CC BY
41PROTECTION SYSTEM AGAINST THE INFRINGEMENT OF INFORMATION SIGNALS IN FIBER COMMUNICATION SYSTEM
Anvarjon Rasulov
Tashkent University of Information Technologies Fergana branch anvarx@inbox.ru
Maxinur Xasanova Tashkent University of Information Technologies Fergana branch butterflytatu@gmail.com
ABSTRACT
One of the most pressing and demanding issues today in the conditions of widespread transformation and digitalization of spheres of human activity is information security and ensuring the integrity of data. The main research and development in the field of information security is aimed at improving efficiency and rationalization. One of the main means of data transmission and operation of information complexes are fiber-optic systems. To date, there have been incidents of illegal intrusion and theft of information, passing through this type of communication. Thus, today there is a problem associated with insufficient information security in fiber-optic data transmission systems. One of the most effective tools to counter acts of illegal interference in systems are artificial intelligence and cryptographic algorithms of information protection. It is the symbiosis of these two tools can qualitatively improve the level of information security in fiber-optic data transmission systems. Thus, the authors of this article pursue the goal associated with the description of an innovative system for protecting information from violations in fiber-optic data transmission systems based on the integration of intelligent cryptographic algorithms.
Keywords: Fiber-optic communication, cryptography, information security, artificial intelligence, information, data transmission system.
Introduction
The work examines the basic information, relevance and effectiveness related to the topic of the study. This research performs work through the application of statistical data and information, as well as empirical and theoretical methods of research. In this article it is used publications and materials of domestic and foreign sources to more fully disclose the topic and obtain reliable data.
Many scientific articles, papers and monographs, each of which explores more thoroughly separate aspects of the structure and systems of information security, are devoted to the approximate presented and related topics of this work. This article is based on scientific conclusions and results obtained by the authors K.A. Kudryavtseva, Sh.U. Uktamzhonov, I.A. Kosimov, D.V. Afanasyeva, A.V. Balanovskaya, O.L. Tsvetkova, A.I. Kreper and others [1-8].
In each of above works, the authors produce research that is of great relevance from the field of information security today. So, for example, questions of scenarios of connection to optical fiber cables and protection against illegal interception of information in communication channels, ways of protection of information signal from unauthorized access, modern threats to information security and more are investigated.
Artificial intelligence (AI) and machine learning technologies are already widely used in information systems to increase labor productivity, increase sales, and training. Their use in protecting against cyber-attacks is becoming one of the key areas in information security.
Total investments in companies that create information security products using AI technologies are $ 3749 million at the end of 2019. At the same time, the global market for information security products using technology AI will reach $ 30 billion in 2025 with an annual increase in 23%.
At the moment, the number of attacks is growing, and the landscape of threats is changing at lightning speed. For example, Kaspersky products reflect more than 700 million online attacks per quarter (data for the second quarter of 2019) worldwide, and Cisco claims to block 20 billion network attacks per day (more than 7 trillion attacks in 2018). It is obvious that with such volumes of malicious activity, cybercriminals are actively using tools to automate cyberattacks, including using artificial intelligence and machine learning technologies to improve and transform them, as well as to bypass known defenses. For example, the well-known Emotet Trojan is an effective prototype. The main channel for its distribution is spam phishing, and the group behind the creation of Emotet could easily use AI to amplify the attack, embedding itself in conversations natively and using natural language text analysis.
Another possible area of malicious use of artificial intelligence is better password guessing or bypassing two-factor authentication. Two years ago, researchers created a bot that was able to bypass CAPTCHA checks with an efficiency of 90% using AI technologies. By using a huge number of different data sources on the darknet to form a knowledge base of artificial intelligence, attackers can make attacks on humans truly effective.
In order to cope with the growing volume of attacks, security vendors are also beginning to actively implement artificial intelligence, machine and deep learning (ML/DL) technologies to detect, predict and respond to cyber threats in real time. Overall, according to Webroot (https://www-cdn.webroot.com), about 85% of security professionals believe that attackers use AI technologies in their attacks.
The relevance of information security in fiber-optic data transmission systems
For a long time, it was believed that fiber-optic communication lines have maximum security and information secrecy, but modern research has shown that there are ways to remove radiation from optical fibers, thus information transmitted through
them can be compromised, deleted or blocked. In accordance with the Federal Law "On communications", telecommunications operators are obliged to ensure the secrecy of communications and protection of communications equipment and facilities from unauthorized access to them. Unauthorized access to means of communication and information transmitted through them entails disciplinary, civil, administrative or criminal liability.
