1110 3844 PDF
1110 3844 PDF
1110 3844 PDF
145
Abstract
Passwords provide security mechanism for authentication and
protection services against unwanted access to resources. A graphical
based password is one promising alternatives of textual passwords.
According to human psychology, humans are able to remember pictures
easily. In this paper, we have proposed a new hybrid graphical
password based system, which is a combination of recognition and
recall based techniques that offers many advantages over the existing
systems and may be more convenient for the user. Our scheme is
resistant to shoulder surfing attack and many other attacks on graphical
passwords. This scheme is proposed for smart mobile devices (like
smart phones i.e. ipod, iphone, PDAs etc) which are more handy and
convenient to use than traditional desktop computer systems.
Keywords: Smart Phones, Graphical Passwords, Authentication,
Network Security.
1. Introduction
One of the major functions of any security system is the control
of people in or out of protected areas, such as physical buildings,
information systems, and our national borders. Computer
systems and the information they store and process are valuable
resources which need to be protected. Computer security
systems must also consider the human factors such as ease of a
use and accessibility. Current secure systems suffer because they
mostly ignore the importance of human factors in security [1].
An ideal security system considers security, reliability, usability,
and human factors. All current security systems have flaws
which make them specific for well trained and skilled users
only. A password is a secret that is shared by the verifier and the
customer. Passwords are simply secrets that are provided by the
user upon request by a recipient. They are often stored on a
server in an encrypted form so that a penetration of the file
system does not reveal password lists [2]. Passwords are the
most common means of authentication because they do not
require any special hardware. Typically passwords are strings of
letters and digits, i.e. they are alphanumeric. Such passwords
have the disadvantage of being hard to remember [3]. Weak
passwords are vulnerable to dictionary attacks and brute force
IJCSI International Journal of Computer Science Issues, Vol. 8, Issue 5, No 2, September 2011
ISSN (Online): 1694-0814
www.IJCSI.org
146
IJCSI International Journal of Computer Science Issues, Vol. 8, Issue 5, No 2, September 2011
ISSN (Online): 1694-0814
www.IJCSI.org
4. Related Work
Haichang Gao et al. [55] have proposed and evaluated a new
shoulder-surfing resistant scheme called Come from DAS and
Story (CDS) which has a desirable usability for PDAs. This
scheme adopts a similar drawing input method in DAS and
inherits the association mnemonics in Story for sequence
retrieval. It requires users to draw a curve across their password
images (pass-images) orderly rather than click directly on them.
The drawing method seems to be more compatible with peoples
writing habit, which may shorten the login time. The drawing
input trick along with the complementary measures, such as
erasing the drawing trace, displaying degraded images, and
starting and ending with randomly designated images provide a
good resistance to shoulder surfing. A user study is conducted to
explore the usability of CDS in terms of accuracy, efficiency and
memorability, and benchmark the usability against that of a
147
Pure Recall
Based
Cognitive
Authenticatio
n [22]
Use Your
Illusion [23]
Cued Recall
Based
Android
Screen
Unlock [34]
GrIDsure
[35]
Hybrid Schemes
Jimmy
Scheme [44,
45]
Suos
Scheme [46]
Story [24]
PassShapes
[33]
PassPoints
[41, 42, 43]
Dj vu [1]
DAS [31]
CCP [47]
PassFace[10,
25]
BDAS [32]
PCCP [48]
PassGo [5]
Inkblot
Authenticatio
n [49]
Photographic
Authenticatio
n [28]
YAGP [36]
Convex Hull
Click [29]
Haptic
Password
[37]
GPI/ GPIS
[30]
Passdoodle
[38, 39]
CDS [55]
Two Step
Authenticat
ions [59]
Our
Proposed
System
3D Scheme
[50]
Passlogix
[40]
Picture
Password
Fig. 2 Classification of Graphical Password Based Systems
IJCSI International Journal of Computer Science Issues, Vol. 8, Issue 5, No 2, September 2011
ISSN (Online): 1694-0814
www.IJCSI.org
5. Problem Domain
There are many problems with each of the graphical based
authentication methods. These are discussed below:
5.1 Problems of Recognition Based Methods:
Dhamijia and Perrig proposed a graphical password based
scheme Dj Vu, based on Hash Visualization technique [11].
