SECURITY ANALYSIS OF MOBILE AUTHENTICATION USING QR-CODES

David C. Wyld et al. (Eds) : NETCOM, NCS, WiMoNe, CSEIT, SPM - 2015
pp. 151–160, 2015. © CS & IT-CSCP 2015 DOI : 10.5121/csit.2015.51612
SECURITY ANALYSIS OF MOBILE
AUTHENTICATION USING QR-CODES
Siwon Sung1,2
, Joonghwan Lee1,2
, Jinmok Kim1,2
, Jongho Mun2
and
Dongho Won2
1
Samsung Electronics, Suwon, Korea
{siwon.sung, joonghwan.lee, jinmok.kim}@samsung.com
2
College of Information and Communication Engineering,
Sungkyunkwan University, Suwon, Korea
{jhmoon, dhwon}@security.re.kr
ABSTRACT
The QR-Code authentication system using mobile application is easily implemented in a mobile
device with high recognition rate without short distance wireless communication support such
as NFC. This system has been widely used for physical authentication system does not require a
strong level of security. The system also can be implemented at a low cost. However, the system
has a vulnerability of tampering or counterfeiting, because of the nature of the mobile
application that should be installed on the user’s smart device. In this paper we analyze the
vulnerabilities about each type of architectures of the system and discuss the concerns about the
implementation aspect to reduce these vulnerabilities.
KEYWORDS
Authentication, Security, QR-Code, Mobile
1. INTRODUCTION
The one-dimensional barcodes have been used in limited areas such as identifier of goods due to
the limitation of information capacity. The two-dimensional barcode has been widely used in
various fields because of great increase of the information capacity. Thus the user authentication
system using the QR-Code becomes possible. Therefore the widely use of smart devices has
promoted to spread the QR-Code authentication system. Users could generate the QR-Code
image for authentication easily through mobile application on their smart devices. The system is
mainly used in a low cost authentication system such as gate access control in buildings and
unmanned book rental system in libraries. However, this system is vulnerable to attempts to
disguise access to someone else’s identity because of the nature of easily replicable QR-Code. In
case of mobile device, it is very easy to duplicate through screen capture and be transferred to
other user via instant messenger. Thus it is very difficult to eradicate these user violations
because of the limitations by those characteristics. In order to overcome these vulnerabilities,
various considerations such as security policy, authentication protocols, network security are
required. In addition, the system needs also to consider reverse engineering because the
vulnerable attributes of the mobile application. The reverse engineering can occur on the various
components to configure the system. Particularly the mobile application running on the smart

152 Computer Science & Information Technology (CS & IT)
device platform is extremely vulnerable to reverse engineering. Even if an application use very
secure authentication scheme, the security might not meet the security levels by lower
implementation maturity. Whereas the system uses vulnerable methods, fine implementation
recommendations help to prevent those vulnerabilities. In this paper, we suggest some
recommendations in implementation points of view to prevent those threats. The remainder of the
paper is organized as follows. In Section 2, related works are presented. We introduce the general
architecture of the QR-Code authentication system using mobile application in Section 3. The
attacker models are described in Section 4. The vulnerabilities in the software implementation
point of the view are discussed in Section 5. The result of the forgery attack on real system is
reported in Section 6. Then the countermeasure to reduce this software threats is given in Section
7. Finally, we conclude this paper in Section 8.
2. RELATED WORK
QR-TANs is a transaction authentication technique based on QR-Code, allow the user to directly
validate the content of a transaction within a trusted device [1]. Liao et al. [2] proposed a scheme
based on two-dimensional barcode not only eliminates the usage of the password verification
table, but also is a cost effective solution. Lee et al. [3] proposed an authentication system used
Mobile OTP with the combination of QR-Code. Oh et al. [4] suggests a technology to authorize
users three kinds of QR-Code. Kao et al. [5] try to develop a safe and efficient authentication way
by using mobile device and implement it for access control system that enhances the security of
physical access control systems. 2CAuth is a two factor authentication scheme that enhances
secure usage of application information and preserves usability, without sacrificing user’s privacy
[6]. Lee et al. designed secured QR-Login user verification protocol for smart devices that are
ready to communicate with QR-Code and proposed a way to keep critical data safe when using
the Internet [7]. Kale et al. proposed a anti phishing single sign-on authentication model using
QR-Code [8]. Malik et al. introduced the idea of a one-time password, which makes unauthorized
access difficult for unauthorized users [9].
3. ARCHITECTURE OF AUTHENTICATION SYSTEM
The QR-Code authentication system using mobile application (BAS-MA) consists of five
participants: the users U, the mobile application MA, QR-Code generator QG, the service
provider SP and authentication center AC. We summarize the notations and acronyms used in this
paper in Table 1.
Table 1. Notations and Acronyms
Ui User
Ci User credentials for Ui
Bi QR-Code for authentication issued to Ui
QG QR-Code generator
SP Service provider
AC Authentication center
BAS-MA QR-Code authentication system using mobile application
AC-AR Authentication center based architecture

