CASPIAN JOURNAL
MANAGEMENT AND HIGH TECHNOLOGIES
DEVELOPMENT OF SECURE COMMUNICATION MECHANISM BASED ON MACHINE LEARNING METHODS
 |
Read | Vlasov Roman S., Lukashchik Elena P., Osipyan Valery O. DEVELOPMENT OF SECURE COMMUNICATION MECHANISM BASED ON MACHINE LEARNING METHODS // Caspian journal : management and high technologies. — 2020. — №2. — pp. 108-117. |
Vlasov Roman S. - Kuban State University, roma1202@mail.ru
Lukashchik Elena P. - Kuban State University, lep_9091@mail.ru
Osipyan Valery O. - Kuban State University, v.osippyan@gmail.com
Development of a secure data transmission channel is one of the solutions to the problem of customer data security, which remains a priority for modern information services. One of the solutions seems to be the creation of a mechanism that can adapt to the current state of the channel and, in order to enhance security, automatically change its cryptographic characteristics. To solve complex problems that go beyond simple functional specifications, machine learning methods are increasingly being used. In this work, during creating a system for the secure transmission of data over a secure channel, neural networks are used. The organization of the proposed cryptosystem corresponds to the classical scenario in the security field. This system consists of three neural networks: A, B and E. Networks A and B want to communicate securely using symmetric encryption. Network E wants to eavesdrop on their messages in order to receive the original message. The main distinguishing feature of the proposed security system is that it is not based on a strict cryptographic algorithm, but learns how to communicate securely over time. The adversarial functioning of neural networks leads to constant modification of the encryption system. The proposed system demonstrates the capabilities of a new security method - neural, when artificial intelligence itself develops such encryption that will be inaccessible for hacking. To date, there are very few examples of the use of neural network technology in cryptography due to its specificity. However, given the potential of the advanced information technologies, it can be argued that the application of machine learning methods in the era of quantum computers, into which we are gradually diving, will be necessary for creating optimal security systems.
Key words: безопасная коммуникация, защищенный канал связи, криптография, искусственный интеллект, машинное обучение, нейронные сети, состязательность, secure communication, secure communication channel, cryptography, artificial intelligence, machine learning, neura