Keyboard handwriting authentication: benefits and problems of use. Keyboard handwriting is a natural change associated with the improvement of the user's skills in working with the keyboard and mouse or, conversely, with their deterioration due to the aging of the body

Moscow State Technical University named after N. E. Bauman

Department of Automated Information Processing and Control Systems

Essay

in the discipline "Information Security"

"Keyboarding as a means of authentication"

Performed:

student of group IU5-114

Pavlyuchenko K. G.

Checked:

Kesel S. A.

Moscow 2014

Introduction 3

More about Method 4 itself

Testing method 8

Conclusion 11

Advantages of the method: 11

Disadvantages: 11

References 13

Introduction

In this abstract we will consider one of the authentication methods - procedures for verifying the authenticity of the access subject. Authenticator is some parameter provided to the system for verification.

There are 3 types of authenticators:

Unique knowledge (password, pin code)

Unique item (key, smart card)

A unique characteristic of the subject himself (static - fingerprints, retinal image, behavioral - for example, voice authentication)

The latter type also includes keyboard handwriting.

This is a behavioral biometric characteristic that is described by the following parameters:

Input speed - the number of characters entered divided by the typing time

Input dynamics - characterized by the time between keystrokes and the time they are held

Frequency of input errors

Key usage - for example, which function keys are pressed to enter capital letters

More details about the method itself

All people perceive current events differently. Try to estimate in a short time the number of dots or vowels in long words, the sizes of horizontal and vertical lines - how many subjects there will be, so many opinions. These features of the human psyche are also suitable for identification. True, depending on the state and well-being of a person, the obtained values ​​will “float”, so in practice they rely on an integral approach, when the result is summed up according to several checks, taking into account work with the keyboard. For example, the identification method may be as follows: vertical or horizontal lines appear on the screen for a few seconds. Their size and number are random. The user types the appropriate numbers, in his opinion. Thus, it is determined: the characteristics of the keyboard handwriting, it is assessed how close the indicated length and number of lines are to reality, attention and counting accuracy (how the length of one line is correctly compared with the neighboring one). Finally, the results are compared with the standard. In this method, errors in determining sizes are not so important; the main thing is that they are repeated both during setup and identification.

From the point of view of using hidden monitoring of computer security systems, it is of interest to classify the psychophysical parameters of the user, which include: keyboard handwriting, mouse signature, reaction to events occurring on the screen. We will only focus on considering the use of keyboard handwriting for personal identification.

One of the rather difficult tasks that many people solve every day is quickly typing texts from a computer keyboard. Typically, fast keyboard input can be achieved by using all fingers of both hands. At the same time, each person has his own unique keyboard style. Keyboard style is a set of dynamic characteristics of working on the keyboard.

Not many people realize that when communicating with a computer, the user’s individuality is manifested in the speed of typing, the habit of using the main or additional part of the keyboard, the nature of “double” and “triple” keystrokes, favorite computer control techniques, etc. And this is not the case nothing surprising - this is akin to the ability of music lovers to distinguish by ear pianists performing the same piece, or the work of telegraph operators using Morse code.

This identification method is popular in the USA to prevent children from accessing the Internet through home computers. Even if a child spied or found out the parents’ password, he will not be able to use it. This method can also be used for additional protection when organizing access to computer systems.

Keyboard handwriting recognition consists of selecting the appropriate standard from a list of standards stored in the computer's memory, based on an assessment of the degree of similarity of the handwriting parameters of one of the operators authorized to work with this computer to this standard. Solving the problem of user identification comes down to solving the problem of pattern recognition.

The classical statistical approach to recognizing a user by keyboard handwriting (a set of keywords) revealed a number of interesting features: the dependence of handwriting on letter combinations in a word, the existence of deep connections between a set of individual characters, and the presence of “delays” when entering characters.

A very important characteristic of biometric identification is the length of the passphrase. Practice shows that the passphrase should be easy to remember and contain from 21 to 42 keystrokes. When synthesizing a passphrase, it is permissible to use words with meaning.

In addition, it is possible to analyze such features as the dependence of the speed of entering words on their meaning, the relative time of pressing various keys, etc. Moreover, in some cases they are even more informative: for example, the test taker’s reaction to various terms will indicate his area of ​​interest. Indeed, a chemist will type “hydrogen” or “compound” faster than “program” or “excavator”. And the fashion designer will be more familiar with words such as “mannequin” or “pattern”.

The collection of biometric information about the user’s work when analyzing keyboard handwriting occurs by measuring the intervals between keystrokes and the time they are held, after which the results obtained are formed into a matrix of intercharacter intervals and a vector of key hold times. After collecting biometric information, the obtained data is compared with its reference values.

How can one identify the individual characteristics of keyboard handwriting? Yes, just like with graphological examination: you need a reference and studied text samples. It is better if their content is the same - the so-called password or key phrase. Of course, it is impossible to distinguish a user by two or three, even ten pressed keys. We need statistics.

When typing a key phrase, the computer allows you to record many different parameters, but for identification it is most convenient to use the time spent entering individual characters. The times of pressing the keys tl, t2,....tn are different and, accordingly, the values ​​of these parameters can be used to identify the characteristic features of the user's keyboard handwriting. In addition, you can use the intervals between pressing adjacent keys as controlled parameters. The controlled parameters significantly depend on how many fingers the user uses when typing, on the user’s characteristic combinations of movements of various fingers, and on the characteristic movements of the hands when typing. For example, if you force the user to work with one finger of one hand, then the keyboard handwriting almost completely loses its individuality. In this case, keypress times no longer reflect the individuality of people, since the intervals between keypresses become proportional to the distance between the keys, and overlapping keypresses of adjacent keys becomes impossible.

