Abstract Current authentication systems suffer from many weaknesses. Textual passwords are commonly used; however, users do not follow f ollow their requirements. Users tend to choose meaningful words from dictionaries, which make textual passwords easy to break and vulnerable to dictionary or brute force attacks. any available graphical passwords have a password space that is less than or equal to the textual password space. !mart cards or tokens can be stolen. any biometric authentications have been proposed; however, users tend to resist using biometrics because of their intrusiveness and the effect on their privacy. oreover, biometrics cannot be revoked. "n this paper, we present and evaluate our contribution, i.e., the #$% password. The #$% password is a multifactor authentication scheme. To be authenticated, we present a #$% virtual environment where the user navigates and interacts with various ob&ects. The sequence of actions and interactions toward the ob&ects inside the #$% environment constructs the user's #$% password. The #$% password can combine most existing authentication schemes such as textual passwords, graphical passwords, and various types of biometrics into a #$% virtual environment. The design of the #$% virtual environment and the type of ob&ects selected determine the #$% password key space.
Acknowledgement (s " write this acknowledgement, " must clarify that this is not &ust a formal acknowledgement but also a sincere note of thanks and regard from my side. " feel a deep sense of gratitude and affection for those who were associated with this seminar. )ithout their co$operation and guidance this seminar could not have been conducted properly. " am also indebted to my friends and family for their constant support and their priceless reviews which helped me to take this seminar to its current level.
Acknowledgement (s " write this acknowledgement, " must clarify that this is not &ust a formal acknowledgement but also a sincere note of thanks and regard from my side. " feel a deep sense of gratitude and affection for those who were associated with this seminar. )ithout their co$operation and guidance this seminar could not have been conducted properly. " am also indebted to my friends and family for their constant support and their priceless reviews which helped me to take this seminar to its current level.
TABLE O OF F CONTENTS Abstract *. #% passwords
i *
*.*. "ntroduction
*
*.*. +elated )orks
*.-. !cheme
*.-.*. /verview
*.-.-. !election and "nputs
0
*.-.#. #$% virtual 1nvironment %esign 2uidelines
3
*.-.. (pplications
*-
*.#. !ecurity (nalysis
*#
*.#.*. 4assword !pace !i5e
*
*.#.-. 4assword %istribution 6nowledge
*
*.#.#. (ttacks and Countermeasures
*7
*.. 1xperimental results
-8
*..*. 1xperimental 9irtual #% environment
-8
*..-. User !tudy
-*
-. :iterature +eview
--
#. Conclusion and uture work
-#
References
-<
INTO!"CTION The dramatic increase of computer usage has given rise to many security concerns. /ne ma&or security concern is authentication, which is the process of validating who you are to whom you claimed to be. "n general, human authentication techniques can be classified as=
Textual Knowledge Based What you KNOW
Password Graphical Password
ATM cards
Human Authenticat ion techniques
Token Based -What you HAVE
Keys
ID cards !i"#erpri" ts$
Biometrics What you AE
Pal%pri"t s Ha"d #eo%etry !ace$ Iris$ Voice$ eti"a reco#"itio"
Figure 1 – Human Authentication Techniques - Classification Te#tual $asswords
+ecall$based techniques require the user to repeat or reproduce a secret that the user created before. +ecognition based techniques require the user to identify and recogni5e the secret, or part of it, that the user selected before. /ne of the most common recall$based authentication schemes used in the computer world is textual passwords. /ne ma&or drawback of the textual password is its two conflicting requirements= the selection of passwords that are easy to remember and, at the same time, are hard to guess.
6lein
>-?
collected the passwords of nearly *< 888 accounts that had alphanumerical passwords, and
he reached the following observation= -<@ of the passwords were guessed by using a small yet
well$ formed dictionary of # A *8 words. urthermore, -*@ of the passwords were guessed in the first week and #0 passwords were guessed within the first *< min. 6lein
>-?
stated that by
looking at these results in a system with about <8 accounts, the first account can be guessed in - min and
>-?
showed that even though the full
textual password space for eight$character passwords consisting of letters and numbers is almost *
- A *8 possible passwords, it is easy to crack -<@ of the passwords by using only a small subset of the full password space. "t is important to note that 6lein's experiment was in *338 when the processing capabilities, memory, networking, and other resources were very limited compared to today's technology.
