Neural Based Stream Cipher Design to Secure Mobile Phone Communications

Abstract

Keystream generators produce a sequence that appears to be random, which is combined with the plaintext message using the output function. Stream ciphers consist of a keystream generator and an output function. Stream ciphers have been and still widely used for encrypting military communications, satellite communications and also for voice encryption in mobile phone networks and this may be related to their speed of encryption and decryption, since each bit of the message is encrypted independently as soon as it is read. Many types of attack have been proposed for breaking various LFSR based stream ciphers such as A5 family stream ciphers which are used to ensure the confidentiality in mobile phone networks were the most secure version known as A5/1 suffers from multiple attacks including time memory tradeoff attacks [2], correlation attacks [3][4] and algebraic attacks [1]

This paper proposes a new way based on neuron networks to build pseudo-random generators based on LFSRs. The new technique is used in this paper to overload some issues found in the A5 / 1 generators used to secure mobile communications so that it is reused to ensure data privacy in new generations of mobile telephony such as 3G, 4G and 5G. We establish some general comparisons between the output sequences of the conventional version of the A5/1 algorithm and the new proposed design to prove that the new neural generator named NeuralA5 is more secure than the conventional version of A5 /1 and has good character of the random.