A chaotic neural network based encryption algorithm for MPEG-2 encoded video signal

Abstract

In this paper, a cipher system based on Chaotic Neural Network (CNN) is used to encrypt and construct a stream cipher of compressed MPEG-2 video signal. The symmetric CNN cipher algorithm transforms the plaintext (compressed video data) into an unclear form under the control of key; this algorithm has high security and simple architecture with low cost hardware implementation. However, when the size of input layer of the neural network is increased, the required execution time for encryption and decryption process will be decreased. Two main contributions points are discussed in this paper. The first one is related to chaos properties and its effect on CNN cipher algorithm. Results show that a very small key modification in the receiver side will lead to an unintelligible video scene with low PSNR value of -18.363 dB. The second contribution point is related to the bit rate control and video quality of the whole system model. It can also be shown from results that by increasing video quality value then the PSNR and compressed bit rate values will increase, but with penalty vale of compression ratio. The whole system model can control the bit rate and video quality depending on communication channel conditions. The proposed system model is suitable for secure video transmission applications and wireless multimedia communication.