%ra$&'cal $asswords
9arious graphical password schemes have been proposed. 2raphical passwords are based on the idea that users can recall and recogni5e pictures better than words. owever, some of the graphical password schemes require a long time to be performed. oreover, most of the graphical passwords can be easily observed or recorded while the legitimate user is performing the graphical password; thus, it is vulnerable to shoulder surfing attacks. Currently, most graphical passwords are still in their research phase and require more enhancements and usability studies to deploy them in the market.
B'ometr'c s
any biometric schemes have been proposed; fingerprints, palmprints, hand geometry, face recognition, voice recognition, iris recognition, and retina recognition are all different biometric schemes. 1ach biometric recognition scheme has its advantages and disadvantages based on several factors such as consistency, uniqueness, and acceptability. /ne of the main drawbacks of applying biometrics is its intrusiveness upon a user's personal characteristic. oreover, retina biometrical recognition schemes require the user to willingly sub&ect their eyes to a low$intensity infrared light. "n addition, most biometric systems require a special scanning device to authenticate users, which is not applicable for remote and "nternet users.
(! )asswords
The #$% password is a multifactor authentication scheme. "t can combine all existing authentication schemes into a single #$% virtual environment. This #$% virtual environment contains several ob&ects or items with which the user can interact. The type of interaction varies from one item to another. The #$% password is constructed by observing the actions and interactions of the user and by observing the sequences of such actions. "t is the user's choice to select which type of authentication techniques will be part of their #$ % password. This is achieved through interacting only with the ob&ects that acquire information that the user is comfortable in providing and ignoring the ob&ects that request information that the user prefers not to provide. or example, if an item requests an iris scan and the user is not comfortable in providing such information, the user simply avoids interacting with that item. oreover, giving the user the freedom of choice as to what type of authentication schemes will be part of their #$% password and given the large number of ob&ects and items in the environment, the number of possible #$% passwords will increase. Thus, it becomes much more difficult for the attacker to guess the user's #$% password.
ELATE! *O+S any graphical password schemes have been proposed
>?B>0? >*8?B>*-?
,
>?
. Dlonder introduced the first
graphical password schema. Dlonder's idea of graphical passwords is that by having a predetermined image, the user can select or touch regions of the image causing the sequence and the >?
location of the touches to construct the user's graphical password. (fter Dlonder , the notion of graphical passwords
was
developed.
any
graphical
password
schemes
have
been
proposed.
D&'( Vu eco#"iti o" )ased
Graphic al passwor ds
Pass*aces
+tory +che%e Passpoi"t
ecall )ased
Draw a +ecret ,DA+-
Figure 2 – Graphical Passor!s - Classification %hami&a and 4errig
>7?
proposed %E&F 9u, which is a +ecognition$based graphical password system
that authenticates Users by choosing portfolios among decoy portfolios. These portfolios are art randomi5ed portfolios. 1ach image is derived from an 0$D seed. Therefore, an authentication server does not need to store the whole image; it simply needs to store the 0$D seed. (nother recognition$ based graphical password is 4assfaces
>0?
. 4assfaces simply works by having the user
select a subgroup of k faces from a group of n faces. or authentication, the system shows m faces and one of the faces belongs to the subgroup k. The user has to do the selection many times to complete the authentication process. (nother scheme is the !tory scheme
>3?
, which requires
the selection of pictures of ob&ects Gpeople, cars, foods, airplanes, sightseeing, etc.H to form a story
line. %avis et al.
>3?
concluded that the user's choices in 4assfaces and in the !tory scheme result in a
password space that is far less than the theoretical entropy. Therefore, it leads to an insecure authentication scheme.
The graphical password schema of Dlonder
>?
is considered to be recall based since the user must
remember selection locations. oreover, 4ass4oint
>*8?B>*-?
is a recall$based graphical password
schema, where a background picture is presented and the user is free to select any point on the picture as the user's password Guser's 4ass4ointH. %raw a !ecret G%(!H, which is a recall$based graphical password schema and introduced by Iermyn et al.
>*#?
, is simply a grid in which the
user creates a drawing. The user's drawings, which consist of strokes, are considered to be the user's password. The si5e and the complexity of the grid affect the probable password space. :arger grid si5es increase the full password space. owever, there are limitations in grid complexity due to human error. "t becomes very hard to recall where the drawing started and ended and where the middle points were if we have very large grid si5es. /ne important type of authentication is based on who you are or, in other words, biometrics. Diometric recognition systems have been exhaustively studied as a way of authentication. ingerprints, palmprints, face recognition, voice recognition, and iris and retina recognition are all different methodologies of biometric recognition systems. •
uman properties are vulnerable to change from time to time due to several reasons such as aging, scarring, face makeup, change of hairstyle, and sickness Gchange of voiceH.
