Rubal Jain and Chander Kant /International Journal of Advances in Scientific Research 2015; 1(07): 283-288.

283

International Journal of Advances in Scientific Research

ISSN: 2395-3616 (Online)
Journal DOI: 10.7439/ijasr Review Article

Attacks on Biometric Systems: An Overview

Rubal Jain*1 and Chander Kant2

1Research Scholar, Department of Computer Science and Applications, K.U., Kurukshetra, India
2Assistant Professor, Department of computer Science and Application, K.U., Kurukshetra, India

*Correspondence Info:
Rubal Jain
Research Scholar,
Department of Computer Science and Applications,
K.U., Kurukshetra, India
E-mail: [email protected]

Abstract
Biometrics is a pattern recognition system that refers to the use of different physiological (face,
fingerprints, etc.) and behavioral (voice, gait etc.) traits for identification and verification purposes. A
biometrics-based personal authentication system has numerous advantages over traditional systems such as
token-based (e.g., ID cards) or knowledge-based (e.g., password) but they are at the risk of attacks. This paper
presents a literature review of attack system architecture and makes progress towards various attack points in
biometric system. These attacks may compromise the template resulting in reducing the security of the system
and motivates to study existing biometric template protection techniques to resist these attacks.
Keywords: Biometrics, Biometric Attacks, Biometric Traits, Biometrics System Template, Generic Threats,
Architecture, Template Protection Techniques.
1. Introduction
Biometric is a science through which system
can uniquely identify an individual on the basis of his
physiological (face, iris, fingerprint, hand geometry,
retina etc.) and behavioral (gait, voice, signature,
keystroke etc.) traits [1]. The use of biometric traits as
an authentication technology has become widespread
from door access to e-commerce due to the need of
better security in many fields. Biometric systems are
more convenient to use as compared with traditional
authentication systems such as token based (e.g., ID
cards) or knowledge based (e.g., passwords) because it
alleviates the need to remember long passwords and to Fig.1. Identification Process
carry tokens with itself. It also guards the user against
repudiation. Biometric-based personal authentications In verification mode (shown in fig.2), system
system may operate in two different modes: carries out a one-to-one comparison to set up an
identification and verification modes [1]. individual’s identity. In other words, the user claims
In identification mode (shown in fig.1), an identity and the system verifies whether the claim is
system carries out a one-to-many comparison to set up genuine or not on the basis of validating a sample
an individual’s identity. In other words, the user’s collected against a previously collected biometric
input is compared with all the templates stored in sample for the individual. The purpose of verification
system database. The purpose of identification is to is to answer the question “Am I who I say I am”?
answer the question: “Who am I?”. Identification
systems are costly to deploy and needs processing
time to find a match within database.

IJASR|VOL 01|ISSUE 07|2015 www.ssjournals.com

Rubal Jain and Chander Kant / Attacks on Biometric Systems: An Overview 284

predefined security threshold it will accept the

individual otherwise reject it.