The unique features of keyboard handwriting are identified by two methods:

by typing a key phrase;

for typing “free” text.

Each of them necessarily has configuration and identification modes. During setup, benchmark characteristics of the user's input of key phrases, such as the time spent on individual characters, are determined and remembered. And in the identification mode, the reference and resulting sets are compared after eliminating gross errors.

“Free” text is typed using a wide variety of phrases (the key phrase is usually the same), which has its advantages, allowing you to obtain individual characteristics unnoticed, without focusing the user’s attention on the password phrase.

The choice of verification scheme depends on the application with which it is used. For example, if an accountant wanted to get a short certificate, but the computer instead offers to type 2-3 pages of “free” text to make sure that it really is the right person. There will be no nerves here and it will only cause irritation, and as a result, the user will try in every possible way to avoid such an identification system.

On the other hand, someone with security clearance can work with such a program all day long, leaving the computer from time to time. And to prevent attackers from taking advantage of the exposed system at this moment, it is advisable to periodically conduct a “secret check.” Such systems allow you to constantly monitor whether a legitimate user is sitting at the computer.

It should be noted that when using these methods, it becomes possible not only to confirm the authenticity, but also to analyze its condition. The described approach to protection against unauthorized access allows you to:

monitor the physical condition of employees;

end the practice of violating security rules when working with passwords;

provide an easier and equally secure method of logging into the network.

Send your good work in the knowledge base is simple. Use the form below

Students, graduate students, young scientists who use the knowledge base in their studies and work will be very grateful to you.

INTRODUCTION

Traditional identification and authentication methods based on the use of wearable identifiers, as well as passwords and access codes, have a number of significant disadvantages due to the fact that attributive and knowledge-based identification characteristics are used to establish the authenticity of the user. This disadvantage is eliminated by using biometric identification methods. Biometric characteristics are an integral part of a person and therefore cannot be forgotten or lost. An important place among biometric products is occupied by devices and programs based on the analysis of dynamic images of a person (authentication by the dynamics of a handwritten signature, by keyboard handwriting, by working with a computer mouse, etc.).

1.1 General principles for constructing biometric systems dynamic identification/authentication

Dynamic systems of biometric identification/personal authentication are based on the use of certain dynamic parameters and characteristics of a person (gait, handwriting and keyboard handwriting, speech) as signs.

Biometric systems based on the analysis of individual characteristics of movement dynamics have much in common. This allows you to use one generalized scheme to describe all biometric systems of this class, which is shown in Fig. 1 and reflects the main stages of information processing.

Posted on http://www.allbest.ru/

Rice. 1. Generalized block diagram of the personal identification system based on the characteristics of movement dynamics

The first stage of processing is the conversion of non-electrical quantities (coordinates of the tip of the pen, sound pressure, hand position) into electrical signals. Next, these signals are digitized and entered into a processor that performs software processing of the data. During software processing, the amplitudes of the input signals are scaled, bringing them to a certain reference value. In addition, the signals are brought to a single time scale, the signals are split into separate fragments, and then the signal fragments are shifted to optimal alignment with the reference location.

After bringing the scales and shifting of signal fragments to the reference value, the vector of functionals is calculated (vector of controlled biometric parameters v=(v 1, v 2, ..., v k)).

The first five information processing blocks listed above operate according to the same algorithms, regardless of the operating mode of the biometric system itself. It is for this reason that they form a sequential connection of blocks without branches. The operating mode of the system (training or authentication) determines the set of operations carried out with the already generated vector of parameters v=(v 1, v 2, ..., v k).

If the biometric system is in training mode, the vectors of biometric parameters v enter the block of training rules, which forms a biometric standard of the individual. Since dynamic personality images have significant variability, several examples of implementations of the same image are required to form a biometric standard. In the simplest case, a biometric standard can be formed in the form of two vectors: a vector of mathematical expectations of controlled parameters m(v) and a vector of dispersions of these parameters (v).

In authentication mode, the vector of controlled biometric parameters v obtained from the presented image is compared by a decision rule with a biometric standard. If the presented vector is close to the biometric standard, a positive authentication decision is made. If the presented vector differs significantly from its biometric standard, admission is denied. If the authentication protocol is not too strong, the user is given additional re-authentication attempts.

The type of decision rule used by the system and the type of biometric standard are inextricably linked. When developing a system, based on the selected decision rule, the type of biometric standard is determined.

The application of the principles of biometric personal identification in information security systems has led to the creation of biometric identification/authentication systems (BIS) for access to information objects (in particular, personal computers). Users of such facilities must undergo a biometric identification/authentication procedure to gain access to them.

The quality of the BSI work is characterized by the percentage of errors during the admission procedure. The BSI distinguishes between three types of errors:

· FRR (False Reject Rate) or error of the first type - the probability of erroneous refusals to an authorized user (erroneous refusal to one’s own);

· FAR (False Accept Rate) or error of the second type - this is the probability of admitting an unregistered user (erroneous admission of a “stranger”);

· EER (Equal Error Rates) - equal probability (rate) of errors of the first and second types.

Depending on the requirements for BSI, the formation of the user’s biometric standard is also carried out with a given degree of rigor. The samples presented by a given user must correspond to some average statistical characteristic for a given user. That is, after collecting some initial statistics, the presentation of bad samples (samples with large deviations from the statistical average) should be rejected by the system. The ratio of samples accepted by the system to the total number of samples presented characterizes the degree of stability of the biometric parameters of a given user.