•
4eople tend to resist biometrics for different reasons. !ome people think that keeping a copy of the user's fingerprints is not acceptable and is a threat to the user's privacy. "n addition, some users resist the idea of a low$intensity infrared light or any other kind of light directed at their eyes, such as in retina recognition systems.
•
Diometrics cannot be revoked, which leads to a dilemma in case the user's data have been forged. Unlike other authentication schemes where the user can alter hisJher textual password in case of a stolen password or replace hisJher token if it has been stolen or forged, a user's biometrics cannot be revoked.
SC,E-E "n this section, we present a multifactor authentication scheme that combines the benefits of various authentication schemes. )e attempted to satisfy the following requirements. *. The new scheme should not be either recall based or +ecognition based only. "nstead, the scheme should be a combination of recall$, recognition$, biometrics$, and Token$based authentication schemes. -. Users ought to have the freedom to select whether the #$% password will be solely recall$, biometrics$, recognition$, or token$based, or a combination of two schemes or more. This freedom of selection is necessary because users are different and they have different requirements. !ome users do not like to carry cards. !ome users do not like to provide biometrical data, and some users have poor memories. Therefore, to ensure high user acceptability, the user's freedom of selection is important. #. The new scheme should provide secrets that are easy to remember and very difficult for intruders to guess. . The new scheme should provide secrets that are not easy to write down on paper. oreover, the scheme secrets should be difficult to share with others. <. The new scheme should provide secrets that can be easily revoked or changed. Dased on the aforementioned requirements, we propose our contribution, i.e., the #$% password authentication scheme. #.% )assword O.er.'ew The #$% password is a multifactor authentication scheme. The #$% password presents a #$% virtual environment containing various virtual ob&ects. The user navigates through this environment and interacts with the ob&ects. The #$% password is simply the combination and the sequence of user interactions that occur in the #$% virtual e nvironment. The #$% password can combine recognition$, recall$, token$, and biometrics$based systems into one authentication scheme. This can be done by designing a #$% virtual environment that contains ob&ects that request information to be recalled, information biometrical data to be verified. or
to
be
recogni5ed, tokens to
be
presented,
and
example, the user can enter the virtual environment and type something on a computer that exists in Gx*, y*, 5*H position, then enter a room that has a fingerprint recognition device that exists in a position Gx-, y-, 5-H and provide hisJher fingerprint. Then, the user can go to the virtual garage, open the car door, and turn on the radio to a specific channel. The combination and the sequence of the previous actions toward the specific ob&ects construct the user's #$% password. 9irtual ob&ects can be any ob&ect that we encounter in real life. (ny obvious actions and interactions toward the real$life ob&ects can be done in the virtual #$% environment toward the virtual ob&ects. oreover, any user input Gsuch as speaking in a specific locationH in the virtual #$% environment can be considered as a part of the #$% password. )e can have the following ob&ects= *. ( computer with which the user can type; -. ( fingerprint reader that requires the user's fingerprint; #. ( biometrical recognition device; . ( paper or a white board that a user can write, sign, or %raw on; <. (n automated teller machine G(TH that requests a token; . ( light that can be switched onJoff; 7. ( television or radio where channels can be selected; 0. ( staple that can be punched; 3. ( car that can be driven; *8. ( book that can be moved from one place to another; **. (ny graphical password scheme; *-. (ny real$life ob&ect; *#. (ny upcoming authentication scheme. The action toward an ob&ect Gassume a fingerprint recognition deviceH that exists in location Gx*, y*, 5*H is different from the actions toward a similar ob&ect Ganother fingerprint recognition deviceH that
exists in location Gx-, y-, 5-H, where x*KL x-, y*KL y-, and 5*KL 5-. Therefore, to perform the legitimate #$% password, the user must follow the same scenario performed by the legitimate user. This means interacting with the same ob&ects that reside at the exact locations and perform the exact actions in the proper sequence. #.% )assword Select'on and In$uts :et us consider a #$% virtual environment space of si5e 2 x 2 x 2. The #$% environment space is represented by the coordinates Gx, y, 5H M >*, . . . , 2? x >*, . . . , 2? x >*, . . . , 2?. The ob&ects are distributed in the #$% virtual environment with unique Gx, y, 5H coordinates. )e assume that the user can navigate into the #$% virtual environment and interact with the ob&ects using any input device such as a mouse, keyboard, fingerprint scanner, iris scanner, stylus, card reader, and microphone. )e consider the sequence of those actions and interactions using the previous input devices as the user's #$% password. or example, consider a user who navigates through the #$% virtual environment that consists of an office and a meeting room. :et us assume that the user is in the virtual office and the user turns around to the door located in G*8, -, 3*H and opens it. Then, the user closes the door. The user then finds a computer to the left, which exists in the position G, #, *0H,(nd the user types N(:C/O.P Then, the user walks to the meeting room and picks up a pen located at G*8, -, 08H and draws only one dot in a paper located in G*, *0, #8H, which is the dot Gx, yH coordinate relative to the paper space is G##8, *#8H. The user then presses the login button. The initial representation of user actions in the #$% virtual environment can be recorded as follows= G*8, -, 3*H (ction L /pen the office door; G*8, -, 3*H (ction L Close the office door; G, #, *0H (ction L Typing, NP; G, #, *0H (ction L Typing, N(P; G, #, *0H (ction L Typing, N:P; G, #, *0H (ction L Typing, NCP; G, #, *0H (ction L Typing, N/P;
G, #, *0H (ction L Typing, NOP; G*8, -, 08H (ction L 4ick up the pen; G*, *0, 08H (ction L %rawing, point L G##8, *#8H.