The drawback of this scheme is that the server needs to store a
large amount of pictures which may have to be transferred over
the network, delaying the authentication process. Another
weakness of this system is that the server needs to store the
seeds of portfolio images of each user in plaintext. Also, the
process of selecting a set of pictures from picture database can
be tedious and time consuming for the user [7]. This scheme was
not really secure because the passwords need to store in database
and that is easy to see.
Sobrado and Birget developed a graphical password technique
that deals with the shoulder surfing problem [3]. In their first
scheme the system displays a number of pass-objects (preselected by user) among many other objects as shown in Fig: 3.
To be authenticated, a user needs to recognize pass-objects and
click inside convex hull formed by all the pass objects. They
developed many schemes to solve the shoulder surfing problem
but the main drawback of these schemes is that log in process
can be slow.
Another recognition based technique is proposed by Man et al
[63]. He proposed a shoulder-surfing resistant algorithm which
is similar to that developed by Sobrado and Birget. The
difference is that Man et al has introduced several variants for
each pass-object and each variant is assigned a unique code.
Thus during authentication the user recognize pre-selected
objects with an alphanumeric code and a string for each passobject. Although it is very hard to break this kind of password
but this method still requires the user to memorize alphanumeric
codes for each pass-object variants.
Passface is another recognition based system. It is argued by
its developer that it is easy for human beings to remember
human faces than any other kind of passwords. But Davis et al
[12] have found that most users tend to choose faces of people
from the same race. This makes the Passface password
148
6. Proposed System
Taking into account all the problems and limitations of graphical
based schemes, we have proposed a hybrid system for
IJCSI International Journal of Computer Science Issues, Vol. 8, Issue 5, No 2, September 2011
ISSN (Online): 1694-0814
www.IJCSI.org
149
Step 6
After stroke merging, the system constructs the hierarchy.
Step 7
Seventh step is the sketch simplification.
Step 8
In the eighth step three types of features are extracted from the
sketch drawn by the user.
Step 9
The last step is called hierarchical matching.
During registration, the user selects the user name and a textual
password in a conventional manner and then chooses the objects
as password. The minimum length for textual password is L=6.
Textual password can be a mixture of digits, lowercase and
uppercase letter. After this the system shows objects on the
screen of a PDA to select as a graphical password. After
choosing the objects, the user draws those objects on a screen
with a stylus or a mouse. Objects drawn by the user are stored in
the database with his/her username. In object selection, each
object can be selected any number of times. Flow chart of
registration phase is shown in Figure 6.
During authentication, the user has to first give his username and
textual password and then draw pre-selected objects. These
objects are then matched with the templates of all the objects
stored in the database. Flow chart of authentication phase is
shown in Figure 7. The phases during the authentication like the
pre-processing, stroke merging, hierarchy construction, sketch
simplification, feature extraction, and hierarchical matching are
the steps proposed by Wing Ho Leung and Tsuhan Chen in their
paper [13]. They propose a novel method for the retrieval of
hand drawn sketches from the database, finally ranking the best
matches. In the proposed system, the user will be authenticated
only if the drawn sketch is fully matched with the selected
objects template stored in the database. Pre-processing of hand
IJCSI International Journal of Computer Science Issues, Vol. 8, Issue 5, No 2, September 2011
ISSN (Online): 1694-0814
www.IJCSI.org
150
In case, if user draws very large or a very small sketch then the
system performs size normalization which adjusts the symbols
or sketches to a standard size. The Stroke merging phase is use
to merge the strokes which are broken at end points. If the end
points are not close, then that stroke is considered as open stroke
and it may be merged with another open stroke if the end point
of one stroke is close to the end point of the other. The strokes
are then represented in a hierarchy to simplify the image and to
make it meaningful for further phases [13]. In the next step of
sketch simplification, a shaded region is represented by a single
hyper-stroke. After sketch simplification three types of features
are extracted from the user re-drawn sketch. These features are
hyper stroke features, Stroke features, and bi-stroke features.