To experimentally test the FRR characteristics, the system is sequentially presented n times with the biometric characteristics of users who have successfully completed registration. Next, the ratio of the number n 1 of unsuccessful attempts (system denial of access) to the total number of attempts n is calculated. The specified ratio provides an estimate of the error probability FRR. The estimate is considered reliable at values ​​of n 1/FRR.

To experimentally test the FAR characteristics, the system is sequentially presented m times with the biometric characteristics of users who have not been registered. Next, the ratio of the number n 2 of successful attempts (positive authentication decision) to the total number of attempts m is calculated. The specified ratio provides an estimate of the error probability FAR. The estimate is considered reliable at values ​​m 1/FAR.

Biometric identification systems are essentially an addition to standard password identification (when a user logs into the system). However, in the future, a decrease in the percentage of password identification to the total number of identification systems and an increase in the share of biometric authentication systems is predicted.

Today, the most famous are two biometric authentication systems: the “ID - 007” system, developed by an American company

“Enigma Logic” and the “Cobra” system, developed by the Russian FSB Academy.

Systems of this type implement three main functions: 1) collection of information; 2) information processing (mechanisms for comparisons with reference values); 3) making decisions based on authentication results.

The first and third functions in the “ID - 007” and “Cobra” systems are implemented algorithmically in the same way (the difference is made by some coefficients), but the second function - information processing or mechanisms for comparisons with reference values ​​- are fundamentally different. Comparison of the newly obtained values ​​of key hold times with reference values ​​in both systems is carried out using an additive characteristic. Comparison of intercharacter intervals with reference values ​​in the “ID - 007” system is performed according to the principle of additive relation, and in the “Cobra” system - according to the principle of multiplicative relation.

1.2 Authentication based on keystroke analysis and painting with the mouse

identification authentication access protection

The main biometric characteristics of CS users that can be used for their authentication include:

fingerprints;

geometric hand shape;

iris pattern;

retinal pattern;

geometric shape and size of the face;

voice timbre;

geometric shape and size of the ear, etc.

One of the first to propose the idea of ​​user authentication based on the characteristics of their work with the keyboard and mouse was S.P. Rastorguev. When developing a mathematical model of authentication based on users' keyboard handwriting, it was assumed that the time intervals between pressing adjacent symbols of a key phrase and between pressing specific key combinations in it obey a normal distribution law. The essence of this authentication method is to test the hypothesis about the equality of the distribution centers of two normal general populations (obtained when setting up the system for the user’s characteristics and during his authentication).

Let's consider the option of user authentication by typing a key phrase (the same in the configuration and authentication modes).

The procedure for setting up the characteristics of a user registered in the CS:

user selection of a key phrase (its symbols should be evenly spaced across the keyboard);

typing a key phrase several times;

calculation and storage of estimates of mathematical expectations, variances and number of observations for time intervals between sets of each pair of adjacent key phrase symbols.

The user authentication procedure can be carried out in two ways. First option of the authentication procedure:

typing a key phrase by the user several times;

exclusion of gross errors (using a special algorithm);

calculation of estimates of mathematical expectations and variances for time intervals between pressing each pair of adjacent symbols of a key phrase;

solving the problem of testing the hypothesis about the equality of variances of two normal populations for each pair of adjacent symbols of a key phrase (using a special algorithm);

if the variances are equal, then solving the problem of testing the hypothesis about the equality of the distribution centers of two normal populations with an unknown variance for each pair of adjacent symbols of the key phrase (using a special algorithm);

calculating the probability of user authenticity as the ratio of the number of combinations of adjacent keys for which hypotheses are confirmed (items 4 and 5) to the total number of combinations of adjacent key phrase symbols;

Second option for the authentication procedure:

enter the key phrase once;

solving the problem of testing the hypothesis about the equality of variances of two normal populations for time intervals between pressing adjacent symbols of a key phrase;

if the variances are equal, then eliminating the time intervals between pressing adjacent symbols of the key phrase, which differ significantly from the reference ones (obtained during setup);

calculating the probability of user authenticity as the ratio of the number of remaining intervals to the total number of intervals in the key phrase

comparison of the obtained probability estimate with the selected threshold value for making a decision on user admission.

Instead of using a key phrase that is constant for the KS user, authentication can be carried out by typing pseudo-random text. In this case, the keyboard is divided into fields and the concept of distance d ij between keys i and j is introduced, which means the number of keys located on the straight line connecting i and j. The i key belongs to the t field if

d ij k

We will call the quantity k the degree of the field m (if k = 0, then m is a separate key). Let us denote by x ij the time interval between keystrokes belonging to fields i and j.

Let us introduce the following assumptions:

The keystroke characteristics of one field are closer to each other, the smaller the k;

for a user working with two hands, obtaining the characteristics of the keyboard handwriting is possible by studying work with only one half of the keyboard;

the key phrase can be any set of characters;

the number of fields must be the same in configuration and authentication modes.

Setting procedure for typing pseudo-random text:

generating and displaying text to the user from a fixed set of words, the symbols of which are scattered as widely as possible across the keyboard;

user typing;

fixing and storing x ij values, which are then used to calculate the statistical characteristics of keyboard handwriting.

The authentication procedure is the same as the authentication procedure used when entering a passphrase.

The reliability of authentication based on the user's keyboard handwriting is lower than when using his biometric characteristics.