"a#
"b#
Figure $ – "a# %napshot of a proof-of-concept $-& 'irtual en'ironment( here the user is t)ping a te*tual passor! on a 'irtual computer as a part of the user+s $-& passor!, "b# %napshot of a proofof-concept 'irtual art galler)( hich contains $ pictures an! si* computers To simplify the idea of how a #$% password works, ig. shows a state diagram of a possible #$% password authentication system. (/! 0'rtual En.'ronment !es'gn %u'del'nes
%esigning a well$studied #$% virtual environment affects the usability, effectiveness, and acceptability of a #$% password system. Therefore, the first step in building a #$% password system is to design a #$% environment that reflects the administration needs and the security requirements. The design of #$% virtual environments should follow these guidelines.
Figure . – %tate !iagram of a possible $-& passor! application *. +eal$life similarity= The prospective #$% virtual environment should reflect what people are used to seeing in real life. /b&ects used in virtual environments should be relatively similar in si5e to real ob&ects Gsi5ed to scaleH. 4ossible actions and interactions toward virtual ob&ects should reflect real$life situations. /b&ect responses should be realistic. The target should have a #$% virtual environment that users can interact with, by using common sense.
-. /b& ect unique ness and distinctio n= 1very virtual ob&ect or item in the #$% virtual environment is different from any other virtual ob&ect. The uniqueness comes from the fact that every virtual ob&ect has its own attributes such as position. Thus, the prospective interaction with ob&ect * is not equal to the interaction with ob&ect -. owever, having similar ob&ects such as -8 computers in one place might confuse the user. Therefore, the design of the #$% virtual environment should consider that every ob&ect should be distinguishable from other ob&ects. ( simple real$life example is home numbering. (ssume that there are -8 or more homes that look like each other and the homes are not numbered. "t would be difficult to distinguish which house was visited a month ago. !imilarly, in designing a #$% virtual environment, it should be easy for users to navigate through and to distinguish between ob&ects. The distinguishing factor increases the user's recognition of ob&ects. Therefore, it improves the system usability. #. Th re e$dimensio nal virtual environme nt si5e = ( #$% virtual environment can depict a city or even the world. /n the other hand, it can depict a space as focused as a single room or office. The si5e of a #$% environment should be carefully studied. ( large #$% virtual environment will increase the time required by the user to perform a #$% password. oreover, a large #$% virtual environment can contain a large number of virtual ob&ects. Therefore, the probable #$% password space broadens. owever, a small #$% virtual environment usually contains only a few ob&ects, and thus, performing a #$% password will take less time. . Oumb er of ob&e cts GitemsH and their types= 4art of designing a #$% virtual environment is determining the types of ob&ects and how many ob&ects should be placed in the environment. The types of ob&ects reflect what kind of responses the ob&ect will have. or simplicity, we can consider requesting a textual password or a fingerprint as an ob&ect response type. !electing the right ob&ect response types and the number of ob&ects affects the probable password space of a #$% password. <. !ystem import ance= The #$% virtual environment should consider what systems will be protected by a #$% password. The number of ob&ects and the types of ob&ects that have been used in the #$% virtual environment should reflect the importance of the protected system.