Computer Science & Information Technology (CS & IT)
MA-AR Mobile application center based architecture
Ait Internal attacker
Aex External attacker
BAS-MA is classified into two kinds of architectures based on the location of the
• Authentication Center based Architecture
• Mobile Application based Architecture
In the AC-AR, QG is located in
location. The basic configuration of
Ui can get a valid access to SP is the following:
1. Ui is willing to use the SP.
the system.
2. MA requires Bi for authentication to
3. AC retrieves user data from
carrying Ci.
4. QG transform the Ci to Bi without any modification on
5. AC passes through the Bi to
6. MA displays the Bi on its screen and submits to
7. SP decodes the Bi and extracts
8. AC verifies Ci and return with an authentication callback.
The Bi is generated in the AC on remote. The
displays Bi on its screen. The MA
is just provided with a bitmap image. If the
secured to the reversing attack because they are created and managed on
Figure 1. Basic procedure of Authentication Center based Architecture
Otherwise, the QG is associated with
to interact with the external QG
Computer Science & Information Technology (CS & IT)
Mobile application center based architecture
Internal attacker
ternal attacker
is classified into two kinds of architectures based on the location of the QG
Authentication Center based Architecture
Mobile Application based Architecture
is located in AC that could be regarded as a secure server in the remote
on. The basic configuration of AC-AR is shown in Figure 1. The procedure through which a
is the following:
. Ui get authentication through the user authentication scheme of
for authentication to AC.
retrieves user data from User Data Storage then calculates Ci. AC asks the
without any modification on Ci and returns a Bi.
to MA via secure channel.
on its screen and submits to SP.
and extracts Ci. SP asks the AC the verification carrying the C
and return with an authentication callback.
on remote. The MA has just downloaded the image Bi
MA does not have any logic for generating Bi or processing
is just provided with a bitmap image. If the MA is authorized to download the Bi, the
versing attack because they are created and managed on AC in remote.
Basic procedure of Authentication Center based Architecture
is associated with MA tightly in MA-AR. The MA includes QG itself or is able
QG in the same device. Figure 2 shows the operation flows that the
153
QG :
that could be regarded as a secure server in the remote
gure 1. The procedure through which a
get authentication through the user authentication scheme of
asks the QG a Bi
Ci.
from AC and
or processing Ci and
, the Ci can be
in remote.
itself or is able
gure 2 shows the operation flows that the

MA is granted the permission for SP. The procedure on MA-AR is different to that of the AC-AR,
yet it has the following differences:
1. Ui try to get a permission to access SP.
2. MA requests a Ci to generate Bi.
3. AC assemble Ci using the User Data Storage and return to MA.
4. MA askes QG existed in mobile device a Bi carrying Ci.
5. QG generate Bi containing Ci then pass it to MA.
6. MA gives the Bi to SP.
7. SP open the Bi and find Ci then SP request a verification to the AC with the Ci.
8. AC checks the equality of the Ci and response.
The MA is responsible to manage the Ci in secure and to generate the Bi itself. The QG is exposed
to threat to get to be attempt software attack.
Figure 2. Basic procedure of Mobile Application based Architecture
4. ATTACK MOTELS
The attackers who threaten the BAS-MA can be divided into two groups:
4.1. The Internal Attacker
The internal attacker Ait is a member of the organization provided services by SP and has the
correct permissions can access the SP. Ait may be a malicious user himself willing to access to SP
by the identity of the other user Uj . The Ait also tries to let an unauthorized person to get the
authentication to the system by sharing his identity. The Ait can try various attacks on the system
because they know very well about the BAS-MA, and can examine the MA on their mobile
devices. Ait is able to perform repeated incorrect attempts to authentication system with less doubt
of the system administrator.