However, this authentication method also has its advantages:

the ability to hide the fact of using additional user authentication if a password phrase entered by the user is used as a key phrase;

the possibility of implementing this method only using software (reducing the cost of authentication tools).

Now let’s look at an authentication method based on painting with a mouse (using this manipulator, of course, it is impossible to perform a real user signature, so this signature will be a fairly simple stroke). Let's call a painting line a broken line obtained by connecting points from the beginning of the painting to its completion (neighboring points should not have the same coordinates). We calculate the length of the painting line as the sum of the lengths of the segments connecting the points of the painting.

Let us introduce the concept of a break in a painting line, the sign of which will be the fulfillment of the condition

where is the distance between two adjacent points of the painting line; d -- length of the entire line; k -- number of points in the line.

To eliminate breaks in the painting line, Rastorguev proposed an algorithm for smoothing it, which consists of adding additional points to the line at the points of its breaks. Each additional point with coordinates x a and y a added between points i-1 and i of the painting line must satisfy the condition

min(

Using a smoothed painting line, you can identify all the closed contours in it (using a special algorithm).

The customization procedure for user characteristics may consist of the following steps:

input of several reference paintings;

for each painting, obtaining the number of points in it and the length of its line, determining the number and location of breaks in the line of the painting;

for each line of painting, performing smoothing, obtaining the number and location of closed contours;

calculation of the average value of the obtained characteristics of the painting and their permissible deviations.

The authentication procedure consists of the following steps:

input of painting;

calculation of the number of points and length of the painting line;

obtaining the number and location of breaks in the painting line;

smoothing the painting line;

obtaining the number and location of closed loops;

comparison of the obtained characteristics of the painting with the reference ones;

making a decision on the user’s admission to work in the CS.

Similar to authentication based on keyboard handwriting, the authenticity of the user by his signature with a mouse is confirmed primarily by the pace of his work with this input device.

The advantages of authenticating users by their mouse handwriting, similar to using keyboard handwriting, include the possibility of implementing this method only using software; The disadvantages are the lower reliability of authentication compared to the use of biometric user characteristics, as well as the need for the user to have fairly confident mouse skills.

A common feature of authentication methods based on keyboard handwriting and mouse writing is the instability of their characteristics for the same user, which can be caused by:

natural changes associated with the improvement of the user’s skills in working with the keyboard and mouse or, conversely, with their deterioration due to the aging of the body;

changes associated with an abnormal physical or emotional state of the user.

Changes in user characteristics caused by reasons of the first type are not abrupt, and therefore can be neutralized by changing the reference characteristics after each successful user authentication.

Changes in user characteristics caused by reasons of the second type can be abrupt and lead to the rejection of his attempt to enter the CS. However, this feature of authentication based on keyboard handwriting and mouse writing can also become an advantage when it comes to users of computer systems for military, energy and financial purposes.

A promising direction for the development of methods for authenticating CS users based on their personal characteristics may be confirmation of the user’s authenticity based on his knowledge and skills characterizing the level of education and culture.

1 .3 Designed software

In this case, the designed software must provide a reliable mechanism for implementing the developed user authentication system using keyboard handwriting, carry out coordinated interaction with the operating system, have a convenient interface with users and minimal requirements for system resources.

The hardware will be a specialized computer in which the main algorithmic sections of the software implementation of the system, which require the greatest time resources, will be implemented at the hardware level. Such areas include, first of all, an algorithm for collecting biometric characteristics, selecting reference matrices of users from RAM, an algorithm for comparing reference characteristics with the obtained characteristics, and a mechanism for generating control signals based on the authentication result.

The developed hardware implementation of the authentication algorithm consists of: a clock pulse generator, a clock pulse divider, a real-time clock, a keyboard interrupt controller, a pulse adder, statistics accumulation RAM, standards RAM, a buffer for storing intermediate data, a matrix processor, a results accumulation buffer, and a generation module control pulses, interface for interaction with the control computer.

An important factor influencing the quality of authentication is the use of non-standard means of entering information, for example, the method of entering symbolic information into a computer using a six-key keyboard shown in Fig. 2. (where a) is a left-hand keyboard, b) a right-hand keyboard). These tools make it possible to fix the fingers on the information keys and eliminate their inter-key movements, which helps to focus attention on the more “subtle” biometric parameters of a person during the dynamic process of entering information into the computer.

The dynamic capabilities of the human hand on a standard keyboard are quite wide - this is due to the fact that to press a key you need to move your fingers at interkey distances of up to 40 mm, and the proposed method of entering character information in a six-element code based on right and left circular scanning limits these capabilities, placing more stringent requirements on the user's experience with the keyboard. The proposed method for increasing the efficiency of authentication is implemented in the developed device for entering symbolic information.

Figure 2. - Device for entering symbolic information into a computer in a six-element code based on right and left circular scanning

The keyboard of the device, shown in Fig. 2, consists of keys 1-6, which are installed on panel 7 and diverge radially around its circumference, and the size of the keys is not the same depending on the size of the corresponding fingers and the congruent palmar side of the hand at wrist level.

It should be noted that when using the proposed device and method, it becomes possible not only to confirm the authenticity, but also to analyze the user’s state.

The described approach to protection against unauthorized access allows you to:

· monitor the physical condition of employees;

· stop practicing violating security rules when working with passwords;

· provide an easier and equally secure method of logging into the network.

From the point of view of using hidden monitoring of computer security systems, it is of interest to classify the psychophysical parameters of the user, which include: keyboard handwriting, mouse signature, reaction to events occurring on the screen.