Contrary to the opinion that in a fiber-optic line to make a covert withdrawal of information is impossible, the methods of such connection exist and their implementation is possible in each of the enterprises, which use in the data transmission optical technology. Also, in this article we will consider methods of protection against these illegal connections. Considering the second type of threats - voice information interception, we can conclude that leakage of voice information can occur not only in operating, but also in non-operational, but laid fiber-optic networks, if an intruder artificially introduces a signal which will be modulated by acoustic waves into the cable [2].
It should be noted that the equipment used by an intruder does not necessarily have to be specialized for unauthorized data capture, it can be a variety of publicly available standard equipment, for example for the installation of communication lines. The main methods for protecting traffic from leakage in fiber-optic communication lines can be divided into three main groups of methods for protecting against the interception of such information by an intruder:
1. Physical means of information protection;
2. Hardware means of information protection;
3. Cryptographic protection of information.
Information protection is provided not by influence on leakage channel parameters, but by probabilistic transformation of information before its transmission via communication channel. Impossibility of information recovery by an intruder is based upon the property that leakage channel has lower bandwidth, than user's normal channel. Encryption method is chosen so that the number of errors arising in the leakage channel greatly increases, providing the effect of noise transmission signal, while the main channel provides a reliable connection Cryptographic method includes a method that makes the information for an attacker little useful - this is quantum cryptography, which is reflected just in the fiber-optic technology. Quantum cryptography is based on the Heisenberg uncertainty principle - it is impossible to measure one parameter of a photon without distorting another. Therefore, an intruder will not be able to change the state of the transmitted photons, as this could cause him to be exposed, by the fact of additional interference on the receiving side.
SCIENTIFIC PROGRESS VOLUME 2 I ISSUE 5 I 2021 _ISSN: 2181-1601
Analysis of integration of artificial neural networks to improve the efficiency of cryptographic algorithms for information protection
Intelligent technologies, in particular artificial neural networks (ANN), which have enormous potential in solving various complex computational tasks, are most actively studied and integrated in modern information security systems. Colossal relevance of integration of artificial intelligence in these tasks is one of the highest in the modern world within the field of study. This factor is related to the fact that intelligent technology is used not only to solve problems of mathematical and engineering nature, but also successfully proven themselves in solving problems from information security, encryption, decryption and other processes.
ANN rather firmly enters the life of the modern man in solving various kinds of problems, and are also used where primitive algorithms are inefficient or even impossible tool. The list of tasks, the solution of which is based on the use of neural networks, includes: text recognition, contextual advertising on websites, spam filtration, monitoring of suspicious transactions in the banking system, image restoration and many others [3].
Artificial neural networks are a key area of development from the field of artificial intelligence for solving information security problems. ANNs are a mathematical model that has its own implementation at the software and hardware level. Fig. 1 illustrates the scheme of a simple artificial neural network:
Fig. 1. Schematic diagram of a simple neural network. Green - input neurons; blue - hidden neurons; yellow - output neuron
Let us consider the mathematical meaning of artificial neural networks. In the mathematical interpretation ANNs are represented as a non-linear function. Under w it is characterized by connections, it is through them that signals from some neurons are fed into input signals of other neurons. Each artificial neural network neuron includes a single output, called a synapse. It should be noted that each output of a neuron is connected (or can be connected) with an unlimited number of outputs of other neurons (Fig. 2). The following mathematical model of artificial neuron is presented for understanding:
y = f(Z?=i(wi-xi + bi)), (1)
Where: Wj - represent the weights of the corresponding inputs;
Xj - represent signals at the inputs of the neuron;
bi - represent the input and weight of the displacement neuron.
Fig. 2. Schematic of an artificial neuron
Cryptography has a distinctive feature relative to other methods of protection of information flows, namely concentration of algorithmic work on physical processes and methods. Information and ciphers received by means of physical methods can be transferred and formed on the basis of objects of quantum mechanics. All processes as a whole, in this method of information encryption, take place by means of execution of physical methods. One of the examples of quantum-cryptographic algorithms work is movement of some number of electrons in electric field or photons in fiber-optic communication lines. Such a circuit includes a quantum channel and special equipment placed at both ends of the system. Fig. 3 schematically depicts the principle of operation of such a scheme for transmitting information flows [4].