In the last step of hierarchical matching, the similarity is
evaluation the top to bottom hierarchical manner. The user is
allowed to draw in an unrestricted manner. The overall process
is difficult because free hand sketching is a difficult job. The
order in which the user has selected the objects does matter in
our proposed system i.e. during the authentication phase, the
user can draw his pre-selected objects in the same order as he
had selected during the registration phase. So, in this way the
total combinations of each password will be 2n 1, n being the
number of objects selected by the user as password during the
registration phase.
IJCSI International Journal of Computer Science Issues, Vol. 8, Issue 5, No 2, September 2011
ISSN (Online): 1694-0814
www.IJCSI.org
151
IJCSI International Journal of Computer Science Issues, Vol. 8, Issue 5, No 2, September 2011
ISSN (Online): 1694-0814
www.IJCSI.org
Graphical Password
Schemes/ Systems
Blonders Scheme [62]
DAS [31]
BDAS [32]
Qualitative DAS [65]
Syukri Algorithm [64]
PassPoints [41, 42, 43]
PassFace [10, 25]
PassGo [5]
Passlogix [40]
PassMap [66]
Passdoodle [38, 39]
Viskey SFR
Perrig and Song [11]
Sobrado and Birget [3]
Man et al Scheme [63]
Picture Password Scheme [60]
CDS [55]
WIW [57]
Association based scheme
[58]
Dj Vu [1]
Haptic Password Scheme [37]
YAGP [36]
Photographic Authentication
[28]
Two Step Authentication [59]
Our Proposed System
Type of Scheme
Brute Force
Attack
Dictionary
Attack
152
Recognition Based
Y
N
Y
N
Table Recall
1: Comparison
Pure
Based Of Graphical
N password Schemes
Y resistant toYdifferent AttacksN
Pure Recall Based
N
Pure recall Based
N
Pure recall Based
N
Y
Y
N
Cued Recall Based
Y
N
Y
N
Recognition Based
Y
Y
Y
N
Pure Recall Based
Y
Cued Recall Based
Y
N
Y
N
Pure Recall Based
Y
N
N
Pure Recall Based
N
Pure Recall Based
Y
N
Y
N
Recognition Based
Y
N
Y
N
Recognition Based
Y
N
Y
N
Recognition Based
Y
N
N
Y
Recognition Based
Y
N
Y
N
Hybrid
Recognition Based
-
Phishing Attack
or
Social
Engineering
Y
Y
Y
Y
Y
Y
Y
Y
Y
N
Y
Y
Y
Y
N
N
N
N
N
N
N
N
N
N
N
N
-
Recognition Based
Recognition Based
Pure Recall Based
Pure Recall Based
Y
Y
Y
Y
Y
Y
Recognition Based
Y
Y
Y
Y
Y
Note: Y= Yes resistant to attack N=No not resistant to attack
N
Y
Y
Y
Hybrid
Hybrid
References
[1] Rachna Dhamija and Adrian Perrig, Deja Vu: A User Study.
Using Images for Authentication In Proceedings of the 9th
USENIX Security Symposium, August 2000.
[2] Authentication: http://www.objs.com/survey/authent.htm [Last
Visited on 15/05/2011].
IJCSI International Journal of Computer Science Issues, Vol. 8, Issue 5, No 2, September 2011
ISSN (Online): 1694-0814
www.IJCSI.org
153
IJCSI International Journal of Computer Science Issues, Vol. 8, Issue 5, No 2, September 2011
ISSN (Online): 1694-0814
www.IJCSI.org
154