Identification of keyboard handwriting consists of selecting the appropriate standard from the list of standards stored in the computer’s memory, based on assessing the degree of similarity of the handwriting parameters of one of the operators authorized to work with this computer to this standard. The task of identifying a user comes down to solving the problem of pattern recognition. The classical statistical approach to recognizing a user by keyboard handwriting when typing keywords revealed a number of interesting features: a significant dependence of handwriting on letter combinations in a word; the existence of deep connections between a set of individual symbols; presence of “delays” when entering characters.

The results obtained allow us to draw a conclusion about the feasibility and effectiveness of using this method for identifying and authenticating a user using keyboard handwriting.

If the computer is already equipped with tools for analyzing biometric characteristics, the cost of biometric authentication systems will be entirely determined by the cost of the software, which, in turn, depends on the circulation and should decrease significantly in the future. One of the prerequisites for this is the possibility of developing such software by small and medium-sized companies that can create competition in this market sector.

CONCLUSION

Currently, the most relevant are systems based on biometric methods of differentiation and access control. One of the important areas of biometrics is the authentication of users by their keyboard handwriting. Its area of ​​application is systems in which there is keyboard input of information or control via a keyboard: computer systems and networks, cellular communications, systems of national importance, etc.

However, many issues of user authentication based on their keyboard handwriting have not been studied. Existing software implementations of such systems are characterized by insufficient authentication reliability. The development of new methods, algorithms and their hardware and software implementations that increase the efficiency of identification and authentication systems is relevant.

Increasing the reliability of user authentication can be achieved through the development of a new authentication algorithm in telecommunications systems and networks - a poly-Gaussian algorithm, which makes it possible to explore new parameters of keyboard handwriting while simultaneously increasing the registered user information. However, the use of this algorithm is hampered by the insufficient development of methods for its implementation.

Bibliography

1. Ivanov A.I. Biometric identification of a person by the dynamics of subconscious movements: Monograph. - Penza: Penz Publishing House. state Univ., 2000. 188 p.

2. Bryukhomitsky Yu.A., Kazarin M.N. handwriting authentication system / collection of proceedings of a scientific and practical conference with international participation “information security”. Taganrog: TRTU publishing house, 2002.

3. Khorev P.B. Methods and means of information security in computer systems: Textbook. aid for students higher textbook institutions - M.: Publishing center "Academy", 2005. - 256 p.

Posted on Allbest.ru

...

Similar documents

Using electronic keys as a means of user authentication. Analysis of identification and authentication methods from the point of view of the technologies used in them. Installation and configuration of "Rutoken" authentication tools, driver management.

course work, added 01/11/2013


General principles of authentication in Windows. Local and domain registration. Authentication in Linux. Access rights to files and registry. Transactions, primitives, chains and policies. Basic components of a security descriptor. Password storage and encryption.

course work, added 06/13/2013


Problems of using passwords in the enterprise. General concepts and technologies of identification and authentication. The operating principle and structure of the SecureLogin software from ActiveIdentity. Automatic password generation, phishing and pharming.

course work, added 01/22/2015


Data security concept. Basic technologies for network authentication of information based on reusable and one-time passwords: access authorization, audit. Certification authorities, public key infrastructure, digital signature, program codes.

course work, added 12/23/2014


Translation of form fields. A method of authentication on the Web that requires visitors to provide a username and password. Form for data transfer. Using a database to store passwords. Development of a script for visitor authentication.

lecture, added 04/27/2009


Development of proposals for the implementation of biometric authentication of users of a linear computer network. The essence and characteristics of static and dynamic methods of user authentication. Threat elimination methods, protection service parameters.

course work, added 04/25/2014


Classification and main characteristics of biometric means of personal identification. Features of the implementation of static and dynamic methods of biometric control. Means of authorization and authentication in electronic security and safety systems.

course work, added 01/19/2011


Concept of biometric authentication process. Technology and probability of authentication errors by fingerprint, iris or retina, by the geometry of a person’s hand and face, by facial thermogram, by voice, by handwriting.

presentation, added 05/03/2014


Development of plug-in authentication modules as a means of user authentication. Linux-PAM module included in Linux distributions. Operating principle, administration, time and resource limitations. Overview of authentication plugins.

course work, added 01/29/2011


Characteristics of biometric access control systems (BACS) and a generalized scheme of their functioning. Static and dynamic authentication methods. Integration of BSKD with video surveillance systems. Application of BSKD to protect data transmission systems.

All people perceive different events differently. This feature of the psyche is suitable for identification. Usually, in practice, several methods are used so that the person’s condition cannot affect the correctness of identification. For example, horizontal and vertical lines with a random number and size appear on the screen during identification. The user must score certain results for his opinion. The results are checked against the template. The results may not be correct, the main thing is that they coincide.

Today, many complex tasks come down to quickly typing text on the keyboard. Using two hands, each person creates an individual approach to typing. Keyboard handwriting is a set of dynamic parameters for working on the keyboard. The user's individuality depends on the speed of typing characters, different habits regarding keystrokes, etc. This is how music lovers distinguish audio works by ear. This method can be used for example or serve as one of. The classical statistical approach to keyboard handwriting showed a number of features:

• dependence of handwriting on combinations of letters in a word
• deep connections between a set of different characters
• delays— when entering characters

Another important parameter for such identification is the presence of a passphrase. Such a phrase should have from 21 to 42 keystrokes and be easy to remember. The collection of such information about the user when analyzing handwriting is realized by measuring the interval between the user pressing and holding keys, this is shown in Fig. 1.

