No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
USERS' IDENTIFICATION THROUGH KEYSTROKE DYNAMICS BASED ON VIBRATION PARAMETERS AND KEYBOARD PRESSURE
... One principle drawn from past studies is the idea of extracting attributes from the keypresses to have qualities to compare. Examples of how characteristics of the keypress can be used in this way are in [22] (which also measured vibration), and [23], with best results 1.67% FAR at 0% FRR for [23] and 0.6% error rate for [22], which was an extensive study using neural methods. This paper differs from most papers using analogue keystroke dynamics because the system is on a keypad, not a full keyboard. ...
... In addition, the keys should return as fast as possible after the press, ideally moving up with the finger that pressed the key, so the data recorded on the upstroke shows how the button was released, rather than how the button is returned to the resting position. The number pad should also contains analogue sensing elements; previous studies have used piezoelectric sensors [24], or force sensitive resistors [22,25] to measure the pressure of a button press. In this case study, we chose linear motion potentiometers mostly chosen for the ease of creating buttons. ...
... Various algorithms can be used to build a distinctive profile that capture the keystroke characteristics of each user. Possibly the most effective algorithms involve machine learning, which although harder to implement than statistical methods can provide better results [13], [22]. Such methods may require a lot of processing power, possibly too much to be included. ...
This paper proposes a portable hardware token for user authentication, it is based on the use of keystroke dynamics to verify users in a bio-metric manner. The proposed approach allows for a multifactor authentication scheme in which users are not allowed access unless they provide the correct password and their unique bio-metric signature. The proposed system is implemented in hardware and its security is evaluated.
... T Due to the low number of features with high quality, such approaches give a high percentage of errors FRR and FAR. 3. The most highly accurate methods of pattern recognition ("deep" learning networks, convolutional neural networks, evolutionary neural networks, etc.) require huge amounts of training sample (thousands and tens of thousands of examples) and therefore it is difficult to apply them in biometric authentication. The recognition machine is guaranteed to be trained on a small number of examples of the user's pattern (10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20). 4. Iterative algorithms for artificial neural networks (ANNs) learning lose their stability when the ANN structure becomes more complicated. ...
... As candidates for biometric features, the detailing coefficients obtained for the decomposition of functions related to biometric patterns with different bases using the fast Mallat algorithm were considered (Fig. 3). As bases, the following were used: Haar, Daubechies (D4, D4, D6, D8, D10, D12, D14, D16, D18, D20), Simlet (orders 4 to 10), Coiflet (6,12,18,24,30), Morlet, Mayer, Shannon. Briefly consider the features of these bases suitable for fast wavelet transform. ...
... In training the user recognition system based on the training sample data, it is possible to construct a probability density function of the feature (approximated by some distribution law, if it's known, or the empirical density through relative-frequency histograms). The average informativeness of a group of features is determined through the areas of intersection of probability densities functions [12] by formula (1): ...
... If the feature space is "plane" then the shortest path from the image to the center of the template can be found using the Euclidean measure. If the ranges of values of particular features vary significantly in scale, the space ceases to be "ideally plane", but it can be normalized by calculating distances with the help of the Pearson measure (3) [7]. Chi-module measure provides similar results while recognizing images (4). ...
... In this case an aspect of variability of a personal biometric pattern in a course of time is taken into account. This aspect was not considered in previous experiments [1,7,10,13]. This test scenario meets real conditions. ...
... For comparison, we give graphs of signatures verification errors by the Bayes-Hamming networks ( Figure 6). They show that the use of a network of one type of functional gives more errors and a slightly different picture of the characteristic curves, which indicates that the measures (6), (7) and (8) are not completely correlated. If a biometric system is tested immediately after its learning (or using testing samplings) the hybrid wide networks divide users' patterns almost error-free. ...
This paper suggests a model of a hybrid wide neural network based on perceptrons, quadratic form networks and multidimensional difference and hyperbolic Bayes functionals. It is experimentally proved that this model is highly efficient when used for biometric authentication and generation of a digital signature activated biometrically. The paper suggests methods of generating keys of a digital signature and personal authentication by handwritten patterns, a key stroke manner and facial parameters. Comparatively high rates of reliability for taken solutions were achieved that were estimated taking into account the variability of dynamic biometric patterns over time.
