Размер шрифта Цветовая схема Изображения


For admission call or message:


To come to Stavropol write:


Efficient and secure system for confidential data storing and processing

Project Name

An efficient, secure and fault-tolerant system for sensitive data distributed storage and processingwith controlled redundancy for designing mobile clouds on low-power computing devices.


Leaderand Major Members

Leader: Mikhail Babenko – PhD, Physical and Mathematical Sciences; Associate Professor, Department of Applied Mathematics and Mathematical Modeling

Team members: Deryabin M. A., Kucherov N. N., Kuchukov V. A., Kuchukova N. N., Nazarov A. S., Kuchukova E. A., Golemblevskaya E. I.

Support Mechanisms

The project enjoyed supported based on the results of contest for grants (2019) from the Russian Research Foundation for initiative research by young scientists under the President’s program of research projects launched by leading researchers, including young scientists (Project Number: 19-71-10033)

Departments and Partners involved
Department of Applied Mathematics and Mathematical Modeling, Institute of Mathematics and Natural Sciences

Project Summary

Security and reliability are very important issues in terms of maintaining the confidentiality of data stored and processed encrypted. Complexencoding algorithms require a lot of hardware resources. Using classic cryptographic primitives will lead to issues operating mobile devices: the battery will run outquickly while the device memory will be full.

The proposed data protection system allows avoiding these restrictions and offers a high level of data security with minimumresources. The speed of data encryption in real time will not onlydecrease, yet will also increasemany times. This effect can be achieved through using artificial neural networks and modular arithmetic.

The proposed system is adaptive, i.e. it can be optimized for the available infrastructure (technical features of the device, such as its performance, memory capacity, or the communication channel used), which is very convenient. The user can choose one criterion for adjusting the system following the environment parameters and their goals. In the future, researchers plan to offer potential for adjusting the system to several criteria at once, which will take studying how individual factorsaffect each other.

Another important advantage of the system is its reliability – it is not very sensitive to technical failures. The proposed method for error detection, localization and correction exceeds the available methods of projection and syndrome by 68% and 52%, respectively, in the worst-case scenario, where the error detection time is the longest.

Current Outcomes

  1. A data encoding algorithm has been developed, which allows increasing the data encoding speed for homomorphic encryption algorithms by more than one magnitude order, due to the use of a finite ring neural network.
  2. A modification of the data decoding algorithm has been proposed, which allows increasing the decoding speed for homomorphic encryption algorithms by more than two orders of magnitude, based on the use of polyadic code and a finite ring neural network.
  3. Two weighted code error correction scenarios have been developed. The first scenario is based on weighted access structures, with its properties examined. The other scenario relies on generalization of the idea by I. Y. Akushsky and D. I. Yuditsky and the entropy approach, which allows building weighted codes for detecting, localizing and correcting errors. The advantage of the proposed entropy approach involves an increase in the number of detected and corrected errors.
  4. The issue of choosing residual class system modules has been examined in order to reduce the computational complexity of the Barrett reduction algorithm. The link between the selected RCS bases and the accuracy of the outcomes has been detected. Two scenarios have been proposedfor selecting RCS bases, which allow calculating the result of multiplication by p module within the range [0, 2p) and [0, 4p), respectively.
  5. A generalizing mathematical model of the core function by I. Ya. Akushsky has been designed. Subclasses of core functions by I. Ya. Akushskyhave been identified to compare numbers, identify the sign of a number, and find its rank, respectively. The selected subclasses of core functions allow reducing the computational complexity of algorithms if compared to the Wang method, the diagonal function, the Chinese remainder theorem, the Vu method, etc. Using the core function allows effective implementation of operations for comparing and determining the sign of a number with encrypted numbers.
  6. Cryptographic properties of homomorphic ciphers have been investigated. A scenario for attacking the existing homomorphic codes using an open-text attack has been designed.
  7. The issue of dynamic problems distribution has been examined, and it has been shown that simultaneous use of RCS and homomorphic ciphers will reduce the uncertainty of the time required to obtain the result for matrix calculations.
  8. Anextremum search mechanism based on a mathematical model of representation of a neural network as an information transmission system has been developed. In the future, we will employ this approach to improve algorithms for matrix calculations with encrypted numbers.

Through the project implementation, 5 papers have been published in journals indexed by Web of Science, 2 software registration certificates have been obtained, and an application for an international patent has been submitted.

Major publications

  1. Tchernykh A., Babenko M., Chervyakov N., Miranda-López V., Avetisyan A., Drozdov A. Y., Rivera-Rodriguez R., Radchenko G., Du, Z. Scalable Data Storage Design for Non-Stationary IoT Environment with Adaptive Security and Reliability // IEEE Internet of Things Journal, 2020, Q1, IF=9.515
  2. Babenko M., Deryabin M., Tchernykh A. The Accuracy Estimation of the Interval-Positional Characteristic in Residue Number System // 2019 International Conference on Engineering and Telecommunication (EnT), IEEE, 2019, pp. 1-5
  3. Vershkov N.A., Kuchukov V.A., Kuchukova N.N., Babenko M.G. The Wave Model of Artificial Neural Network // IEEE Conference of Russian Young Researchers in Electrical and Electronic Engineering (EIConRus), 2020, pp. 542-547
  4. Kucherov N.N., Deryabin M.A., Babenko M.G. Homomorphic encryption methods review // IEEE Conference of Russian Young Researchers in Electrical and Electronic Engineering (EIConRus), 2020, pp. 370-373
  5. Nazarov A. S. Development of methods and algorithms for building fault-tolerant distributed data storage systems based on modular arithmetic: dis. – 2019.
  6. Certificateofsoftwareregistration2019661480 of19/07/2019.Softwareprogram for modular neural network data decoding. Authors: Babenko, M. G., Kuchukova E. A.
  7. Certificateofsoftwareregistration2019661394 of19/07/2019. Software program for modular neural network data encoding. Authors: Babenko, M. G., KuchukovaE. A.

Expected Outcomes

The project framework implies studying the algorithms of division by constant and division with remainder for encoded numbers, modular operations for their practical application, as well as encoding and decoding algorithms.

The obtained outcomes will be presented at international conferences, with four papers published as well as an application submitted for an international patent.

Potential Application

The proposed approach can be used on low-power devices to ensure high reliability and security of data storage.