Fig.2. Verification Process

Each biometric system has four basic
modules whether it is identification system or
verification system. In the following section 2
biometrics system along with its four basic modules
(sensor module, feature extractor module, matcher
module and decision module) is explained. Fig.3. Biometric System
There is no doubt biometrics-based
2. Biometric System authentication systems may overcome the limitations
All the biometric systems have four basic of traditional systems (alleviates the need to remember
modules which are sensor module, feature extractor passwords, to carry tokens, etc.) but they also
module, matcher module and decision module [2]. possesses some limitations such as they are vulnerable
These four modules are necessary in any biometric to attacks. There are eight different attack points in
system to acquire and process raw biometric data and biometric system which can be attacked and the
convert it into some useful information. The block following section 3 deals with these attack points.
diagram of biometric system is shown in fig. 3.
2.1. Sensor Module 3. Attacks on biometric system
In this type of module raw biometric data is Biometric based authentication systems that
captured by the sensor and it scans the biometric trait uses physiological (face, iris etc.) and behavioral traits
to convert it into digital form. After converting it to (voice, signature etc.) are becoming increasingly
digital form, this module transmits the data to feature popular and utilized in many applications to increase
extraction module. the security of the system. Traditional systems are
2.2. Feature Extraction Module unable to distinguish between an authorized person
It processes the raw data captured by sensor and intruder who can fraudulently access the system.
and generate a biometric template. It extracts the Biometric systems are more convenient to use because
necessary features from the raw data which needs there is no need to remember any password and with a
much attention because essential features must be single biometric trait different account can be secured
extracted in an optimal way. It basically removes noise without the burden of remembering passwords.
from the input sample and transmits the sample to Biometric systems offer great advantages over
input sample to the succeeding module known as traditional systems but they are vulnerable to attacks
matcher module. [3]. There are eight attack points in biometric system
2.3. Matcher Module which can be attacked as shown in fig.4. These attack
This module compares the input sample with points are divided into two categories: Direct attacks
the templates being stored in the database using and indirect attacks.
matching algorithm and produces match score. The
resulting match score is transmitted to the decision
module, which decides whether to accept the
individual or not.
2.4. Decision Module
After accepting the match score from matcher
module, it compares the matching score against the
predefined security threshold. This module accepts or
rejects the individual on the basis of predefined
security threshold. If match score is greater than Fig.4. Attack points on biometric system
IJASR|VOL 01|ISSUE 07|2015 www.ssjournals.com
Rubal Jain and Chander Kant / Attacks on Biometric Systems: An Overview 285

3.1. Direct attacks matcher later on. It is known as “Attack on the channel
It refers to the attacks that do not require any between the feature extractor and matcher”.
specific knowledge about the system operation such as 3.2.4. Type 5 attack
matching algorithm used, feature vector format, etc. It A matcher module is vulnerable to type 5th
includes only type 1 attack which is referred as attack which is known as “Attack on matcher module”
“Sensor Attack”. [5]. It is attacked to generate the high matching score
3.1.1. Type 1 attack as selected by the imposter to bypass the biometric
The sensor module is vulnerable to type 1 authentication system regardless of the values obtained
attack which is known as “Attack at the sensor”. In from the input feature set.
this attack, a fake biometric trait such as an artificial 3.2.5. Type 6 attack
finger or facial image is presented to the sensor by an It occurs when the imposter compromises
imposter to bypass recognition systems [4]. An with the security of the database by adding new
imposter can also physically damage the recognition templates, modifying existing templates and removing
system and flood the system with bogus access existing templates [5]. It is not an easy task to attack
requests. It is very easy to attack at the sensor because system database because templates are protected by
no specific knowledge about the system operation is digital mechanisms such as steganography,
needed and there is no digital protection mechanisms watermarking, etc. To make successful attack on
such as watermarking, cryptography are used at the system database some knowledge of inner working of
sensor level. Sensors are unable to distinguish between the system must be needed.
fake and real characteristics of an individual and can 3.2.6. Type 7 attack
be fooled easily by using synthetic fingerprints and Attack can be made possible only when
facial image of a person. template is transmitting through communication
3.2. Indirect attacks channel between system database and matcher
Unlike direct attacks, these are the attacks module. It occurs when imposter modifies or tampers
where information about the inner working of the the contents of the transmitted template. An imposter
authentication system is required to make an attack intercepts the channel to steal, replace or alter
successful. It includes all the remaining seven points biometric template. It is known as “Attack on the
of attack (2, 3, 4, 5, 6, 7, 8.) that can be attacked by an communication channel between system database and
impostor in a biometric-based authentication system. the matcher”.
3.2.1. Type 2 attack 3.2.7. Type 8 attack
When the sensor acquires a raw biometric An imposter may override the result declared
data, it sends the raw data to feature extractor module by the matcher module. In this attack, imposter may
for pre-processing through a communication channel. tamper the match score which is transmitted through
This channel is in between sensor and the feature communication channel between matcher module and
extractor module. It is intercepted to steal the application device. It tampers the match score to
biometric trait and stored somewhere. The previously change the original decision (accept or reject) of the
stored biometric trait is replayed to the feature matcher module.
extractor to bypass the sensor. This is known as After studying these eight attack points
“replay attack” [4]. author observed that most of the time adversary
3.2.2. Type 3 attack attacks on the templates which are to be stored in
The feature extractor module is vulnerable to database. These templates which are stored in database
type 3 attack which is known as “Attack on feature can be tampered by adding new templates to database,
extractor module”. When the sensor acquires a raw modifying existing templates in database and
biometric data, it sends the raw data to feature removing existing templates from the database. In the
extractor module. An imposter pressurize the feature following section 4, the author had discussed about
extractor module to produce the feature values chosen architecture of attack system, which explains how to
by the intruder instead of producing the feature values made successful attack by tampering with synthetic
generated from the original data obtained from the fingerprint templates.
sensor.
3.2.3. Type 4 attack 4. Architecture of attack system
This attack is similar to attack of type 2 but A template is a digital reference of distinct
difference is in that, an imposter intercepts the traits that represents a set of important features
communication channel between the feature extractor extracted from the biometric data of an individual. Its
and matcher modules and steal the feature values of nature is compact in database, due to which it infers
genuine user [4]. These values can be replayed to the
IJASR|VOL 01|ISSUE 07|2015 www.ssjournals.com
Rubal Jain and Chander Kant / Attacks on Biometric Systems: An Overview 286