... The market of biometric protection means is increasing very rapidly. Studies in this area are also progressive, there are all new ways for person recognition: gait and gestures [2], parameters of electroencephalograms [3], finger's pressing force on each key during password phrase typing [4]. Searching for new approaches is came due to the drawbacks of traditional authentication technologies. ...
... The point is valid within the "fuzzy extractors" concept [1,5,6] or artificial neural networks [8], consisting of "classical" neurons based on the weighted sum function (1). This is also true for many proximity measures (functionals), in particular for quadratic forms and their analogues (for example, the Pearson measure, Euclid measure, chi-module) [4,15]. Networks of quadratic forms also lose power if the vector of input biometric parameters contains dependent values. ...
A model of neurons for biometric authentication, capable of efficient processing of highly dependent features, based on the agreement criteria (Gini, Сramer-von-Mises, Kolmogorov-Smirnov, the maximum of intersection areas of probability densities) is proposed. An experiment was performed on comparing the efficiency of neurons based on the proposed model and neurons on the basis of difference and hyperbolic Bayesian functionals capable of processing highly dependent biometric data. Variants of construction of hybrid neural networks, that can be trained on a small number of examples of a biometric pattern (about 20), are suggested. An experiment was conducted to collect dynamic biometric patterns, in the experiment 90 people entered handwritten and voice patterns during a month. Intermediate results on recognition of subjects based on hybrid neural networks were obtained. Number of errors in verification of a signature (handwritten password) was less than 2%, verification of a speaker by a fixed passphrase was less than 6%. The testing was carried out on biometric samples, obtained after some time period after the formation of training sample.
... The market of biometric protection means is increasing very rapidly. Studies in this area are also progressive, there are all new ways for person recognition: gait and gestures [2], parameters of electroencephalograms [3], finger's pressing force on each key during password phrase typing [4]. Searching for new approaches is came due to the drawbacks of traditional authentication technologies. ...
... The point is valid within the "fuzzy extractors" concept [1,5,6] or artificial neural networks [8], consisting of "classical" neurons based on the weighted sum function (1). This is also true for many proximity measures (functionals), in particular for quadratic forms and their analogues (for example, the Pearson measure, Euclid measure, chi-module) [4,15]. Networks of quadratic forms also lose power if the vector of input biometric parameters contains dependent values. ...
... Dynamic biometric patterns include: keyboard handwriting, handwritten passwords and signatures, gait characteristics, voice, electroencephalogram, cardiogram, computer mouse operation, parameters of head tremor, facial expressions, face, neck and other thermograms, finger pressure screen, keys. All these patterns in various degrees characterize both the person and his condition (psychophysiological and emotional) [4,5,6,7]. ...
... Within the framework of the theory of "broad" ANN, procedures for assessing the informativeness of characteristics (through areas of intersection of functions of probability densities) have been applied [7]. For the first time it has been proposed to create synapses taking into account the informativeness of the characteristics, and to determine the number of neuron inputs, starting from the general informative character of the signs [31]. ...
... The method can be used to identify and differentiate the individual characteristics based on biometrical features [13]. As further enhancement, researchers developed a keyboard that can record the user characteristics based on pressure parameter with a combination of time for individuals, while the press any key on the keyboard [14]. ...
Cloud computing enables the data owners to store and host their data on the cloud storage servers, which is to accessed by data consumers from the cloud storage servers whenever needed. The approach provides not only better accessibility but also unlimited storage capacity to the users. Even though there are numerous benefits of the approach, but security remains the prime concern for cloud computing and cloud service users. There are few solutions that provide protection in cloud computing, but even after having these solution, trust in cloud service providers remains the prime concern for cloud users. To overcome such a situation and to enhance the faith amongst the user and cloud service provider, a hybrid monitoring scheme (SecHMS) have been proposed and evaluated in this paper. The hybrid monitoring scheme (SecHMS) uses public-key cryptography and hashing technique to provide data security in cloud computing. The hybrid scheme (SecHMS) constantly monitors the stored data on behalf of end-users. As user’s data is getting continuously monitored, it leads to the enhancement of trust for the end-user in cloud computing systems. A thorough analysis has been done on different size files, and results have been demonstrated to show the correctness of SecHMS scheme.