#.% )assword A$$l'cat'ons Decause a #$% password can have a password space that is very large compared to other authentication schemes, the #$% password's main application domains are protecting critical systems and resources. 4ossible critical applications include the following. *. Critical servers= any large organi5ations have critical servers that are usually protected by a textual password. ( #$% password authentication proposes a sound replacement for a textual password. oreover, entrances to such locations are usually protected by access cards and sometimes 4"O numbers. Therefore, a #$% password can be used to protect the entrance to such locations and protect the usage of such servers. -. Ou cle ar and milita ry fa cilitie s= !uch facilities should be protected by the most powerful authentication systems. The #$% password has a very large probable password space, and since it can contain token$, biometrics$, recognition$, and knowledge$based authentications in a single authentication system, it is a sound choice for high level security locations. #. (irplanes and &etfighters= Decause of the possible threat of misusing airplanes and &etfighters for religion$political agendas, usage of such airplanes should be protected by a powerful authentication system. The #$% password is recommended for these systems. "n addition, #$% passwords can be used in less critical systems because the #$% virtual environment can be designed to fit any system's needs. ( small #$% virtual environment can be used in many systems, including the following= *. (Ts; -. 4ersonal digital assistants; #. %esktop computers and laptop logins; . )eb authentication.
SEC"ITY ANALYSIS To analyse and study how secure a system is, we have to consider how hard it is for the attacker to break such a system. ( possible measurement is based on the information content of a password space, which is defined in
>*#?
as Nthe entropy of the probability distribution over that space given
by the relative frequencies of the passwords that users actually choose.P )e have seen that textual password space may be relatively large; however, an attacker might only need a small subset of the full password space as 6lein
>-?
observed to successfully break such an authentication
system. (s a result, it is important to have a scheme that has a very large possible password space as one factor for increasing the work required by the attacker to break the authentication system. (nother factor is to find a scheme that has no previous or existing knowledge of the most probable user password selection, which can also resist the attack on such an authentication scheme.
Figure / - Passor! space of the $-& passor!( te*tual passor!( Passfaces( an! &A% ith gri! si0es of / / an! 1 1, 3ength is the number of actions an! interactions for a $-& passor!( the number of characters for te*tual passor!s( the number of selections for Passfaces( an! the number of points that represent the stro4es for &A%, The length is up to eight "characters5actions( interactions( inputs5selections#,,
(/! S'1e
)assword
S$ace
/ne important factor to determine how difficult it is to launch an attack on an authentication system is the si5e of the password space. To determine the #$% password space, we have to count all possible #$% passwords that have a certain number of actions, interactions, and inputs toward all ob&ects that exist in the #$% virtual environment. )e assume that the length of the #$% password is :max, and the probability of the #$% password of si5e greater than :max is 5ero. To measure the #$% password space, we will calculate Q G:max, 2H on a #$% virtual environment that has the space G2 x 2 x 2H for a #$% password of a length Gnumber of actions, interactions, and inputsH of :max or less. "n the following expression, (C represents the possible actions toward the #$% virtual environment, whereas Q represents the total numbers of possible #$% passwords of length :max or less=
"n the following expression G-H, 6max is the number of ob&ects in the #$% virtual environment=
)here *i L * 7 ( ) i L ) 7 ( and 0 i L 0 7 ( only if i L 7. The design of the #$% environment will determine the value of 6max. The variable m represents all possible actions and interactions toward all existing ob&ects 6i. owever, g G AC H counts the total number of actions and inputs toward the #$ % virtual environment, whereas m, as we mentioned before, counts the actions and interactions toward the ob&ects. (n example of g G AC H can be a user movement pattern, which can be considered as a part of the user's #$% password. The function is the number of possible actions and interactions toward the ob&ect /i based on the ob&ect type Ti. /b&ect types can be textual password ob&ects, %(! ob&ects, or any authentication scheme.
The function f is determined from the ob&ect type. "t counts the possible actions and interactions that the ob&ect can accept. "f we assume that an ob&ect N6eyboardP is in location Gx8, y8, 58H of type L textual password, f will count the possible characters and numbers that can be typed, which is around 3# possibilities. (s we mentioned before, an ob&ect type is one of the important factors that affects the overall password space. Therefore, higher outcomes of function f mean larger #$ %password space si5e.
