We continue to research new resources of behavior analysis: fitness trackers, ECG and EEG.
One of the main tendencies of modern education is its distant format. But the practical use of this new technological and content format might have some problems. One of them is how to organize effective proctoring thus guaranteeing the quality of education. The first necessity of any examination is to identify a student. As a teacher is not able to memorize all students face to face, so there is a necessity in a student identification and legal support of this process.
To maintain this problem there is an automatic identification during tests. Students are identified not just by login-password system, but by biometric factors. This biometric identification is considered effective as it minimizes cheating, and also monitors students behavior during examination.
Let's analyze applied resources of biometrics as a method of identification of a person based on his physiological or behavioral characteristics. 
There are many biometric characteristics as well as many ways of their identification. Each way has pros and cons.
Modern ways of biometric identifications are divided into two groups: static and dynamic. Static technologies are based on personal unique physiological features: fingerprint, palm shape, location of the veins on the front of the hand, retina, iris, shape and temperature records of face, DNA. At the heart of the dynamic methods of biometric identification arebehavioral or physiological changing characteristics. They include: special features of handwriting and keyboard handwriting, voice, characteristics of EEG, ECG and eye movements.
With all diversity only two of them are used in proctoring systems — identification by keyboard typing and by face image (2D- or 3D-image).
But some other ways of biometric identification can be used in proctoring as well.
Identification by any parameter should meet the following conditions:
-Its university – any person should possess this detected parameter (excluding unique single situations).
-Its unique character – it should vary from one person to another.
-Its constancy — the parameter should not change with time, for example, when getting older.
The process of biometric identification is based on a comparison of the identified object data and the biometric standard. Such a comparison is not possible without a pre-recording and saving of this biometric information. The main instruments of automatic biometric method are scanner for a biometric characteristic measuring and algorithm to compare it with the same pre-registered characteristic (so-called biometric standard).
The majority of biometric identification methods are faced with the same types of problems:
- Selection of a mathematical algorithm for information processing with maximum accuracy;
- To create devices as convenient as efficient;
- To overcome interference or distortions arising in the process of registration indicators.
Since these problems are common for all biometric identification methods and differ only in the degree of solutions success, let's take a closer look at the specific features of biometric technologies.
The main research object of fingerprinting is plane reflections of human pectinate skin – papillate fingerprints patterns. Their structure is usually studied in handprints of hand surfaces obtained using the black paint on the white paper. Currently, special scanners are developed to identify a person without the use of paint and they are much faster.
In comparison with other common biometric systems, fingerprinting has several advantages:
- Convenience for user;
- Ease at use and development of algorithms;
- The reliability of the method;
- Accessibility (including price), and application speed.
One disadvantage of this method is a stereotype that fingerprint scanning is associated primarily with the work of law- enforcement agencies to find criminals. However, data security technologies are quite different from the methods of criminalist fingerprinting.The imprint of the electronic template It is impossible to reconstruct imprint from the electronic template.
The structure of papillate fingerprints patterns is very individual, has nothing in common with other people, even conjoined twins has different papillate fingerprints patterns
Researchers find some difficulties in the use of fingerprint identification:
- Change of fingerprints via injuries and illnesses that requires more flexible automated identification system
- The complexity of some scanners – fingers humidity, purity and other physical characteristics may affect the success of identifying.
- Fingerprint identification is static -a long continuous proof of identity is impossible. This method is used in systems for receiving a single personal access.
We should notice, that this method is not suitable for proctoring systems because of the possible fraud — the student can pass authentication, but control measures can be carried out by another person.
Identification by facial structure
Face detection is divided into 2D and 3D-methods, characterized by the construction of the control points system on the face image with the possibility of space extrapolation and creation of three-dimensional model or not (2D). 
At the moment, the above methods have problems with recognizing faces in low light conditions , wearing some items (hats, glasses, mustaches, beards, etc.). Some of these problems are proposed to solve with the help of cameras, working in the infrared scale.
This trend in identification is developing most rapidly, and causes the greatest interest among the developers of mobile applications and devices, and is already used in proctoring systems. This effect is caused by the simplicity of the required equipment (video cameras), as well as naturalness of human face recognition.
