Design of a novel chaotic neural network based encryption system for security applications

Abstract

Cryptography is one of the vital domains in computer systems. It is continually evolving with a tremendous amount of active research. With the widespread use of technology, there is a constant requirement for faster, secure cryptosystems. The advance of neural networks has provided us with a wide range of opportunities to optimize cryptosystems. Chaotic neural networks (CNN) are sensitive to their initial conditions. This unique property allows us to improve the security of cryptosystems. In this paper, a chaotic neural network is used to design encryption and decryption algorithms. The algorithms are experimented on using a cipher block chaining mechanism. The cryptographic algorithm was optimized by thoroughly evaluating performance parameters such as autocorrelation, cross-correlation and throughput. This paper experiments with various chaotic maps such as 1-D Logistic map, complex squaring map and cubic map to identify the most efficient map for the outlined CNN in this paper.This paper provides a symmetrical chaotic neural network for a cryptographic application that works for a wide range of inputs. The inputs are not sensitive dependent on characters. Also, the network works seamlessly with UTF-8-based characters.