Picture 1

As can be seen from Fig. 1, when typing a key phrase, the computer can record the time spent on different characters. Also, the parameters depend on how the user dials, whether with one finger, five or all. The unique characteristics of keyboard handwriting are revealed using:

• free text typing
• by typing a key phrase

It should be noted that the implementation of such methods not only helps to identify users, but also allows them to analyze their status. This approach allows:

• monitor the physical well-being of users
• provide a simple and reliable method of identification
• simplifies the problem when working with identification passwords

Practice

Keyboard handwriting- a behavioral biometric characteristic that describes the following aspects:

• input dynamics is the time between keystrokes and the time they are held
• Input speed is the result of the number of characters divided by the typing time.
• Key implementation - for example, which keys are pressed to type capital letters
• input error rate

A simple program can be implemented to test keystroke authentication capabilities. You need to come up with a passphrase that subjects will enter repeatedly. The input speed will fluctuate within certain limits for a certain period of time. Based on such statistics, you can get some mat. expectation of speed and its dispersion (Fig. 2).

Figure - 2

Now the results can be compared with speed tests of other subjects. Taking into account the different speeds at which the subjects typed the passphrase, they all created a certain speed range within which all results fell. The range is large (Fig. 3), so there will be some probability of false positives.

Advantages of implementing keyboard handwriting for authentication:

• Does not require any additional actions from the user. The user still enters his password when logging in; additional authentication can be implemented based on this password
• Ease of implementation and implementation.
• Possibility of hidden authentication - the user may not be aware that he is undergoing additional verification of keyboard handwriting

• Strong dependence on a specific keyboard. If the keyboard is replaced, the user needs to configure the program again
• Application training required
• Also depends on the user's condition. For example, if the user is sick, he may have different keyboard handwriting parameters

Ayupova A.R. 1, Yakupov A.R. 2, Shabalkina A.A. 3

1 ORCID: 0000-0002-6820-1605, candidate of physical and mathematical sciences, 2 student of the Faculty of Economics and Mathematics, 3 student of the Faculty of Economics and Mathematics,

Neftekamsk branch of Bashkir State University

AUTHENTICATION BY KEYBOARD HANDWRITING: BENEFITS AND PROBLEMS OF USE

annotation

Authentication is one of the oldest and at the same time pressing problems in the field of information security. The rules for creating passwords that have already become dogmas cannot withstand the criticism of expertsIT-industry. The article discusses the possibility of eliminating some of the shortcomings of the password method using a keyboard handwriting recognition system. Keystroke refers to dynamic (behavioural) biometric characteristics that describe the subconscious actions that are habitual to the user. The advantages and disadvantages of this system are also discussed in detail.

Keywords: authentication, identification, information protection, keyboard handwriting, password.

Ayupova A.R. 1, Yakupov A.R. 2, Shabalkina A.A. 3

1 ORCID: 0000-0002-6820-1605, PhD in Physics and Mathematics, 2 Student of the Faculty of Economics and Mathematics, 3 Student of the Faculty of Economics and Mathematics,

Neftekamsk branch of Bashkir State University,

KEYBOARD RHYTHM AUTHENTICATION: BENEFITS AND PROBLEMS OF USE

Abstract

Authentication is one of the oldest and at the same time topical problems in the field of information security. The rules of passwords forming that have already become dogmatic doesn’t stand up to scrutiny of the specialists of the IT industry. The paper deals with the possibility of eliminating some of the shortcomings of the password method with the help of a keyboard rhythm recognition system. Keyboard rhythm refers to dynamic (behavioral) biometric characteristics describing subconscious actions, typical for a user. Also the advantages and disadvantages of this system are considered in detail.

Keywords: authentication, identification, information protection, keyboard rhythm, password.

During the general informatization, the tasks of information protection acquire particular importance and significance. Modern information security is impossible without the process of verifying the authenticity of entered data - authentication. The password authentication method has become widespread, but it has a number of disadvantages:

The solution to some problems is a combination of a password and a person’s keyboard handwriting. Keyboarding is a unique style of entering characters. The input style is determined by: the speed of holding the key, the time between keystrokes, the features of entering double or triple keystrokes, etc. , .

Using a keyboard style provides a noticeable advantage for the user - a simplified password. It will be enough for a person to remember a convenient combination, for example, the password length can consist of 4 to 8 characters. Taking into account the specifics of the input, the threat of password theft by an attacker becomes less of a priority, since the input speed can vary significantly.

However, do not forget that a person will never enter a password in exactly the same way. Biometric authentication methods always allow for some inaccuracy. This flaw can lead to the following errors:

– FRR (false rejection rate) – error of the first type – the probability of not allowing a registered user into the system;

– FAR (false access rate) – error of the second type – the probability of allowing an attacker into the system under the guise of a registered user.

Another disadvantage is the need to train the program so that it remembers the features of the user's keyboard handwriting and reduces the likelihood of FRR and FAR. Obviously, the more the program learns, the less likely it is for errors to occur, and the higher the security of the system.

But the most serious problem with using keyboard handwriting is the strong dependence of the result on the psychophysical state of the user. If a person is not feeling well, they may not be able to authenticate as their input speed may be significantly impaired. One cannot ignore the dependence on the user's experience and the technical characteristics of the keyboard.

It is not recommended to use too long passwords here, as this worsens the results due to the fact that a person thinks about the input, remembering the combination - first of all, the value of the time parameter between keystrokes changes. This is explained by the fact that hand movements when entering a memorized short password are controlled by the subconscious thinking process, therefore, the time for entering such a password will be approximately the same in different time periods.