Fig. 3. Cryptographic algorithm of information protection
As seen from the diagram, the key principle of quantum-cryptographic algorithms operation is uncertainty of quantum system behavior. The main idea of this principle
lies in the fact that there is no possibility to express simultaneously coordinate and momentum of a particle without parallel distortion of another.
By means of work of quantum processes nowadays various communication systems and means of transferring of information flows are widely implemented and developed, having an ability of hundred percent detection of eavesdropping and interception of information. This ability is achieved by means of the following factor: any attempt to measure interconnected parameters of a quantum system introduces disturbances into it, in parallel destroying initial data.
In recent years researches in the field of construction of methods of protection of information with use of theory of cryptography and noise-resistant coding are actively conducted and exactly these systems are most actively exposed to computer attacks. Traditionally existing information security systems do not have the possibility of self-learning and use only certain rules, laid in their software or hardware. Creation of perspective information security systems is identified recently with use of intellectual tools, such as: expert systems, fuzzy logic systems, neural networks, genetic algorithms. These approaches implement evolutionary properties of adaptation, self-organization, learning, possibility of inheritance and representation of information security experts' experience in the form of a system of fuzzy rules available for analysis [5].
Based on the theory of artificial intelligence and artificial neural networks, as well as studying the basic operation of cryptographic algorithms for information encryption, we can propose the following, shown in Fig. 4 algorithm of intellectual cryptographic protection of information in fiber-optic data networks.
Figure 4. Intelligent encryption algorithm for network communication
The increasing emergence of unwanted (malicious) software that exploits new vulnerabilities has increased the requirements for modern information security systems and has led to the use of artificial intelligence systems. Intelligence tools are actively used to solve information security problems. Classification and clustering are the main tasks solved by intelligent tools for information security (IS) of telecommunications channels in space communications, because constant monitoring of system vulnerabilities and threat fields of channels is required [6].
Performance assessment of information security system based on the principle of intelligent cryptography
AI has a significant impact on many sectors of our society and the economy (for example, predicting police, justice, precision medicine, marketing, political propaganda). AI sectoral applications are characterized by various problems and cannot be properly discussed in this report, which provides a general overview of the main issues related to the interaction between data protection and AI. Thus, this last section briefly sheds only on two main areas: the public sector and the workplace. In particular, in most cases, the introduction of AI technologies in the organization's information security reduces the time to identify problems and respond to incidents, as well as expenses for personnel management. Operators note an increase in the effectiveness of detecting unknown threats, as well as the speed of analysis and detecting malicious activity at endpoints and in applications [8].
AI Applications increases a number of specific questions when used in the public sector, largely due to the imbalance of power between citizens and administration and the necessary services provided. Moreover, the adoption of comprehensive and unclear solutions AI from governments and their agencies indicate that they are more difficult to comply with their accountability obligations, not only regarding the data in processing
[9].
This state of affairs, apparently, justifies the adoption of more stringent guarantees, except for the transfer of special committees or audit. Protection should also contain a evaluation process that critically evaluates the need for the proposed AI solutions and their suitability for the delivery of services by government agencies or private companies operating on their behalf. This process requires "at least they [AI applications] should be available for public audit, testing and consideration, as well as in accordance with the accountability standards."
To achieve this goal, state procurement procedures can impose specific duties of transparency and previous evaluation by AI suppliers.
In addition, procurement procedures may also solve issues related to the protection of IP protection of trade secrets that introduce certain contractual exceptions to increase transparency and make a possible AI audit.
As for the consequences of AI for the future of work, leaving in the direction of its influence on the labor market, AI solutions can affect the relationship in the workplace. First, they can increase the control of the employer over employees in a situation that is often characterized by an imbalance of power.
Moreover, the use of hidden and unregulated data processing forms can convert the workplace to a social experiment by raising additional important questions about the role of transparency, ethics committees and voluntary participation in data processing.
Finally, devices paid to employees by employers may have double use. For example, in the workplace at the workplace, you can wear wearable well-being devices to collect biological data designed to protect the health of the employee, but employees can also use them outside work to track their sports fitness. If only the consequences for data protection and individual freedom are not studied, such double use can blur the boundaries between the work and personal life, increasing the issues of common control and the right to shut down. In Fig. 5 is an efficiency of using artificial intelligence technologies for different scenarios:
14,00%
12,20%
12,00% 10,00% 8,00% 6,00% 4,00% 2,00% 0,00%
8,20%
6,10%
I4,70% 4,10%
I. I ill
Detecting and Application Analyzing user and Endpoint Anti-phishing
responding to projection and device behavior protection
cyber attacks vulnerability
management
Fig. 5. Efficiency of using artificial intelligence technologies for different scenarios compared to traditional antimalware systems
Having examined the relevance of the use of intellectual systems, in particular, artificial neural networks in information protection tasks, it is necessary to focus in more detail on the aspect of the integration of network data integration into the area studied.
