A Neural Key Generator for a Public Block Cipher

Abstract

In this paper, the applicability of using an artificial Spiking Neural Network with a symmetric block cipher is explored. Our goal is to develop a novel neural block cipher where the keys are generated by a spiking neural network and can have any desired block length. With the new algorithm the private keys do not have to be exchanged and present a stronger process of key scheduling. The system allows a rapid change in encryption keys and a network level encryption to be done at very high speed without the problem of factorization of other systems. The block cipher will be transformed in a public cryptosystem, less vulnerable to brute force attacks, and it is hoped to be also resistant to linear attacks since the spiking neuron network architecture brings non-linearity to the encryption/decryption process.