Authentication means proving a person’s existence and distinguishing a real person from a non-real person. For years, authentication was based on a person’s ID, such as password, national code, ID number, etc. With the passage of time and the remarkable advancement of technology, the identification of individuals took on an identity aspect. And the presence of solutions such as people’s faces, their fingerprints, etc., so that the criterion for recognizing a unique element. It paved the way for increasing the security of individuals’ identities with online communications and the growing need for office automation. The issue of offline and electronic authentication was raised as an important need of human societies. In this regard, authentication based on biometric metrics through artificial intelligence algorithms to identify individuals has led experts to try to improve this method.
Authentication based on biometric metrics means using individuals’ behavioral and physiological characteristics for validation and identification. This article briefly describes the new ways to implement this issue and the challenges ahead.
A variety of methods of offline authentication
Authentication means distinguishing the real person from the unreal person. Authentication mechanisms on most smartphones come with a four-digit password, which is a very basic method with high hacking capability.
There are three important factors to use in authentication:
1- Information that a person has of himself.
2- Knowledge that a person has in general.
3- traits he inherits from his genetics.
Authentication based on biometric metrics means using individuals’ behavioral and physiological characteristics for validation and identification. And includes two categories of behavioral characteristics of individuals and physiological characteristics of individuals.
The advantages of non-biometric authentication methods include:
- The ease of implementation of the method in various systems with low cost.
- Ease of use for the user.
- The ability to adapt to different devices.
However, since these methods are not from characteristics, they will have an association with challenges such as providing virtual security to users.
The advantages of biometric authentication methods include the following:
- These methods have a high degree of trust and accuracy in identifying the real person against profiteers.
- Biometric identifiers include hereditary traits specific to the user, and it is impossible to forge these traits.
- Despite the high cost of providing a platform for implementing these detectors at the beginning of their emergence. This cost is declining, and manufacturers acknowledge this, which will reach its minimum in the coming years.
Given the above, in recent years, special attention has been paid to the issue of authentication based on biometric metrics. Many solutions have been presented to develop this technology, some of these methods in the research.
Criteria for selecting biometric features in the identification
Globality is characteristic in offline authentication
The selected biometric characteristic must be an adjective. That all the people of the world have it. If you select a feature, only people in certain regions of the world have that attribute. This feature can not suggest a comprehensive method of authentication.
Unique characteristic for each person in offline authentication
Although the parameter in question must be universal and have all the people of the world, at the same time, it should not be the same between two people. This feature should be unique to distinguish people from each other.
Possibility of measuring the characteristic and collecting the desired data
One of the important factors determining the desired parameter is its ability to measure and test. It should be the criterion of measurement that can be used to obtain quantitative data with modern equipment. Otherwise, no processing will occur in practice, and the desired result will not be achieved.
Having accuracy and speed in measuring in offline authentication
One of the things that the user expects from any system with any capability is the functional accuracy of the device. The algorithm should be there to include the least measurement error in the algorithm calculations. In the shortest possible time, the processing is performed, and the result of the study is determined. In other words, the algorithm must have a high speed and not lose enough accuracy in return for this speed.
Acceptability in offline authentication
A technology can progress when it has a consumer market. If, for any reason, users are reluctant to use this technology, all efforts are clearly in vain. Therefore, in the first place, it should be able to gain people’s trust in it so that its other capabilities can be test out and have examinations over time.
In addition to all the mentioned features. It is also necessary to state that the efficiency of the chosen method is more related to the costs and difficulties of providing the required resources than the efficiency of the method. And for better efficiency, you have to look for algorithms that use as few resources as possible for implementation. Thus, the selected method should be briefly reviewed for these features to be recognized as a suitable method for implementation on devices in need of authentication.
New methods of offline authentication
Active Authentication Method in offline authentication
In this method, the user’s behavior, including the user’s reaction in selecting items, the time of movements, the rhythm of repetition of each action, and the like, are analyzed and stored during one use and entry into the device. After that, each time the user intends to use the device, new information is receivable and recorded, and compared with the previous information. Identification does not occur if differences are under identification.
Genetic and Evolutionary Feature (GEFE Extraction)
One of the identification methods is GEFE identification. In this method, the user has question to answer to use the device on a trial basis once, and during this use, the movement of his mouse on the monitor and by moving his hand on the touch screen of the mobile phone is record and saved in the profile of that person. It becomes. This information is then record the next time it is useful during work, and the data has validation by matching it with the data recording in his profile.