Figure - Passor! space of the $-& passor!( te*tual passor!( Passfaces( an! &A% ith gri! si0es of / / an! 1 1, 3ength is the number of actions an! interactions for a $-& passor!( the number of characters for te*tual passor!s( the number of selections for Passfaces( an! the number of points that represent the stro4es for &A%, The length is up to e ight "characters5actions( interactions( inputs5selections#,
Figure 8 - obser'ing the number of possible actions5interactions of a $-& passor! ithin a $-& en'ironment specifie! in %ection 9-A compare! to the to critical points of te*tual passor!s, Point :a; is the bit si0e of
=2>
"$ 1 # !ictionar) of eight-character te*tual passor!s, Point
:b; represents the full passor! space of eight-character te*tual passor!s, ig. 7 shows the points where the #$% password exceeds two important textual password points. 4oint NaP shows that by having only two actions and interactions as a #$% password, the #$% password exceeds the number of textual passwords used by 6lein
>-?
to break -<@ of textual
passwords of eight characters. 4oint NbP represents the full textual password space of eight characters or less. "t shows that by performing only four interactions, actions, and inputs as a #$% password, the #$% password space exceeds the full textual passwords of eight characters or less. rom the previous equations, we observe that the number of ob&ects and the type of actions and interactions determines the probable password space. Therefore, the design of the #$% virtual environment is a very critical part of the #$% password system. igs. and < illustrate the resulting password space of the proposed #$% password compared to textual password, 4assfaces, and %(! of a grid of < x < and *8 x *8, respectively. Ootice the difference between a #$% passwords built on a simple #$% virtual environment compared to the other authentication schemes. (/! )assword +nowledge
!'str'but'on
!tudying the user's behavior of password selection and knowing the most probable textual passwords are the key behind dictionary attacks. 6lein
>-?
used such knowledge to collect a small
set of # x *8 words that have a high probability of usage among users. The question is how has such information
Ghighly probable passwordsH been found and why. Users tend to choose words that have meaning, such as places, names, famous people's names, sports terms, and biological terminologies. Therefore, finding these different words from the dictionary is a relatively simple task. Using such knowledge yields a high success rate for breaking textual passwords. (ny authentication scheme is affected by the knowledge distribution of the user's secrets. (ccording to %avis et al. >0?
>3?
, 4assfaces
users tend to choose faces that reflect their own taste on facial attractiveness, race, and gender.
oreover, *8@ of male passwords have been guessed in only two guesses. (nother study
>*?
about
user selection of >*#?
%(! eaning, which
concluded that for their secret passwords, users tend to draw things that have
simplifies
the attacker's
task. Currently, knowledge about user behaviors on selecting their #$% password does not exist. 1very user has different requirements and preferences when selecting the appropriate #$% password. This fact will increase the effort required to find a pattern of user's highly selected #$ % password. "n addition, since the #$% password combines several authentication schemes into a single authentication environment, the attacker has to study every single authentication scheme and has to discover what the most probable selected secrets are. or textual password, the highly probable selected textual password might be determined by the use of dictionaries. owever, there are many authentication schemes with undiscovered probable password space. !ince every #$% password system can be designed according to the protected system requirements, the attacker has to separately study every #$% password system. This is because ob&ects that exist in one #$% password system might not exist on other #$% password systems. Therefore, more effort is required to build the knowledge of most probable #$% passwords. Attacks Countermeasures
and
To reali5e and understand how far an authentication scheme is secure, we have to consider all possible attack methods. )e have to study whether the authentication scheme proposed is immune against such attacks or not. oreover, if the proposed authentication scheme is not immune, we then have to find the countermeasures that prevent such attacks. "n this section, we try to cover most possible attacks and whether the attack is valid or not. oreover, we try to propose countermeasures for such attacks.
*. ?rute Force Attac4@ The attacker has to try all possible #$% passwords. This kind of attack is very difficult for the following reasons. a. Time required to lo gin= The total time needed for a legitimate user to login may vary from -8 s to - min or more, depending on the number of interactions and actions, the si5e of the #$% virtual environment, and the type of actions and interactions done by the user as a #$% password. Therefore, a brute force attack on a #$% password is very difficult and time consuming. b. Cost of attacks= "n a #$% virtual environment that contains biometric recognition ob&ects and token$based ob&ects, the attacker has to forge all possible biometric information and forge all the required tokens. The cost of forging such information is very high; therefore, cracking the #$% password is more challenging. oreover, the high number of possible #$% password spaces Gas shown in Table "H leaves the attacker with almost no chance of breaking the #$% password.