It should be noted that according to SANS data, about 30% of information security experts are convinced that artificial intelligence technologies are able to increase the efficiency of detecting unknown threats.
7,30%
Consider in the percentage relative to the standard methods of information security metrics, improved by integrating artificial neural networks (Fig. 6):
16% 14,55% 14% -
12% 10,60%
10% -
8% -
6% 4% 2% 0%
better improved mere effective better capture more efficient
identification of (shorter) time data protection and use of threat use of personnel unknown threats between information
infection and recovery
Fig. 6. Improvement by integrating artificial neural networks compared to traditional antimalware systems
9,25%
■ 7,95%
I
Most companies actively using ANN in order to improve the efficiency of information protection systems confirm that intelligent technologies increase the effectiveness in investigating incidents, a decrease in the response time to threats, increase the efficiency of personnel management, etc. Also, many representatives of the companies are confirmed by the fact of reducing false responses as a result. Integrating neural networks in information security systems [7].
Based on the above data, it is clear that the technologies of ANN make a colossal contribution to the fight against modern information threats. In the overwhelming majority of cases, intelligent technologies reduce the operation time to identify problems and subsequent response to incidents, and also reduce the cost of managing personnel. Inc companies operating in their systems note a significant increase in the efficiency of detecting unknown threats, as well as an increase in the speed of analyzing and detecting malicious activity on servers.
Conclusion
An overview of the state of artificial intelligence segment in information security allows you to make the following conclusions:
The main methods for protecting traffic from leakage in fiber-optic communication lines can be divided into three main groups of methods for protecting against the interception of such information by an intruder:
1. Physical means of information protection;
2. Hardware means of information protection;
3. Cryptographic protection of information.
Artificial neural networks are a key area of development from the field of artificial intelligence for solving information security problems.
All processes as a whole, in this method of information encryption, take place by means of execution of physical methods. One of the examples of quantum-cryptographic algorithms work is movement of some number of electrons in electric field or photons in fiber-optic communication lines. Such a circuit includes a quantum channel and special equipment placed at both ends of the system. Artificial intelligence makes a noticeable contribution to the fight against modern information threats.
As a result of the work, it can be seen that ANN technologies are among the most innovative and breakthrough achievements of science today. These funds are widely introduced in almost all spheres of life of a modern person, ranging from domestic ones, and ending with professional. In this paper, issues related to the integration of artificial neural networks in the information security systems of modern enterprises were in more detail.
REFERENCES
1. Kudryavtseva K. A. Scenarios of connection to fiber-optic cables and protection against illegal interception of information in OV channels // Admiral S.O. Makarov State University of Marine and River Fleet. 2014.
2. Uktamjonov Sh.U., Kosimov I.A., Otakhujaev J.T. Method of protection of information signal from unauthorized access in FOCL // European science. 2019.
3. Katorin Yu. F., Korotkov V.V., Nyrkov Anatoly Pavlovich Information security in data transmission channels in the coastal networks of the automated identification system // Bulletin of the State University of the Maritime and River Fleet Admiral S. O. Makarov. 2012.
4. Afanasyeva D.V., Application of artificial intelligence in data security // Proceedings of Tula State University. Technical Sciences. 2020.
5. Balanovskaya A.V. Analysis of the current state of threats to information security of enterprises // Information Security of Regions. 2015.
6. Bogachkov I.V., Lutchenko S.S., Kopytov E.Yu. Determination of the availability factor of FOCL depending on external influences and diagnostic errors // T-Comm. 2018.
7. Tershukov D.A. Analysis of modern threats to information security // NBI-technologies. 2018.
8. Tsvetkova O.L., Kreper A.I. On the application of artificial neural network theory in solving the problems of information security // Symbol of Science. 2017.
9. Reisman, D., Schultz, J., Crawford, K. and Whittaker, M. 2018. Algorithmic Impact Assessments: A Practical Framework for Public Agency Accountability https: //ainowinstitute. org/aiareport2018. pdf.