It is also possible that an attacker, using a special program - a keylogger, can steal not only the password, but also the user’s keyboard signature.

To test the possibility of actually using a combination of keyboard handwriting methods with passwords, the “Keyboard Handwriting” program was developed (Fig. 1). This program is able to measure basic indicators such as input speed, time between keystrokes, key hold time, as well as their average values ​​to compare the “learned” password with the newly entered one. Additionally, to test the keystroke method, it is assumed that the attacker knows the password, but is trying to bypass the security and penetrate the system.

Rice. 1 – Program interface

This program contains the basic elements of keyboard handwriting:

• sample password;
• password test;
• area for displaying information;
• average values ​​display password entry rates averaged over multiple attempts;
• speed of password entry from start to finish;
• Hold Time displays a table that shows the time in milliseconds (ms) for each key from the time the button is pressed to the time the button is released.

Time between presses displays a table that shows the time in milliseconds (ms) that elapsed from the moment the previous button was released until the next button was pressed.

The program uses methods for calculating dispersion to determine the spread of key press time values, finding the median to determine the “reference” indicators and the average value of the main indicators.

To begin, the user must enter a sample password. Next, you need to enter the password several times in the “Test Password” field. This way the program will remember the handwriting of the password owner.

Next, you need to switch the mode in Settings > Change mode. Now, if you enter a password in the “Password Test” field, the program will compare the existing handwriting of its owner with the handwriting entered by the user in this mode. On the graph you can see red or green dots at the top of each parameter, compared to each other. A red dot indicates that this parameter is different from the handwriting already in memory, and green dots indicate matches.

It is worth noting that the algorithm allows for some error, which is calculated after several trial inputs by the owner. The more times the user enters his password, the better the program will remember his handwriting and the fewer errors there will be.

In Figure 2 you can see an example that the owner of the password was authenticated with only a discrepancy in the time of holding the “w”, “o” and “r” keys (Figure 2).

Rice. 2 – Password owner has been verified

Figure 3 shows the result of an authentication attempt by another person (Figure 3). He was unable to replicate the owner’s handwriting; the results of the indicators are very different. In particular, there are significant differences in the average speed and keystroke times.

Rice. 3 – Another person could not copy the owner’s handwriting

During statistical processing of the program testing results, the following shortcomings were identified: type I and type II errors. In some cases, the pace of entering the user's password coincided with the pace of the attacker, and shortcomings were also identified in the event of a change in state or change of the test subject's keyboard.

Existing software implementations of keyboard handwriting recognition methods are characterized by insufficient reliability of identification and authentication and a high probability of first and second errors.

There are other analogues of software implementation of keyboard handwriting. For example, Igor Aguryanov’s proprietary program, which is capable of comparing the keystroke performance of one user depending on the person’s state and keyboard change. But this program is not intended for authentication, since the comparison of indicators in the above-mentioned program occurs exclusively for one user. Thus, “Keyboard Handwriting” has more advanced functionality.

In the future, it is planned to improve the “Keyboard Handwriting” program: improving the recognition algorithm, increasing the accuracy of the timer.

Keyboard handwriting, according to security experts, is a huge area for further research. And despite the threat of complete de-anonymization, one cannot help but admit that this technology is incredibly effective as a protection tool.

Bibliography /References

1. Sarbukov A.E. Authentication in computer systems / A.E. Sarbukov, A.A. Grushko // Security systems. – 2003. – No. 5(53). – pp. 118–122.
2. Zadorozhny V. Review of biometric technologies / V. Zadorozhny // Information protection. Confidential. – 2003. – No. 5. – P. 26–29.
3. Dovgal V.A. Capturing keyboard handwriting parameters and its features / A.V. Dovgal // Materials of the All-Russian scientific and practical conference “Information systems and technologies in modeling and management”. – Publishing House Typograph. "Arial", 2017.–pp.230-236.
4. Kaluzhin A.S. Confirmation of the user's identity by his keyboard stroke / A.S. Kaluzhin, D.D. Ruder // News of Altai State University. – 2015. – T. 1. – No. 85. – pp. 158-162.
5. Bryukhomitsky Yu.A. Histogram method for recognizing keyboard handwriting / Yu.A. Bryukhomitsky // News of the Southern Federal University. Technical science. – 2010. – T. 11. – No. 112. – P. 55-62.
6. Maznichenko N.I. Analysis of the capabilities of systems for automatic identification of keyboard handwriting / N.I. Maznichenko, M.V. Gvozdenko // Bulletin of the National Technical University Kharkov Polytechnic Institute. Series: Computer science and modeling. – 2008. – No. 24. – P. 77-81.
7. Sidorkina I.G. Three algorithms for controlling access to CSII based on recognizing the operator’s keyboard handwriting / I.G. Sidorkina, A.N. Savinov // Bulletin of the Chuvash University. – 2013. – T.3. - No. 3. – P. 239-301.
8. Eremenko A.V. Two-factor authentication of computer system users on a remote server using keyboard handwriting / A.V. Eremenko, A.E. Sulavko // Applied informatics. – 2015. – T. 6. – No. 60. – pp. 48-59.
9. Dovgal V.A. Review of performance characteristics of data sets used to ensure information security based on keyboard handwriting / V.A. Dovgal // Bulletin of the Adygea State University. Series 4: Natural, mathematical and technical sciences. – 2016. – T. 4. – No. 191. – pp. 157-163.
10. Martynova L.E. Research and comparative analysis of authentication methods / L.E. Martynova, M.Yu. Umnitsyn, K.E. Nazarova and others // Young scientist. – 2016. –No. 19 (123). – pp. 90-93.