Computer Science & Information Technology (CS & IT) 155
4.2. The External Attacker
Otherwise, the external attacker Aex is not participated in the organization. Aex is the attacker
would like to acquire authentication to the system even if they don’t have any right permissions.
Depending on the mobile application distribution policy, the Aex may have difficulties to obtain
the installable packages or binary of the mobile applications. For that reason, Aex distribute a
malware to find vulnerabilities in order to steal someone’s Ci and Bi.
5. VULNERABILITY ANALYSIS
5.1. QR-Code Cloning
In the BAS-MA, the Bi is the key object to achieve the authentication. However, the Bi could be
cloned very easily on MA through screen capture. The Ait might cause the masquerade attack
transferring their image to others via an instant messenger. The Aex also try to get the copies of
the Bi. In fact, it is impossible to completely prevent such cloning because of the nature of the
QR-Code that should be exposed on the screen. The Aex could even take a picture of the Bi using a
separate camera.
5.2. Authentication Hijacking
The permanent authentication such as automatic login is usually enabled for the convenience use
after once successful authentication. Even if the authentication protocol do not support long term
session, the developer tries to implement virtual session by background authentication. In order to
do it, the keeping of the user authentication cookies such as user’s identity and password in the
internal cache are required. These important data in the cache can be easily exposed by the
software attack. The attackers are able to hijack the user authentication through the cached data.
5.3. Stored Data Exfiltration
The Ci is stored in the non-volatile storage area under MA in the MA-AR. The attackers can
extract the values through software attack. If the device is rooted, the attack is easier to occur.
Because of that, it is recommended that the value is not stored permanently or data should be
stored encrypted. Once the encryption key is exposed, the protection using cryptography is
useless. After that, the attackers are able to generate the Bi without big difficulty. The data in
local storage is very vulnerable. In rooted device, the malware get the read permission under
other application’s control. The attacker also can access the file system on the device directly or
inspect the runtime memory.
5.4. Internal Algorithm Disclosure
The developers usually assume that the compiled source code will be safe from the logic
disclosure. However, the powerful tools such as decompiler and logic analyzer are able to
disclose the internal logic. The decompiler is able to recover the source codes using the binaries.
The logic analyzer can draw the flow graph to help the understanding the logic. It can expose
which cryptographic algorithms are used and how to assemble the parameters to construct
credentials. In case of the use of the original designed two-dimensional barcode formats for the
security purpose, the algorithms should not be disclosed and the software modules for the
algorithms should not to be exposed and extracted for other application. If it is possible, the
attackers can produce forgery algorithms to perform same work.

5.5. Network Message Eavesdropping
The attackers have various methods to intercept the messages on the mobile device. The general
purpose smart devices can be installed any application without any restrictions. The Ait could
install a network packet monitoring tool on their mobile device to eavesdrop the messages
between MA and the AC. While the mobile platform operates under some policies that restrict the
packet monitoring permissions, the Ait can attempt to bypass the restriction by using rootkit. In
rooted device, all application could obtain root permission.
5.6. IPC Message Eavesdropping
In the MA-AR, the QG can be located in separated module from MA. This is the case in order to
use the common module to generate barcode provided by the platform. In this case, the IPC
(Inter Process Communication) communication is generally used. If the IPC channel is not
secure, messages Ci through IPC can be exposed to malware. For example, the Broadcast is
popularly used in Android platform to pass the values to other application. However the implicit
broadcast message could be broadcasted to all the application in the device not excluding
malwares.
5.7. Communication Protocol Vulnerabilities
The secure protocols such as SSL/TLS prevent the leakage of the network packet. In this case, the
attacker can manipulate the victim’s network packet to pass through the malicious HTTP Proxy
to incapacitate the secure communication. The HTTP Proxy can perform the man-in-the-middle
attack to obtain the secrete key of the channel. Callegati et al. presented the man-in-the middle
attack to the HTTPS through ARP Spoofing [10]. Aex also can install rogue access point in the
near place that the authentication communication usually happened. Because the authentication
procedure has occurred in fixed places such as the gateway of the buildings or in front of the
unmanned machines, Aex could intercept the communications to proceed authentication through
the rogue access point. Marlinspike et al. showed SSL Strip Attack that do a man-in-the-middle
attack on SSL connections [11].
The Internal logic for communication in the software should be hidden as much as possible. The
address of the destination server, Restful API names, the configuration of the parameters and the
keyword for the message format are usually hard-coded inside mobile applications. The skilful
attacker can reconstruct the protocol completely using a piece of this information.
6. FORGERY ATTACK RESULTS
We performed a forgery attack based on the discussed vulnerability in the previous section
against the real world BAS-MA. The used attacker model was the internal attacker Ait and the
architecture of the targeted system was MA-AR. The Ait registered his device to the AC through
user authentication phase. Then the MA downloaded the Ci and generated Bi for displaying. The
Bi had expired in 5 minutes to prevent replay attack. The MA was able to create a new Bi every 5
minutes without additional network communication with AC. This means that the targeted MA
had stored not only user’s credential but also the refresh logic in itself. The targeted BAS-MA also
used its original two-dimensional barcode format (Inter-Code). The attack procedure is described
as follows. The Ait intercepted the packets of the user authentication phase. The packets were not
encrypted and the network channel also wasn’t secure. Ait could find a plain XML document
containing Ci in the intercepted packets. The Ci had the member identification number and

