Chaotic neural network based MPEG-2 video encryption framework over wireless channel
The increasing demand for retrieving secure and high quality of multimedia service applications corresponding to available bandwidth channel proposes new challenges for system engineering designers to implement efficient and optimum solution ideas. In this thesis, chaos theory property is combine...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Subjects: | |
| Online Access: | http://dspace.unimap.edu.my:80/xmlui/bitstream/123456789/59420/1/Page%201-24.pdf http://dspace.unimap.edu.my:80/xmlui/bitstream/123456789/59420/2/Full%20text.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The increasing demand for retrieving secure and high quality of multimedia service
applications corresponding to available bandwidth channel proposes new challenges for
system engineering designers to implement efficient and optimum solution ideas. In this
thesis, chaos theory property is combined with artificial neural network to construct a
cipher cryptography algorithm called a Chaotic Neural Network (CNN). The proposed
system model framework is developed and modelled by embedding CNN inside video
codec model to produce a secure and a compress bitstream. The proposed video codec
model is designed and implemented based on MPEG-2 standard. The resultant video
signal bitstream is transmitted from source to destination by using Orthogonal
Frequency Division Multiplexing (OFDM) modulation technique. The size of tested
input video signal is 176 × 144 (QCIF standard format). The video sequence frames is
divided into sets of 30, 15, 10, and 5 frames which are fed to the framework model. The
first frame (I-Frame) for each Group of Pictures (GOP) is compressed as still image (i.e.
by using DCT transform, Quantization, Zig-Zag scan, and Huffman entropy coding),
while other frames are compressed by using motion estimation and compensation
algorithm then encoded like (I-Frame). Three Step Search algorithm (TSS) is used as
motion estimation and compensation algorithm in this thesis. Weights and biases of
CNN algorithm are set based on binary sequence generated from the chaotic logistic
map for each iterate. Control parameter and initial value of chaotic logistic map are
used as secret keys of the cipher algorithm. CNN is used to encrypt/decrypt both of
motion and quantized data vectors of video codec model. CNN algorithm shows high
sensitivity behavior for both key and plaintext modification with low PSNR value of
-18.363 dB and high entropy value of 7.833. OFDM model performance is investigated
and simulated over AWGN and 2-path frequency selective Rayleigh fading channel.
Mathematical formulation expression is given and software programming code
implementation is written by using MATLAB to simulate and test the overall system
model framework. The proposed system model framework has the ability to control the
required video quality value factor, bit rate, frames arrangement, and GOP number.
Results indicate that the transmitted bitstream has been protected from known plaintext
attack. Perceptual encryption feature was satisfied and applied successfully. Finally,
subjective and objective measurement metrics are used to verify the performance of
overall system model framework. |
|---|