that template cannot reveal complete information (ii) an existing minutia is replaced with a new
about original biometric data. In addition, the minutia.
templates are stored in an encrypted form, it is difficult (iii) a new minutia is added to the template.
to decipher and find out the contents of the stored (iv) an existing minutia is deleted.
template without knowing the suitable decrypting
keys. However, an attack system presented in the If modified template results in increasing the
literature review conflict these beliefs. match score then declare the modified template as Ti
best
and update the high match score accordingly. The
The architecture of an attack system that current best score is then compared against decision
attacks a minutiae-based fingerprint authentication threshold (S threshold ) set by the matcher and if ((S best
system is shown in fig. 5 [6]. The attack system (Di, Ti best) > (Sthreshold)) then stop the attack; else, the
consists of two subsystems: attacking system and process of modifying the current synthetic template
target system. Both the subsystems consists of and comparing it against the target template, is
different modules such as attacking system consists of repeated until the match score exceeds a decision
synthetic template generator (STG) and attack module threshold.
whereas; target system consists of fingerprint matcher As discussed above biometric systems are
and template database. vulnerable to attacks which can make use of various
generic security threats to harm the integrity of any
biometric system. These threats may result in different
effects on different systems. In section 5, the author
has discussed about various generic security threats
which can be possible to any biometric systems.