The advantages of this method are:
- Availability and low price of the equipment;
- Clear procedures for students;
- Use of electronic devices for monitoring student during an exam;
- The impossibility of fraud at a constant or periodic inspection of students ID.
Problem points of this method are:
- Diversity and different image quality of web-cameras, which complicates the development of detection algorithms;
- Different light conditions and the presence of other objects or hair on students' face.
This method has the greatest potential for the use in proctoring systems, mainly because of dual-use of cameras — the identification and monitoring the behavior.
Identification via thermal maps
Method of identifying a person via thermal maps is paid less attention in the literature, but this method also has perspectives. Its main advantage and specific feature is the use of camera operating in the infrared range, which allows to register parameters even in the dark. D.A. Socolinsky in his work describes in detail the problems and advantages of the method, but he recommends to use this method primarily in multimodal identification system, for example, along with the face recognition system in the visible spectrum . In his works, Diego A. Socolinsky describesthe advantages in the combination of face recognition method with infrared recording. First of all, such a system allows you to deal with the distortion of the image at the low lighting and with plenty of shadows and hot spots in the image, which may complicate the recognition.
Identification by iris
A biometric feature in the systems of identification by iris is its texture. For each person the texture of the iris is unique (the left and right eye also have a different texture), so it is a reliable biometric feature (misidentification rate is 1: 1 200 000). 
One of the problems of identification by iris is the failure of most modern systems to recognize the image of the iris accurately. This is due to eyelashes, eyelids, hot spots, rotation angle of one's eye and others , .
The advantages of this method are:
- The availability and low price of the equipment;
- Use of cameras, taking high-quality images, for identification;
- Clarity of the procedure and the possibility of its transfer into face structure identifying;
- Difficulty to break the system and for fraud;
- Use of electronic devices for monitoring student during exams;
- The impossibility of fraud at a constant or periodic inspection of students ID;
- High accuracy in the identification.
Disadvantages of this method in the proctoring usage are the need for a good camera, student should put his/her face close to the camera and demands on the lighting quality.
Identification by handwriting
Handwriting occupies a special place among other ways of identification, first of all, because writing was the subject of intense study and examinations for a long time. A handwritten signature is used for authentication in the administrative and financial institutions. The signature is the result of a complex process that depends on the psychophysical condition of the signer and the external environment (writing parameters of the device, the support, the noise and distractions). 
There are two methods of automatic handwriting recognition:
- Static (offline) method analyzes the image characteristics of the signature as a result. Characteristics of individual letters and elements are fixed. The signature is recognized as a graphical representation.
- Dynamic (online) method uses a recording device which generates a signal while writing. In this case, the signature is a combination of spatio- temporal characteristics of the tracing process.
As the recording equipment, touch screens and electronic pens are used, sometimes virtual reality gloves are used as well.
The signatures are recognized by the image parameters (points rating, contours, fragments orientation, gray level in toning pixels) or the characteristics of the pen movement (handle position, force, and direction of movement). 
The advantage of this method for proctoring system is the possibility to control the writing of texts by a student, as well as reducing the burden on a proctor (It will not be necessary to check the activity of a student).
However, for such a controversial advantage that can be achieved by other methods, this system may require additional cost and development.
Identification by keyboard typing
The principle of identification by keyboard typing is the ability to analyze temporal characteristics of the keystrokes when you enter a passphrase. Repeated entering of the same phrase carried out by a prepared user usually needs a large number of manipulation on the keyboard subconsciously, which creates the effect of a keyboard typing, thus, the user generates an automatic stereotype actions when entering the password. Monitored parameters of the keyboard input are the time you press each key of your password, as well as the time of intervals between pressing adjacent keys.
EY Kostyuchenko, and RV Meshcheryakov present the results of the study in their work, which found that the recognition by keyboard typing has accuracy of over 97%. 
The large-scale study of the recognition areas, conducted by the National Bureau of the standards in The United States allowed us to make limit assessment: the probability of correct recognition of a user with established skills of typing was 98%, which is sufficient to talk about successful practical application of such systems .
The advantage of user authentication by keyboard typing is its combination with the method of a pass phrase thus a hacker has to know as a pass phrase, as the characteristics of the keyboard typing.