2, ell-%tu!ie! Attac4@ The attacker tries to find the highest probable distribution of #$% passwords. owever, to launch such an attack, the attacker has to acquire knowledge of the most probable #$% password distributions. (cquiring such knowledge is very difficult because the attacker has to study all the existing authentication schemes that are used in the #$ % environment. oreover, acquiring such knowledge may require forging all existing biometrical data and may require forging token$based data. "n addition, it requires a study of the user's selection of ob&ects, or a combination of ob&ects, that the user will use as a #$% password. oreover, a well$studied attack is very hard to accomplish since the attacker has to perform a customi5ed attack for every different #$% virtual environment design. 1very system can be protected by a #$% password that is based on a unique #$% virtual environment. This environment has a number of ob&ects and types of ob&ect responses that differ from any other #$% virtual environment. Therefore, a carefully customi5ed study is required to initiali5e an effective attack. #. !houlder !u rfing (tta ck= (n attacker uses a camera to record the user's #$% password or tries to watch the legitimate user while the #$% password is being performed. This attack is the most successful type of attack against #$% passwords and some other graphical passwords.
owever, the user's #$% password may contain biometrical data or textual passwords that cannot be seen from behind. The attacker may be required to take additional measures to break the legitimate user's #$% password. The refore, we assume that the #$% password should be performed in a secure place where a shoulder surfing attack cannot be performed. . Timing (ttack= "n this attack, the attacker observes how long it takes the legitimate user to perform a correct sign$in using the #$% password. This observation gives the attacker an indication of the legitimate user's #$% password length. owever, this kind of attack alone cannot be very successful since it gives the attacker mere hints. Therefore, it would probably be launched as part of a well$studied or brute force attack. Timing attacks can be very effective if the #$% virtual environment is poorly designed.
E2)EI-ENTAL ES"LTS )e have built an experimental #$% virtual environment that contains several ob&ects of two types. The first type of response is the textual password. The second type of response is requesting graphical passwords. (lmost #8 users volunteered to experiment with the environment. )e asked the users to create their #$% password and to sign$in using their #$% password several times over several days. E#$er'mental 0'rtual (/! En.'ronment
"n our experiment, we have used Iava /pen 2: to build the #$% virtual environment and we have used a *.08$25 4entium Centrino machine with <*-$D random access memory and (T" obility +adeon 388 video card. The design of the experimental #$% virtual environment represents an art gallery that the user can walk through and is depicted in ig. #.
Table 1 - Resulting number of possible $-! passor!s of total length 3ma*
"ser Stud3
)e conducted a user study on #$% passwords using the experimental #$% virtual environments. The study reviewed the usage of textual passwords and other authentication schemes. The study covered almost #8 users. The users varied in age, sex, and educ ation level. 1ven though it is a small set of users, the study produced some distinct results
>*#?
. )e observed the following
regarding textual passwords, #$% passwords, and other authentication schemes. *. ost users who use textual passwords of 3B*- character lengths or who use random characters as a password have only one to three unique passwords. -. ore than <8@ of user's textual passwords are eight characters or less. #. (lmost -<@ of users use meaningful words as their textual passwords. . (lmost 7<@ of users use meaningful words or partially meaningful words as their textual passwords. "n contrast, only -<@ of users use random characters and letters as textual passwords. <. /ver 8@ of users have only one to three unique textual passwords, and over 38@ of users have eight unique textual passwords or less. . /ver 38@ of users do not change their textual passwords unless they are required to by the system. 7. /ver 3<@ of users under study have never used any graphical password scheme as a means of authentication. 0. ost users feel that #$% passwords have a high acceptability. 3. ost users believe that there is no threat to personal privacy by using a #$% password as an authentication scheme.
LITEAT"E E0IE*
•
&tt$455www6'eee#$lore6'eee6org4 (n "111 paper published in -880 was the basic
information source. •
&tt$455www63outube6com4 eticulous details about the #% passwords and #% virtual
environment were the result of various videos available on RouTube. •
&tt$455www6(d.as6 com4 9(! is a network of #% virtual 2alleries for displaying art on
the internet. (ny artist interested in mounting an exhibition can do so in a simple, user friendly way. (ll you have to do is open an account Gfor freeH, choose an appropriate #% gallery and upload your artworks. /nce youSve done that, anyone can visit your exhibition. 1very visitor is represented by a #% character, allowing everyone to see and be seen in the #d gallery space in real$time.