organization code. The AC received the user’s mobile phone number as parameter then returned
the Ci. Anyone with a member’s phone number can obtain member’s credential data. The QG
was included in the install package of the MA. The binary package of MA could be extracted from
the targeted device without root permissions. The Ait could obtain the QG by uncompressing the
package. The QG was a kind of shared object for the targeted device platform. Ait tried to
decompile the binaries to understand the interface and the parameters of the QG. Thus the
interface module was implemented in Java, the analysis didn’t need high difficulty. The free
decompile and analysis tools were available without commercial license. Ait could find the logic
that how to assemble the Ui and timestamps to make the parameters for the QG. The shared
object could be linked and executed with the forgery application on the same platform. Finally,
Ait could produce the forgery application that received the victim’s mobile number as parameter,
then displayed the cloned Bi as Inter-Code format. The Bi could be accepted by the verification on
the real world BAS-MA without any restrictions.
7. RECOMMENDATIONS FOR SECURE IMPLEMENTATION
In Section 5, the various software vulnerabilities on the BAS-MA caused by immaturity
implemented were discussed. The case study of the forgery attack through the software
vulnerabilities on real system was reported in Section 6. Based on this, we suggest the
countermeasures against the premature software implementations.
7.1. Interfered Screen Capturing
The screen capture function is usually enabled on common smart devices. It is recommended to
block the function while the MA shows the Bi. The watermarking on the captured image is also
good to give a warning to the Ait. It might be able to stop the harmful behaviour of the Ait.
7.2. Expiring QR-Code Available Period
The expiration of already issued Bi could cancel the replay attack. Some of salt value has to be
mixed with the Ci before QG generates the Bi. Lee et al. [7] proposed a timestamps based
authentication scheme suited for mobile device environment, in which users can be authenticated
using a QR-Code. A secure authentication system proposed by Shamal et al. [12] that uses a two
factor authentication by combining a password and a camera equipped mobile phone, where
mobile phone is acting as a authentication token. If the nonce for expiring is hard-coded in the
MA, it should be obfuscated against to be disclosed by malicious inspection.
7.3. Maintain Local Storage Cleanliness
The candidate data willing to be stored in local storage should be reviewed carefully. The local
storage is always in danger to be examined by attackers. If it is possible, the empty of the local
area is recommended. The session between MA and AC should be implemented based on secure
protocol. The developer should not arbitrarily implement features beyond the protocol.
7.4. Deliberate Data Storing
The cryptography does not guarantee complete data protection. However, storing data without
encryption is more dangerous. The encryption key supplied from trusted party over secure
channel is better than included in the binary package. The white-box cryptography helps to hide
the key inside the application [13]. If the decryption of the stored value is not required, one-way