5. Generic security threats

All traditional systems and biometric systems
are susceptible to various threats [7]. These threats are
as follows:
i. Denial of service: It refers to act where an imposter
Fig.5. An Attack System Architecture overwhelms the authentication system with bogus
requests so that genuine users cannot use it. An
A minutiae–based fingerprint authentication authentication server that processes access requests
system makes the use of synthetic template generator can be loaded with many bogus access requests, to a
to produce synthetic fingerprint templates, which in point that all computational resources are wasted in
turn makes the use of “Hill Climbing Attack” to find bogus requests and cannot handle valid requests any
out the contents of a targeted fingerprint template (Di) more.
for the ith user. The minutiae points which consist of ii. Circumvention: It occurs when an imposter gains
ridge bifurcations and endings are used to create access to the system that was secured by the
fingerprint minutiae. In general, all the minutiae based authentication process and manipulates the system by
systems use the sequence of location (c, r) of the changing records in an unauthorized manner.
minutiae and orientation  of the minutiae as its iii. Repudiation: A genuine user may access the
attributes but some systems also use ridges flow resources of the system and claims that system had
around the minutiae as supplementary information. been circumvented by the imposter. For example a
The working of attack system begins with STG by corrupt bank officer tampered some financial records
generating a fixed number of synthetic templates (Tij illegally and claims that his/her biometric data has
where j= 1, 2, 3…….m) each of which composed of stolen.
randomly generated minutiae points [7]. These iv. Contamination: It refers to an act where an
templates are compared against the target template via imposter can secretly obtain biometric data of genuine
the matcher and the synthetic template resulting in the users and use it to access the system in an
best match score (Sbest (Di, Ti j)) is retained. The unauthorized manner. For e.g., a biometric data
template (Ti best) with best score is then modified via associated with a specific application can be used in
the following four operations: another application (using a fingerprint for accessing
(i) the r, c and  values of an existing minutia are medical records instead of the intended use of office
perturbed. door access control).

IJASR|VOL 01|ISSUE 07|2015 www.ssjournals.com

 Rubal Jain and Chander Kant / Attacks on Biometric Systems: An Overview 287

v. Coercion: An imposter threatens the legitimate user Cipher text can be mapped back to plain text only with
to take the control of the system in his own hands. For the help of decrypting key. If imposter obtains the
e.g., an ATM user can be threaten at gunpoint. cipher text he/she cannot extract useful information
vi. Collusion: It occurs when root or super user with from it without the help of decrypting key. We use
high privileges makes wrong use of his privileges and cryptographic systems to avoid dictionary attacks that
modify the parameters of system illegally. A root user can easily breaks the security of simple password
with wide access privileges has right to access all the based authentication systems. Biometric
system’s resources. cryptosystems are subdivided into key generation and
All the threats discussed above are used to key binding.
make dangerous attacks on biometric-based  Key generation: In this helper data is only obtained
authentication systems. These attacks are most from the biometric traits and the cryptographic key is
probably made for financial purposes such as hacking directly generated from the helper data.
a bank account secured with biometric authentication  Key release: In this helper data is obtained by
to withdraw an amount from it, so it is necessary to binding a key with biometric template.
resist these attacks. In following section 6, the author 6.3. Steganography and Watermarking
has discussed about various template protection Steganography means covered writing. It
techniques to resist attacks. refers to the process in which cover image is used to
hide the original data [10]. Watermarking technology
6. Techniques to resist attacks is the embodiment of steganography. Steganography
All the techniques used for biometric and watermarking are used to prevent attacks on attack
template protection are known for resisting attacks. points 2 (attack on the channel between sensor and
Some of the known biometric template protection feature extractor) and 7 (attack on channel between
schemes are as follows: matcher and application device). These two techniques
6.1. Liveness detection are same in their hiding method, but differ in the
Liveness detection is a mechanism that is characteristics of the embedded data, host image and
used to detect that input sample feature is provided by medium of data transfer. Watermarking is used in the
live human being or not. It is used to prevent from authentication of ownership claims. Steganography
attacks at sensor. It is an ability to distinguish between can be used for transferring critical biometric
real input sample feature provided by living human information from a client to a server.
being and a fake input feature provided by an artifact 6.4. Cancellable biometrics
[8]. Liveness detection can be applied using software Cancellable biometrics is a technique that
or hardware means. involves intentional and systematic distortion of
 Use of extra hardware to implement liveness biometric template based on a selected non-invertible
detection means to measure various life signs like transform [11]. If transformed template is misplaced
pulse detection, blood pressure, temperature for then it can be cancelled and re-issued by changing
fingerprints and movements of face, eyes for face parameters of template. Cancellable biometrics is used
recognition. The limitation of using extra hardware to prevent form attacks at the attack point 6 (attack on
makes the system too much expensive. system database). It also addresses the issue of non-
 Using software means to use the information already replaceability.
captured to detect life signs. The only used method is
to use information about sweat pores. For this a 7. Conclusion
scanner that can acquire a high-resolution image is In this paper author discussed about a
required. It is practically impossible to reproduce the biometric system along with its modules and then
exact size and position of the pores on an artificial make progress towards various attacks on biometric
mold. system. An author found that most of the attacks
6.2. Biometric cryptosystems makes target to the biometric templates which are
Biometric cryptosystems combines stored in system database. This paper also highlights
biometrics and cryptography to take advantages from various techniques to resist attacks that can be used to
the strengths of both the fields [9]. Cryptography protect biometric templates and also discussed about
provides higher degree of security and biometrics generic security threats to any system. It also gives
eliminates the need to remember any passwords or to idea about steganography, watermarking, cancellable
carry any tokens. In traditional Cryptographic systems, biometrics and biometric cryptosystems techniques to
one or more keys are used to convert a plain text into enhance the integrity of the biometric templates and
cipher text and key is known as encrypting key(s). found that there is no security technique which can