CONCL"SION AN! F"T"E *O+ There are many authentication schemes in the current state. !ome of them are based on user's physical and behavioral properties, and some other authentication schemes are based on user's knowledge such as textual and graphical passwords. oreover, there are some other important authentication schemes that are based on what you have, such as smart cards. (mong the various authentication schemes, textual password and token$based schemes, or the combination of both, are commonly applied. owever, as mentioned before, both authentication schemes are vulnerable to certain attacks. oreover, there are many authentication schemes that are currently under study and they may require additional time and effort to be applicable for commercial use. The #$% password is a multifactor authentication scheme that combines these various authentication schemes into a single #$% virtual environment. The virtual environment can contain any existing authentication scheme or even any upcoming authentication schemes by adding it as a response to actions performed on an ob&ect. Therefore, the resulted password space becomes very large compared to any existing authentication schemes. The design of the #$% virtual environment, the selections of ob&ects inside the environment, and the ob&ect's type reflect the resulted password space. "t is the task of the system administrator to design the environment and to select the appropriate ob&ect that reflects the protected system requirements. (dditionally, designing a simple and easy to use #$% virtual environment is a factor that leads to a higher user acceptability of a #$% password system. The choice of what authentication schemes will be part of the user's #$% password reflects the user's preferences and requirements. ( user who prefers to remember and recall a password might choose textual and graphical passwords as part of their #$% password. /n the other hand, users who have more difficulty with memory or recall might prefer to choose smart cards or biometrics as part of their #$% password. oreover, users who prefer to keep any kind of biometrical data private might not interact with ob&ects that require biometric information. Therefore, it is the user's choice and decision to construct the desired and preferred #$% password. The #$% password is still in its early stages. %esigning various kinds of #$% virtual environments, deciding
on password spaces, and interpreting user feedback and experiences from such
environments will result in enhancing and improving the user experience of the #$% password.
oreover, gathering attackers from different backgrounds to break the system is one of the future works that will lead to system improvement and prove the complexity of breaking a #$% password. oreover, it will demonstrate how the attackers will acquire the knowledge of the most probable #$% passwords to launch their attacks. !houlder surfing attacks are still possible and effective against #$% passwords. Therefore, a proper solution is a field of research.
EFEENCES
>*?
A. !uo, R. hu, and 2. !. /wen, N2raphical passwords= ( survey,P in Proc, 21st Annu, Comput, %ecurit) Appl, Conf,, %ec.
>-? %. 9. 6lein, Noiling the cracker= ( survey of, and improvement to passwords security,P in Proc, B%DE %ecurit) or4shop, *338, pp. #? ODC news, AT Frau!@ ?an4ing on our one)( &ateline Hi!!en Cameras %ho Criminals 6ning ATs, %ec. **, -88#. >?
T. 6itten,
an
)e
on
the
AT . G-88<,
Iul.
**H.
>/nline?
(vailable=
(Tarket4lace.com >
DDC news, Cash achine Frau! up( %a) ?an4s, Oov. , -88.
>?
2. 1. Dlonder, N2raphical password,P U.!. 4atent < <<3 3*, !ep. -, *33.
>7?
+. %hami&a and (. 4errig, N%E&F 9u= ( user study using images for authentication,P in Proc, Ith B%ED %ecurit) %)mp,, %enver, C/, (ug. -888, pp.
>0? +eal User Corporation, The %cience ?ehin! Passfaces. G-88<, /ct.H. >/nline?. (vailable= http=JJwww.realusers.com >3?
%. %avis, . onrose, and . 6. +eiter, N/n user choice in graphical password schemes,P in Proc, 1$th B%DE %ecurit) %)mp,, !an %iego, C(, (ug. -88, pp. *B*.
>*8? !. )iedenbeck, I. )aters, I.$C. Dirget, (. Drodskiy, and O. emon, N(uthentication using graphical passwords= 1ffects of tolerance and image choice,P in Proc, %)mp, Bsable Pri'ac) %ecurit), 4ittsburgh, 4(, Iul. -88<, pp. *B*-. >**? !. )iedenbeck, I. )aters, I.$C. Dirget, (. Drodskiy, and O. emon, N(uthentication using graphical passwords= Dasic results,P in Proc, Human-Comput, Enteraction Ent,, :as 9egas, O9, Iul. -*-? !. )iedenbeck, I. )aters, I.$C. Dirget, (. Drodskiy, and O. emon, N4ass4oints= %esign and longitudinal evaluation of a graphical password system,P Ent, J, Human-Comput, %tu!, "%pecial Essue on HCE Research in Pri'ac) an! %ecurit)#, vol. #, no. *J-, pp. *8-B *-7, Iul.
