attributes are expected to stay identical
between two observations of the fingerprint of a browser, even if they are separated
by nearly 6 months. We validate that fingerprints offer a high performance, as they
only weight a dozen of kilobytes and take a few seconds to collect. We conclude
that browser fingerprints provide satisfying properties for an additional web authen-
tication factor, and can strengthen password-based systems without a major loss of
usability.
Acknowledgements We want to thank Benot Baudry and David Gross-Amblard for their valuable
comments, together with Alexandre Garel for his work on the experiment. This is a preprint of
a contribution published in Innovative Mobile and Internet Services in Ubiquitous Computing,
edited by Leonard Barolli, Aneta Poniszewska-Maranda, and Hyunhee Park, and published by
Springer International Publishing. The final authenticated version is available online at: https:
//doi.org/10.1007/978-3- 030-50399- 4_16.
References
1. Bonneau, J.: The science of guessing: analyzing an anonymized corpus of 70 million pass-
words. In: Symposium on Security and Privacy (2012)
2. Bonneau, J., Herley, C., van Oorschot, P.C., Stajano, F.: Passwords and the Evolution of Im-
perfect Authentication. Communications of the ACM (2015)
3. Eckersley, P.: How unique is your web browser? In: Privacy Enhancing Technologies (2010)
4. Englehardt, S., Narayanan, A.: Online Tracking: A 1-million-site Measurement and Analysis.
In: Conference on Computer and Communications Security (2016)
5. G´
omez-Boix, A., Laperdrix, P., Baudry, B.: Hiding in the Crowd: an Analysis of the Effec-
tiveness of Browser Fingerprinting at Large Scale. In: The Web Conference (2018)
Data-Centric Study of the Adequacy of Browser Fingerprints for Web Authentication 13
6. Laperdrix, P., Avoine, G., Baudry, B., Nikiforakis, N.: Morellian analysis for browsers: Mak-
ing web authentication stronger with canvas fingerprinting. In: Conference on Detection of
Intrusions and Malware & Vulnerability Assessment (2019)
7. Laperdrix, P., Rudametkin, W., Baudry, B.: Beauty and the Beast: Diverting modern web
browsers to build unique browser fingerprints. In: Symposium on Security and Privacy (2016)
8. Maltoni, D., Maio, D., Jain, A.K., Prabhakar, S.: Handbook of Fingerprint Recognition.
Springer-Verlag (2003)
9. Mowery, K., Shacham, H.: Pixel perfect: Fingerprinting canvas in html5. In: W2SP (2012)
10. Preuveneers, D., Joosen, W.: SmartAuth: Dynamic Context Fingerprinting for Continuous
User Authentication. In: Annual Symposium on Applied Computing (2015)
11. Rochet, F., Efthymiadis, K., Koeune, F., Pereira, O.: SWAT: Seamless web authentication
technology. In: The World Wide Web Conference (2019)
12. Spooren, J., Preuveneers, D., Joosen, W.: Mobile Device Fingerprinting Considered Harmful
for Risk-based Authentication. In: European Workshop on System Security (2015)
13. Thomas, K., Li, F., Zand, A., Barrett, J., Ranieri, J., Invernizzi, L., Markov, Y., Comanescu,
O., Eranti, V., Moscicki, A., et al.: Data breaches, phishing, or malware?: Understanding the
risks of stolen credentials. In: Conference on Computer and Communications Security (2017)
14. Unger, T., Mulazzani, M., Frhwirt, D., Huber, M., Schrittwieser, S., Weippl, E.: SHPF: En-
hancing HTTP(S) Session Security with Browser Fingerprinting. In: Conference on Avail-
ability, Reliability and Security (2013)
15. Wang, C., Jan, S.T., Hu, H., Bossart, D., Wang, G.: The next domino to fall: Empirical analysis
of user passwords across online services. In: Conference on Data and Application Security
and Privacy