Implementation of Hash Function Based On Neural Cryptography

In this paper a new hash function is constructed based on multilayer feed forward network with piecewise linear chaotic map. Chaos has been used in data protection because of the features of initial value sensitivity, random similarity and ergodicity. We have used three neuronal layers to prove confusion, diffusion and compression respectively. This hash function takes input of arbitrary length and generate a fixed length hash value.(128 bit, 256 bit or 512 bit). By performance analysis and results we have shown that generated hash function is one way, collision resistant and secure against birthday attacks and man in middle attacks. Keywords—Neural Network, Cryptography, Tree parity machines, Piecewise linear chaotic map, hash function, plain text sensitivity, data signature