... Also, the effect of varying the number of enrollment samples is not investigated. Some methods employ special keyboard which can measure the pressure and apply this information in their authentication procedure [31,32]. In mobile devices, the timing features could be combined 125 with movement sensor information to form sensor-enhanced keystroke dynamics [33,34]. ...
With the increasing need for information security, one of the key options ahead is to provide security based on biometrics. Authentication based on keystroke dynamics is a low cost and convenient biometric authentication technique. In this paper, a method for user authentication based on keystroke dynamics with a novel similarity measure is introduced. Using time–frequency analysis, a similarity measure between an input sample and user reference samples is directly obtained. The input sample is initially converted to a keystroke dynamics signal. Dynamic time warping method is applied to equalize the length of signals. Then, using Wigner distribution, the time–frequency representation of the samples is obtained. Finally, exploiting the correlation coefficient, the similarity between two signals in the time–frequency domain is measured. We also added an update procedure to the proposed method to enhance its performance. The performance of the proposed method is investigated and compared with the state-of-the-art methods. Experimental results show the superiority of the proposed method.
... The results of the study involved a total of 51 volunteers. This touch screen-based study was continued by Sulavko, Eremenko [44] by attempting to identify 20 individuals from 100 volunteers participated. The EER obtained was 0.6%. ...
Feature extraction is an important process before an analysis of a data is carry out. Different behaviour of a user while using the keyboard is a feature that need to be identified in the Keystroke Dynamics (KD) study. Example are the difference between typing time between letters, typing speed and the force of a person pressing the keyboard. Past studies related to feature extraction for KD have been described in this paper. Various features that have been used are listed and the results of the study are compared. The results of this writing are expected to help new researchers in the process of evaluating KD.
В данной статье выдвигается гипотеза о том, что набор текста на клавиатуре каждым конкретным человеком имеет индивидуальные особенности. Развитие этой технологии в дальнейшем поможет предотвратить попытки несанкционированного доступа к персональным данным, банковским счетам и коммерческой тайне. Среди существующих методов биометрической аутентификации предложенный подход относится к категории динамических методов, подвергающихся изменениям со временем. Эта особенность не позволяет злоумышленнику через доступ к сети украсть шаблон почерка пользователя, его скопировать или подделать. Во время работы на клавиатуре человеком используется более 20 различных мышц, что делает стиль набора уникальным. В качестве основных характеристик для проведения аутентификации личности были взяты скорость набора, время удержания клавиши в нажатом положении, время поиска следующей клавиши, периодические опечатки в процессе набора текста и многое другое. Компьютерный почерк может быть зафиксирован в виде различных метрик и проанализирован статистическими методами. Автором статьи раскрывается методика и условия проведения эксперимента. Система ведет подсчет количества нажатий за единицу времени, устанавливает временные метки, производит сбор статистических данных для построения гистограмм. Эксперимент проводится в разное время суток с использованием различных типов клавиатур. Закладывается гипотеза о том, что данные выборки подчиняются нормальному распределению, что подтверждается анализом полученных результатов по критерию согласия Пирсона. Для определения отличий стилей набора испытуемого от своего собственного находится процент совпадений показателей по интегральной формуле Муавра — Лапласа для нормальных распределений, значение которого около 90 %. Аналогичным образом проводится сравнительный анализ полученных результатов с разными пользователями. В данном случае этот показатель будет гораздо ниже и не превышает 60 %. Сравнительный анализ дает возможность аутентифицировать личность и является достаточным информационным показателем для предотвращения попыток несанкционированного доступа.
The paper presents several variations of fuzzy extractors to generate cryptographic keys and password based on parameters of keystroke dynamics. A series of simulation experiments was run to estimate the efficiency of these methods, the best parameters of fuzzy extractors were found. The best result was: FRR=0.061, FAR=0.023 with a key length 192 bits.