The disadvantage is the instability of typing parameters in different physiological states, similar typing parameters of inexperienced users (password is typed very slowly and with one or two fingers). It makes it necessary to limit the application of the method by user authentication since the differences in the standards of a large database will be significantly correlated.
There are works that involve the definition of personality by the characteristics of phrase input. However, these technologies are more accurate and less common, and apparently, less resistant to external influence .
It is worth mentioning that this method of identification is already used in existing proctoring systems. The advantage of this method is that there is no need for additional hardware, it is simple for a user and there is possibility of concealing the identification process under a pass phrase typing.
Identification by voice
Currently there are two types of systems, identify the speaker's voice: the system, analyzing the utterance of a passphrase or phonemes, and systems, analyzing free speech.
At the moment, the speech analysis is used in computers and mobile devices for user authentication. More often in combination with any other modality — facial images , teeth , etc. The positive side of this method is the lack of need for specialized equipment, because now there is a microphone in almost any computer, mobile device or camera.
The main problem of voice recognition has been and remains the technical vulnerability of the registration system for interference and noise from both the external environment and due to the peculiarities of the structure of the microphone. Caused by this fact decrease the accuracy of recognition does not allow using this method without insurance identification system in other biometric modalities.
This method has potential applications in proctoring systems in solving technical problems, connected with noise and voice recognition algorithms.
Identification by ECG
Over the last few years it was possible to observe a number of publications, which propose the use of the ECG as a biometric measurement.ECG signal structure makes it possible to use the ECG in multimodal biometric identification systems, and, in some cases as a unimodal system.
There are several approaches to the analysis of ECG graph:
- Analysis of unique manifestations
- Analysis of differences P, QRS complexes and T, their temporal characteristics.
- Analysis of the structure of the vector cardiogram
- Analysis of biodynamic signature (BDS)
The main problems of the ECG-identification are a relatively high error rate (by 5% or more) and the vulnerability of identification because of physiological changes of a body (circadian rhythms, effects of diseases, stress, changing of a heartbeat).
Advantages of the method: it is impossible to cheat the system by means of artifacts, the ECG signal is almost impossible to imitate due to personal uniqueness, ECG does not depend on external circumstances, the training factor is excluded. 
TW Shen, W.J.Tompkins and Y.H. Hu conducted a study of the feasibility of using an electrocardiogram as a new biometric method of person identification.  As a result of their studies it was proved that the use of the ECG with one lead is enough to identify a particular person from a group of candidates. In contrast to the fingerprint identification, ECG – not a two-dimensional, but a one-dimensional approach, this places high demands to the used algorithms.
Shih-Chin Fang and Hsiao-Lung Chan  compared the accuracy of the identification by ECG recording with one and three leads. Personal identification was conducted by measuring similar and distinctive features in the signal registration. The results showed that in the group of 100 people the identification with one lead is accurate by 93%, while the identification with three-leads -by 99%.
Despite considerable efforts on the development of the ECG as a biometric modality, there remain several important questions. They include factors connected with heart rate variability, changes over time, and the characteristics of an individual's life. 
At the moment, this type of identification is considered to be one of the most promising, as it has a number of advantages:
- Dynamic parameter is analyzed, which allows continuous monitoring and authentication;
- In perspective — compactness and convenience of device, one small sensor bracelet is actually required;
- The ability to be combined with the organism status control systems;
- The inability to break the system;
- The ability to control the psycho-physiological state of a person, which makes it possible to take into account the specific human reactions at a non-standard behavior.
Identification by EEG
Humans' EEG uniqueness question gas been studied by psychiatrists and neurophysiologists since the 1960s. However, the practical application of the EEG in the field of identification appeared recently — since the early 2000s. The main problems for this method was the complexity to allocate the distinguishing characteristics of the EEG signal, the complexity of procedures and required equipment, the need to attract specialists for the procedure, as well as the low accuracy of theidentification (compared to other methods).