hash function could be a good choice. In other hand, hardware-backed approach such as ARM
TrustZone [14] offers a high level security assurance. It provides a completely isolated memory
area and data storages physically or logically. However, the application requires a close
collaboration with hardware manufacturers to use this approach, and there are many restrictions
on the implementation.
7.5. Obfuscation
The software obfuscation is a process to transform the source code into obfuscated code. The
variable names are replaced by meaningless name and the execution flows become skewed
without any logical error. It is a time consuming and laborious process to analyze the obfuscated
binary. The obfuscation is the most effective and comprehensive defence against reversing attack.
The obfuscation tool produces a obfuscated result with the source code or compiled binary. This
can be fully integrated and automated build process. In fact, the developers do not need to worry
to apply the obfuscation. It can be easily accomplished using the 3rd-party tools. In Android, the
ProGuard is included as default in build system. Ensure good performance, it is good to use a
commercial tool. The high performance obfuscation tools replace the strings, the constants and
the hard-coded encryption keys by encrypted themselves. The obfuscation of control-flow had
been proposed by Chow et al. [15]. In case that the MA includes the encryption key, the
obfuscated use of key storage is the essential of the successful encryption.
7.6. Securing Messaging over IPC
When the mobile application structure uses the external QG through IPC, the IPC message
carrying Ci to QG has to be attention not to be eavesdropped. According to the type platforms, the
secure communication between processes might not be supported. The Ci should not be sent
through a non-secure IPC. Chin et al. examine Android application interaction and identify
security risks in application components. They provide a tool, ComDroid, that detects application
communication vulnerabilities [16]. The TaintDroid is an extension to the Android mobile-phone
platform that tracks the flow of privacy-sensitive data through third-party applications [17].
7.7. Detecting Rootkit and Malware
The attack on the rooted device is always more critical. The restricted policies are able to be
cancelled easily. The applications are free to invade the protected area each other. It is
recommended that let MA not launch on the rooted device. A rootkit is a set of malicious tools, in
order to achieve the top level privilege of the system. Kruegel et al. presents a technique that
exploits binary analysis to ascertain, at load time [18]. In case of the difficulty to implement to
detect rootkit or malware itself, the use of 3rd-party solution could be alternative.
7.8. Detecting Rogue Access Point
The rogue AP is a wireless access point installed on a wired enterprise network without
authorization from the network administrator. They are installed by a legitimate user who is
unaware of its security implications or easily smuggled onto enterprise premises by an outsider. It
also allows the attacker to conduct a man-in-the-middle attack. The white-list based detection
approach [19] is appropriate to BAS-MA. The valid access point nearby SP could be listed up. It
also can be an alternative to use the mobile network only such as 3G or LTE instead of WIFI.

7.9. Restricting Allowed Device
The number of authentication enabled devices should be limited against the hijacking. Thus the Bi
could be downloaded in only predefined mobile device. The device unique key is useful to
implement the restrictions. The IMEI (International Mobile Station Equipment Identity) is a
unique number to identity 3GPP (GSM, LTE) mobile device. The IMEI could be a seed of the
device unique key. The key is transmitted to the server along with the request for Bi. The server
verifies the equality of the unique key from mobile device and the registered key in the server.
Therefore, even if an attacker acquires a login cookie, it cannot be used in other devices. Of
course the device unique key should not be stored in the local cookies.
7.10. Forged Verification
The executable is recommended to be signed by trusted certificate authority. This allows that the
platform is able to verify whether the binary is forged. The application can verify the certification
itself using platform API. The signature of the binary package could be accessed using the
PackageManager in Android.
7.11. Limiting Distribution Path
The public on-line application store such as Google Play Store can be accessed without any
limitation to download some mobile applications registered in the store. If the MA is registered on
the public store, Aex can download and install on their own devices in order to analyze the internal
structures. The applications for internal users, such as MA are recommended to distribute on a
private path. The system operators can send a URL containing the application downloadable link
to the valid user via SMS or E-mail.
8. CONCLUSION
We discussed the vulnerabilities on the QR-Code authentication system caused by an immature
software implementation. The main issue of the software security on the QR-Code authentication
system is the protection of the logic and data around the QR-Code generator. These
vulnerabilities can be overcome by a mature implementation. The improvisation implementation
beyond the specification of the protocol is to be eradicated. The locally stored data should be
cherry-picked and encrypted. The obfuscation can complete the robust software implementation.
With these recommendations, the QR-Code authentication system would be a good solution for
physical access system.
REFERENCES
[1] Starnberger, G., Froihofer, L., & Göschka, K. M., (2009) “QR-TAN: Secure mobile transaction
authentication”, Availability, Reliability and Security, pp578-583.
[2] Liao, K. C., & Lee, W. H., (2010). “A novel user authentication scheme based on QR-code” Journal
of Networks, No.5(8), pp937-941.
[3] Lee, Y. S., Kim, N. H., Lim, H., Jo, H., & Lee, H. J., (2010) “Online banking authentication system
using mobile-OTP with QR-code”, Computer Sciences and Convergence Information Technology,
pp644-648.
[4] Oh, D. S., Kim, B. H., & Lee, J. K., (2011) “A study on authentication system using QR code for
mobile cloud computing environment”, Future Information Technology, pp500-507
[5] Kao, Y. W., Luo, G. H., Lin, H. T., Huang, Y. K., & Yuan, S. M., (2011) “Physical access control
based on QR code” Cyber-enabled distributed computing and knowledge discovery pp285-288