IJASR|VOL 01|ISSUE 07|2015 www.ssjournals.com

Rubal Jain and Chander Kant / Attacks on Biometric Systems: An Overview 288

satisfy all the aspects of an ideal biometric template Template Protection Schemes Used in Preventing
protection scheme. There is still need to do research Adversary Attacks on Biometric Fingerprint
work in this field so that an efficient and foolproof Templates”, proc. IJCTT, 2015; 20 (1).
security technique is established. [6] Uludag U. and Jain A. K., “Attacks on biometric
systems: a case study in fingerprints”, proc. SPIE,
References Security, 2004; Vol. 5306, pp. 622–633.
[1] Tiwalade O. Majekodunmi, Francis E. Idachaba, [7] Jain Anil K., Ross Arun, Uludag Umut,
“A Review of the Fingerprint, Speaker “Biometric template security: challenges and
Recognition, Face Recognition and Iris solutions”, proc. Europeon Signal processing
Recognition Based Biometric Identification conference (EUSIPCO), September 2004.
Technologies”, proc. World Congress On [8] Nalinakshi B.G, Sanjeevakumar M. Hatture,
Engineering 2011; 2. Manjunath S.Gabasavalgi and Rashmi P. Karchi,
[2] Jain Anil K., Ross Arun and Salil Prabhakar, “An “Liveness Detection Technique for Prevention of
Introduction to Biometric Recognition”, proc. Spoof Attack in Face Recognition System”, proc.
IEEE Transactions on circuits and systems for IJETAE, 2013; 3 (12).
video technology, 2004; 14 (1). [9] Christian Rathgeb and Andreas Uhl, “A survey on
[3] Abdulmonam Omar Alaswad, Ahlal H. Montaser, biometric cryptosystems and cancellable
and Fawzia Elhashmi Mohamad, “Vulnerabilities biometrics”, proc. EURASIP Journal on
of Biometric Authentication “Threats and Information Security, March 2011.
Countermeasures”, proc. International Journal of [10] Jasleen Kour and Deepankar Verma,
Information & Computation Technology, 2014; 4 “Steganography Techniques –A Review Paper”,
(10): 947-958. proc. International Journal of Emerging Research
[4] U. Latha and K. Rameshkumar, “A Study on in Management &Technology, 2014; 3 (5).
Attacks and Security Against Fingerprint [11] Supriya V G, S Dr Ramachandra Manjunatha,
Template Database”, proc. International Journal “Chaos based Cancellable Biometric Template
of Emerging Trends & Technology in Computer Protection Scheme-A Proposal”, proc.
Science, 2013; 2 (5). International Journal of Engineering Science
[5] Joseph Mwema, Michael Kimwele, Stephen Invention, 2014; 3 (11):14-24.
Kimani, “A Simple Review of Biometric

IJASR|VOL 01|ISSUE 07|2015 www.ssjournals.com