The motion matching method in the GEFE method is based on the local binary patterns method. In this method, the image is divideable into small rectangles, and the center of these rectangles is consider as the origin; If the gray of each of the eight surrounding pixels is greater than the center pixel, that side pixel is several zero, and if the amount of gray is less, the number one is answer to it. Then all these numbers are in a vector in the whole rectangle, and it is possible to process a little on the data.
In the angle-based measurement method, we have to deal with four components: the data recording, the interpreter, the discriminator, and the decision-maker.
The data recorder is the simplest measurement component located on the user’s side and records the user’s movements on the device screen. The interpreter is on the server-side and converts the data received from the recorder into processable data. There are three main categories in this data: direction angle, curvature angle, and curvature angle ratio. These values are calculated by selecting three groups of points in the user’s image traversed with the mouse.
In the cleaning section, two algorithms, SVM (Support Vector Machine) and NMD (Normalized Manhattan Distance), are used. In the first method, the similarity between the receiveable data and the individual’s profile is recording. The average difference between the two data sets is measuring in the second method.
ABM + GEC method (Angle Based Metrics + Genetic and Evolutionary Computation)
The third method complements the second method and has reduced the shortcomings of the second method with genetic algorithms. In the previous method, due to the type of algorithm implemented. Also, the number of features required for data analysis increases unpleasantly and reduces the performance of the existing method. Through a genetic algorithm call X-TOOLS, a connection is between these traits, and it categorizes them. These categorization of data are then evaluate and processing. This way, the complexity of existing data relationships is reducable to about 94.4%.
Signature pattern method (3-D hand gesture signature)
Electronic authentication based on biometric methods is to receive the user’s hand when signing electronically. In this way, the user adjusts the shape of his hand in front of the sensor as if he were signing. In such a case, the receivers installed on the smartphone receive these signals according to the signal generated by the movement of the hand, and by amplifying these signals, they convert them into processable data.
Different signals are receivable based on the hand components’ distance from the telephone surface. The fake person is distinguisheable from the real person by processing these received signals. This method of identification is in the modeling category. Similarly, the user is ask to sign in the air. Then, using sensors on the phone and distance, the user’s hand movement is storeable in a database and compareable to a signature previously register from the user. Any discrepancy can be a reason for the person to be untrue. Of course, in the meantime, data classification algorithms should be useful to increase the algorithm’s accuracy.
According to what has been, there are various methods for recording and analyzing user information. We can mention the user behavior pattern and the model of drawing a person’s signature.
Advantages of authentication methods
Advantages of GEFE and ABM methods and ABM + GEC combination
One of the advantages of the GEFE and ABM methods is that they are compatible with any platform. Also for receiving and measuring information. The measurement of how the mouse moves on the screen and the results are. Much more accurate and reliable than the data obtained from the speed of people on the screen. Still, by comparing the three methods GEFE and ABM and the combination of ABM + GEC. The accuracy of the first method will be higher than the others at the beginning of the test. Still, over time, this accuracy has decreased in competition with the accuracy of the other two methods. Finally, it reaching the lowest value, and the accuracy of the third method was measuring more than others.
Advantages of the signature template method
- This detection is performeable using biometric metrics.
- There is no need for additional hardware to implement the method, and the sensors available on mobile phones, including a variety of speedometers, are useful to implement the algorithm.
- As people move more slowly with age, the sample stored in the database gradually adapts to the new conditions and trains its model so that the algorithm does not lose its functionality after a while.
Based on the above, you can conclude that according to the different algorithms available for authentication base on biometric metrics, which method can be useful for evaluation following the require accuracy and speed.
Conclusion
Today, due to the growing need to increase the security of electronic communications and the comprehensive advancement of technology are needed to be able to share information more confidently in cyberspace and electronics. Over time, having only one password and identification name as authentication will not be, and one must prove its authenticity through its existential nature. Biometric science refers to technology for measuring and analyzing the characteristics of a person’s body to identify a person. All biometric systems have a unique architecture for sample processing and authentication. There are several methods of identification in biometrics, each of which is useful according to accuracy and efficiency.
One of the challenges biometric metrics authentication faces is the high cost of implementing this method on the devices used, which is directly related to the volume of calculations required to verify and record information and, consequently, the supply cost. It has hardware. Optimizing the mathematical relationships governing the methods of recording and retrieving information and the maximum and optimal use of available resources can pave the way for future studies.
In this article, we tried to review some of the most efficient and modern methods of authentication based on biometric indicators and get acquainted with its features and disadvantages. Among the methods studied is the technology of authentication based on the state of individuals’ hands and identification based on their reaction speed, which is based on not having exorbitant costs to companies producing smart devices. We will see a revolution in this industry shortly.