Method and Means of Symmetric Real-time Neural Network Data Encryption

The training of the auto-associative neural network by the method of principal components analysis has been conducted and the matrix of weight coefficients has been calculated. A method of symmetric encryption/decryption of text messages for an auto-associative neural network has been developed. The method provides reducing the time of encryption/decryption of text messages due to the use of pre-calculated weighting matrices. The system of symmetric encryption/decryption of data has been synthesized. Secret keys of the system are the architecture of an auto-associative neural network, a matrix of weighting coefficients, and a bit width of weights masks. Three software tools that teach the auto-associative neural network to encrypt and decrypt data have been implemented for the Raspberry Pi microcomputer using C language. The testing of developed means of encryption/decryption of text messages with different secret keys was performed, which showed the repetitiveness of the obtained results.