In 2014, a group of authors (Marcos Del Pozo-Banos, Jesús B. Alonso, Jaime R. Ticay-Rivas, Carlos M. Travieso)  presented a scoping study on the method of identification by EEG. Its results show that EEG and specificity of the alpha- rhythm signal, obtained with its help, contain unique information that can be used for classification. In general, as noted, the problem of identifying a subject, using ECG, is much deeper and more complicated than it was expected, and as a result, needs careful consideration of all the variables (time, frequency, range, algorithms, etc.). At the moment there are problems with worse recognition over time, because of a bad contact or electromagnetic noise, distorting the recording.
EEG identification is prospective in terms of control of cognitive activities, opening new opportunities in education. However, nowadays, the use of the method is faced with two major problems: the imperfection of mathematical frameworks, the specificity and complexity of the equipment requiring special registration conditions. It can be assumed that further research of this area will bring many discoveries not only in the development of biometric identification, but also in the creation of a «brain-computer» interface.
At the moment, the use of this method in proctoring systems is inappropriate, but it is interesting as an experimental research.
Identification by eye movements
Another method of identification is tracking eye movements of users. For the application of the method eye-tracker is required, a device that remotely detects the position of an eye pupil and eyes direction relative to the device.
In an experimental research, Nguyen Viet Cuong; Vu Dinh; Lam Si Tung Ho reached identification accuracy by 93.56%. 
The advantage of this method is the ability to track the direction of students' eyes, the model of material study for further improvement. However, in terms of identification, this method is not appropriate for proctoring systems as it requires additional expensive equipment. Systems using usual cameras at the moment are extremely inaccurate, and are subject to distortion of an image.
By studying the works of different authors on the most used methods of identification, the following non-specific problems for all methods were found:
- The creation and adjustment of the mathematics and software to create a biometric template which allows identifying a user with high accuracy.
- Creation of equipment meeting the requirements for economic efficiency, convenience and ease of use and providing high identification accuracy. These requirements often contradict — the simpler equipment or procedure is, the higher the probability of system failure.
- The search for new distinctive features in the recorded parameter.
- Creation algorithms that eliminate artifacts and distortion when recording.
These problems are less relevant for static methods of identification, as their specific feature is the stability of the measured parameter in time or the presence of very minor changes. Because of this feature static biometric methods are more accurate in comparison with the dynamic, but static can be hacked easier as to fake static parameter of a person is much easier. Static methods are less demanding on the mathematics in registration, because they do not need to convert, for example, the wave functions and allocate them to specific features necessary for identification. Registered parameter in these methods is studied more detailed. At the same time, static methods are more demanding on the equipment. For example, they need cameras with high-quality optics.
Dynamic identification methods are currently experiencing research growth. One can assume that their weak points will be corrected in the near future. Problems of dynamic methods are primarily determined by the parameter variability in time:
- Selection of features of dynamic characteristics themselves, or their mathematical transformation;
- Overcoming specific changes of parameter that depend on human physiological state (stress, illness, exercise, etc.);
-Dealing with a large number of artifacts and distortion in recording measured parameter;
- Some methods of dynamic biometric identification are inconvenient and require specific conditions or equipment;
- Lack of identification accuracy.
Fingerprint identification and identification by patterns of retinal blood vessels demonstrate the greatest accuracy (over 99% of correct recognition), but they are impractical for use in proctoring systems. Most studies have shown the imperfection of almost any other unimodal techniques. The authors of these studies suggest the use of multi-modal identification, which will compensate the disadvantages of individual methods and complicate system hacking.
It is worth noting that the accuracy of all methods increases significantly when applying user authentication system (identification of a particular person), rather than identification (search for matches in the database). Thus, the system has to determine the possibility of incompliance of an individual to its biometric standard.
This approach, as well as multimodality of a system allows confirming the identity of a student with high accuracy, without high cost and with a great comfort, while offering different possibilities for interaction and activities control in the exam.
Theoretical analysis and monitoring of actual teaching practice led us to the conclusions about the high potential of a multimodal system for the identification of any person in terms of distant education. We submit the following methods as the most prospective in general educational practice: identification by face structure, keyboard typing, ECG, as well as features of an iris and voice. This combination will allow identifying a user with a very high accuracy, to track his psycho-emotional state, and does not require the use of special expensive equipment.