The modern encryption methods are reliable if strong keys (passwords) are used, but the human factor issue cannot be solved by cryptographic methods. The best variant is binding all authenticators (passwords, encryption keys, and others) to the identities. When a user is authenticated by biometrical characteristics, the problem of protecting a biometrical template stored on a remote server becomes a concern. The paper proposes several methods of generating keys (passwords) by means of the fuzzy extractors method based on signature parameters without storing templates in an open way.
This article discusses the problem of user identification and psychophysiological state assessment while writing a signature using a graphics tablet. The solution of the problem includes the creation of templates containing handwriting signature features simultaneously with the hidden registration of physiological parameters of a person being tested. Heart rate variability description in the different time points is used as a physiological parameter. As a result, a signature template is automatically generated for psychophysiological states of an identified person. The problem of user identification and psychophysiological state assessment is solved depending on the registered value of a physiological parameter.
Computing and communication systems have improved our way of life, but have also contributed to an increased data exposure and, consequently, to identity theft. A possible way to overcome this issue is by the use of biometric technologies for user authentication. Among the possible technologies to be analysed, this work focuses on keystroke dynamics, which attempts to recognize users by their typing rhythm. In order to guide future researches in this area, a systematic review on keystroke dynamics was conducted and presented here. The systematic review method adopts a rigorous procedure with the definition of a formal review protocol. Systematic reviews are not commonly used in artificial intelligence, and this work contributes to its use in the area. This paper discusses the process involved in the review along with the results obtained in order to identify the state of the art of keystroke dynamics. We summarized main classifiers, performance measures, extracted features and benchmark datasets used in the area.
The research aims at the development of a method of continuous identification of computer network users during their professional activity by means of data received from standard peripherals (keystroke dynamics, facial images registered by a web camera, mouse dynamics). A new method of identification is proposed. A natural experiment evaluating the reliability of the proposed method has been carried out. The results (FRR = 0.015, FAR = 0.001) suggest that it is possible to approach on the subject recognition reliability to the methods of identification based on open biometric images (e.g., fingerprint) using standard equipment (keyboard, mouse, webcam).
Fuzzy extractors and neural networks used for identification by handwriting dynamics should be used in tandem with another methods. Bayesian statistical inference networks or other networks able to consider a complete matrix of correlation coefficients may become a method that complete them. The paper proposes a simple correlation metric that takes into account correlation coefficients of biometric data in an explicit form. A mean vector, a standard deviation vector and a correlation matrix of biometric parameters should be considered as a complete statistical model of a handwritten signature. These data are enough for tuning a variety of variants of statistical inference networks using Bayes rules.
This paper shows that correlation coefficients obtained from small test samples for biometric data involve considerable uncertainty. This interferes with using them for machine training (setting) of classical quadratic forms and Bayesian networks. A method for symmetrizing correlation relationships is proposed. The requirement on the volume of biometric data is proved to be reduced considerably in this case. As a consequence, the setting (teaching) of quadratic forms and setting of maximum likelihood Bayesian networks become much more stable problems. This enables many-fold reduction in the requirement on the size of the training sample for an “own” biometric image.
Dependence on computers to store and process sensitive information has made it necessary to secure them from intruders. A behavioral biometric such as keystroke dynamics which makes use of the typing cadence of an individual can be used to strengthen existing security techniques effectively and cheaply. Due to the ballistic (semi-autonomous) nature of the typing behavior it is difficult to impersonate, making it useful as abiometric. Therefore in this paper, we provide a basic background of the psychological basis behind the use of keystroke dynamics. We also discuss the data acquisition methods, approaches and the performance of the methods used by researchers on standard computer keyboards. In this survey, we find that the use and acceptance of this biometric could be increased by development of standardized databases, assignment of nomenclature for features, development of common data interchange formats, establishment of protocols for evaluating methods, and resolution of privacy issues.
A systematic review on keystroke dynamics // Journal of the Brazilian Computer Society
- Lorena Ana Pisani Paulo Henrique
- Carolina
Pisani Paulo Henrique, Lorena Ana Carolina. A systematic review on keystroke dynamics // Journal of the Brazilian Computer Society. 2013. 19 (4). Р.