Encryption Based on Neural Networks

Author Name(s): Alexey G. Isavnin, Izida I. Ishmuradova, Anton N. Karamyshev, Denis M. Lysanov, Irina I. Eremina
Author Email: IIIshmuradova@kpfu.ru

Abstract

Due to the active development of information technology, information is of great value to date. Any sphere of public life can be described by information, loss or modernization of which can lead to large losses. Thus, information becomes strategic resource of the state and business of all levels that are interested in its preservation. In addition to the natural risks of information loss (failure of technology, natural disasters, etc.), there is also the desire of criminal structures to illegally kidnap or update information. The problem of protecting information is extremely urgent nowadays. Companies require often applications with the most enhanced protection for sending messages. In this article, we implement a scheme for sending encrypted messages between several parties.On the basis of neural networks, we implement symmetric encryption with the distribution of keys. A scheme for synchronizing neural networks using a tree of a parity machine will be constructed.

Introduction

Due to the active development of information technology, it is of great value to date. Any sphere of public life can be described by information, loss or modernization of which can lead to large losses. When transmitting information over the network, it is processed at each node, according to the rules specified in the data transfer protocol. Also, each node (router, switch, computer) is a system in which various rules of information processing can be present. Thus, when information is transferred within a computer network, a dual problem arises: on the one hand, to ensure the security of information in a single system with unified rules for processing information, and, on the other, to monitor the integrity, confidentiality and availability of information in the aggregate of individual systems with different rules for processing it[1]. There are many modern applications working on the safety standards of industrial control systems that require the safe transfer of sensitive data. It is possible to create a connection using cryptographic protocols. This is done for the fastest and most efficient encryption method using symmetric encryption algorithms. The structure of symmetric encryption is clearly shown in Figure 1.

Conclusion

In symmetric encryption algorithms, we were able to examine and analyze one of the new ways to solve the key exchange problem. This method is based on artificial neural networks on the TPM synchronization. We used a remote server acting as a comparator to apply this method for messaging. We conducted experiments and tested our system, and also brought forward our arguments and suggestions for improving further work. Such actions will allow to get rid of comparison keys on the server, make the system much safer and to prevent unpleasant incidents with intruders [7,8].

727 total views, 1 views today

Download PDF File

About the author: admin