List of references in English /References in English

1. Sarbukov A. E. Autentifikaciya v kompyuternih sistemah / A. E. Sarbukov, A. A. Grusho// Sistemi bezopasnosti. – 2003. – No. 5(53). – R.118–122.
2. Zadorojnii V. Obzor biometricheskih tehnologii / V.Zadorojnii // Zaschita informacii. Confident. – 2003. – No. 5. – R. 26–29.
3. Dovgal V.A. Zahvat parametrov klaviaturnogo pocherka i ego osobennosti /A.V. Dovgal // Materiali vserossiiskoi nauchno_prakticheskoi konferencii “Informacionnie sistemi i tehnologii v modelirovanii i upravlenii” [Materials of all-Russian scientific-practical conference “Information systems and technologies in modeling and control”]. –Izd-vo Tipograf. "Arial", 2017.–P. 230–236.
4. Kaluzhin A.S. Podtverzhdenie lichnosti pol’zovatelya po ego klaviaturnomu podcherku /A.S. Kaluzhin, D.D. Ruder // Izvestiya Altajskogo gosudarstvennogo universiteta. – 2015. – T. 1. – No. 85. – R.158-162.
5. Bryuhomickij YU.A. Gistogrammnyj metod raspoznavaniya klaviaturnogo podcherka / YU.A. Bryuhomickij // Izvestiya Yuzhnogo federal’nogo universiteta. Tekhnicheskie nauki. – 2010. – T. 11. – No. 112. – R. 55-62.
6. Maznichenko N.I. Analiz vozmozhnostej sistem avtomaticheskoj identifikacii klaviaturnogo podcherka / N.I. Maznichenko, M.V. Gvozdenko // Vestnik Nacional’nogo tekhnicheskogo universiteta Har’kovskij politekhnicheskij institut. Seriya: Informatika i modeling. – 2008. – No. 24. – R. 77–81.
7. Sidorkina I.G. Tri algoritma upravleniya dostupom k KSII na osnove raspoznavaniya klaviaturnogo podcherka operatora / I.G. Sidorkina, A.N. Savinov // Vestnik Chuvashskogo universiteta. – 2013. – T.3. - No. 3. – R. 239–301.
8. Eremenko A.V. Dvuhfaktornaya autentifikaciya pol’zovatelej komp’yuternyh sistem na udalennom servere po klaviaturnomu pocherku / A.V.Eremenko, A.E. Sulavko // Prikladnaya informatika. – 2015. – T. 6. – No. 60. – R. 48–59.
9. Dovgal' V.A. Obzor harakteristik proizvoditel’nosti naborov dannyh, ispol’zuemyh dlya obespecheniya informacionnoj bezopasnosti na osnove klaviaturnogo pocherka / V.A. Dovgal’ // Vestnik Adygejskogo gosudarstvennogo universiteta. Seriya 4: Estestvenno-matematicheskie i tekhnicheskie nauki. – 2016. – T. 4. – No. 191. – R. 157-163.
10. Martynova L.E.. Issledovanie i sravnitel’nyj analiz metodov autentifikacii / L.E. Martynova, M.Y.U. Umnicyn, K.E. Nazarova i dr. //Molodoj uchenyj. – 2016. No. 19 (123). – R. 90-93.

We provide consumer lending online. The issue of preventing fraud is one of the main ones for us. Often, potential scammers think that the security systems of fintech companies are lower than those of traditional financial institutions. But this is a myth. It's exactly the opposite.

In order to reduce the level of potential fraud to zero, we have developed a behavioral biometrics system that can identify a person by their keyboard handwriting. The pilot version was introduced in Russia in August 2017.

How it works

The system allows user authentication based on behavioral patterns characteristic of each person. Among them: speed and dynamics of typing, transition time between keys, typos and features of the mouse cursor moving across the screen. As soon as a person registers in his personal account, the system analyzes his behavior and makes a snapshot, which is used for subsequent authentication on the site.

The new program eliminates cases of fraud. If, when entering your personal account, the user’s handwriting does not match the impression in the database, such a profile is added to the list of risk profiles that require additional verification. Those. if the system has reason to believe that a person is not who he claims to be, then it can request additional data or send this request for verification, where a living person will decide to approve or deny the loan.

The peculiarity of the system is that the analyzed behavioral patterns cannot be faked: even if a person changes the keyboard, they may change slightly, but in the aggregate their structure will remain unchanged.

How it was implemented

The system was developed in-house by the holding company based on big data analysis technology. The development and implementation of the program took about 4 weeks, the pilot lasted 6 months. The accuracy of the pilot version of the biometric system was 97.6%. The system is constantly being improved and filled with new data, which will increase accuracy to the maximum extent. The biometric system forms the basis of ID Finance fraud scoring, which allows you to identify fraudsters. Biometrics are used not only for authentication, but also for initial registration, when the client’s handwriting is compared with all the impressions in the database.

What did you get?

Thanks to the anti-fraud scoring system, it was possible to reduce the cost of one issued loan by almost 25% by reducing verification costs and the cost of requests to information sources. The approval rate of applications increased by 28%. The economic effect of implementation is estimated at more than 90 million rubles. in Russia at the end of 2018. The economic effect from keyboard handwriting recognition technology alone in the ID Finance group in seven countries is estimated at $2.8 million at the end of 2018.