[6] Harini, N., & Padmanabhan, T. R., (2013) “2CAuth: A new two factor authentication scheme using
QR-code”, International Journal of Engineering and Technology, No.5, pp1087-1094.
[7] Lee, Y., Kim, J., Jeon, W., & Won, D., (2012) “Design of a simple user authentication scheme using
QR-code for mobile device”, Information Technology Convergence, Secure and Trust Computing,
and Data Management, pp241-247.
[8] Kale, V., Nakat, Y., Bhosale, S., Bandal, A., & Patole, R. G., (2015) “A Mobile Based Authentication
Scheme Using QR Code for Bank Security”
[9] Malik, J., Girdhar, D., Dahiya, R., & Sainarayanan, G., (2014) “Multifactor Authentication Using a
QR Code and a One-Time Password”, Journal of Information Processing Systems, No.10.
[10] Callegati, F., Cerroni, W., & Ramilli, M., (2009) “Man-in-the-Middle Attack to the HTTPS
Protocol”, IEEE Security & Privacy, pp78-81.
[11] Marlinspike, Moxie, (2009) “New tricks for defeating SSL in practice”, BlackHat DC.
[12] Shamal, S., Monika, K., & Neha, N., (2014) “Secure Authentication for Online Banking Using QR
Code”, IJETAE–International Journal for Emerging Technology and Advance Engineering
[13] Chow, S., Eisen, P., Johnson, H., & Van Oorschot, P. C., (2003) “White-box cryptography and an
AES implementation”, Selected Areas in Cryptography, pp250-270.
[14] Winter, J., (2008) “Trusted computing building blocks for embedded linux-based ARM trustzone
platforms”, Proceedings of the 3rd ACM workshop on Scalable trusted computing, pp21-30.
[15] Chow, S., Gu, Y., Johnson, H., & Zakharov, V. A., (2001) “An approach to the obfuscation of
control-flow of sequential computer programs”, Information Security, pp144-155.
[16] Chin, E., Felt, A. P., Greenwood, K., & Wagner, D., (2011) “Analyzing inter-application
communication in Android”, Proceedings of the 9th international conference on Mobile systems,
applications, and services, pp239-252.
[17] Enck, W., Gilbert, P., Han, S., Tendulkar, V., Chun, B. G., (2014) “TaintDroid: an information-flow
tracking system for realtime privacy monitoring on smartphones”, ACM Transactions on Computer
Systems.
[18] Kruegel, C., Robertson, W., & Vigna, G., (2004), “Detecting kernel-level rootkits through binary
analysis”, Computer Security Applications Conference, pp91-100.
[19] Park, J., Park, M., & Jung, S., (2013). “A whitelist-based scheme for detecting and preventing
unauthorized AP access using mobile device”, The Journal of Korean Institute of Communications
and Information Sciences, No.38, pp632-640.

SECURITY ANALYSIS OF MOBILE AUTHENTICATION USING QR-CODES

More Related Content

What's hot (19)

Similar to SECURITY ANALYSIS OF MOBILE AUTHENTICATION USING QR-CODES (20)

More from cscpconf (20)

Recently uploaded (20)

SECURITY ANALYSIS OF MOBILE AUTHENTICATION USING QR-CODES