SVD Watermarked Video Frame Transmission System Using Chaotic   Interleaving


 This paper suggests a chaotic Baker interleaving method for minimizing the channel errors on the transmitted data and setting some encryption. The evaluation of the suggested approach is checked by the transmission of various SVD watermarked video frames over Rayleigh fading channels in the presence of additive white Gaussian noise with interleaving by chaotic Baker map. The obtained results indicate that the received SVD watermarked video frames give higher peak signal to noise ratios (PSNRs) using chaotic interleaving compared to other methods.


Introduction
SVD watermarked video communication over wireless networks suffers from losses due to either packet dropping or fading-motivated bit errors. Thus, the robustness performance of SVD watermarked video transmission over wireless channels becomes a recent considerable hot issue due to restricted resources and the existence of heavy channel errors [1]. This work suggests a chaotic Baker interleaving technique for efficient transmission of SVD watermarked frame video over (OFDM) wireless systems with convolution code and equalization. Cases of additive white Gaussian noise and Rayleigh fading channels are considered.
SVD watermarking is an attractive technique that used for reliable multimedia transmission. The common use of the SVD watermarked video is utilized in more applications and become of interest, especially when SVD watermarked is used in cryptography with conjunction. SVD technique is the technique that inserts the hiding secret media. When mixing these methods, the hiding secret data is embedded, authenticated and encrypted. SVD adds hiding secret data which introduce authentication and encryption. The video frame is set to a group of nonoverlapping blocks and the watermark is embedded into these blocks [2][3].
The video data is transported over channels mixed with heavy conditions, e.g. burst errors that occur in groups. The burst errors effect should be minimized thus, it is used interleaving on binary data of the SVD watermarked video frames at the side of the transmitter and equalization and convolution code at the side of the receiver [4][5][6][7]. To build strong interleavers, chaotic maps scheme that depending on chaotic Baker map can be used for the watermarked SVD video frames in binary format prior to the modulation [8][9][10]. The advantages of chaotic maps randomization are minimizing the effects of the channel without the need for coding schemes for detecting the error and correcting it [11], achieving encryption in the transmitted data which adds a security to the watermarked SVD video transmission process [12][13].
Using linear equalization and convolution code eliminate the effects of the ISI and MAI [14][15][16]. While using the joint randomization method, "equalization and convolution code" enhance the robust of the OFDM systems [17]. The randomization is applied on the watermarked SVD video frames and convolution coding before transmission. At the receiver, the equalization and convolution decoding steps are applied. The suggested method estimates the signal to noise ratio (SNR) to measure the efficiency of the received watermarked SVD video frames [18]. This paper is organized as follows. Section 2 explains the background of data hiding. Section 3 explains the hiding and video scheme. Section 4 shows the suggested SVD watermarked scheme. Section 5 explains the OFDM model which used for transmission of watermarked compressed video frame. Section 6 presents the convolution code. Section 7 discusses the linear equalization. Section 8 explains the chaotic map Baker randomization. Section 9 shows the suggested model of OFDM system with the hybrid scheme. Section 10 shows the metric parameters for system performance evaluation. Section 11 shows the simulation results. At the end, Section 12 presents the conclusions remarks.

Background of Hiding Secret Data
Watermarking is an important part of data hiding, which exists to shield copyright content in multimedia. When the multimedia content is shared over the Internet exchange of multimedia content copyright infringement issues are comes into the picture. This paper gives survey of the latest methods that used for the Digital Image watermarking. These techniques can also be used for music files as well as video files. Every digital watermarking technique have certain characteristics such as imperceptibility, and robustness against various image manipulations like compression, filtering, rotation, scaling cropping etc. [19]. Digital video watermarking techniques can be broadly classified into two major categories [20]:  [22].
The goal of SVD is embedded data. The objectives of SVD scheme are detectability, robustness and capacity of the hidden data, these advantages made SVD is the important mostly than other techniques such as watermarking and cryptography.

Hiding and Video Scheme
Each video contains frames, which are compressed via a changed from the RGB coordinate color to the Y CB CR method [21]. Each part of the Y CB CR is splitted into non-overlapping blocks with size 16 x 16 pixels.
SVD is the technique which made hiding secret objects in invisible manner.
The merits of SVD based techniques are robustness, payload, invisibility, and the requirements of the size of the file. At embedding mechanism, the embed data isn't showed, and no one can found it so it is named data hiding. The hidden bits represent the payload which determines the invisibility requirement. The SVD achieves the robustness for data hiding detection. The size of the file after that embedding is unchanged. The suggested method in our work has two processes which are embedding processes and extraction processes [2][3].

SVD-Based Video Watermarking
The SVD of video frame is calculated to get orthogonal of two matrices U, V and one diagonal matrix S [2]. In [1], the W watermark is added into the matrix S then a new SVD mechanism is employed on the matrix S+kW to get U w , S w and V w , where k is the factor scale which is controlling the strength of the embedded watermark to the original video frame. Then, the video watermarked frame F w is obtained by multiplying the matrices U, S w , and VT.

Watermark Embedding:
The steps of embedding the watermark are as follows: 1-Split the video original frame (F matrix) into non-overlapping blocks.
2-Perform SVD on each block (Bi matrix) to get the SVs (Si matrix) of each block.
3-Add the video watermark frame (W matrix) to the S matrix of each block.
4-Made SVD on each D i matrix to get the SVs of each (S wi matrix).
5-Use the (Swi matrix) of each block to made the watermarked blocks in the spatial domain.
6-Combine the watermarked blocks back into one matrix to make the watermarked video frame in the spatial domain (Fw matrix).

Watermark Detection:
Having distorted video frame Fw * and Uwi, Vwi, Si, matrices, the steps to get the corrupted watermark are as follow: 1. Split the watermarked video frame (Fw* matrix) into blocks which having the same size that used in the embedding process.
2. Made SVD on each none overlapped watermarked block (Bwi* matrix) to get the SVs of each one (Swi* matrix).
4. Extract the corrupted watermark (Wi* matrix) from the Di matrices.

OFDM System Model
OFDM is communication technology is used in wireless systems of communications. High-speed data that transported via OFDM is changed from series to parallel N sub channels data [17].Then modulation, making an inverse fast Fourier transform (IFFT) on the output of data modulated that generate the OFDM multi-carrier signal. Guard intervals inserting inside frames to reduce the inter symbol interference (ISI) by the multipath fading channels, [17]. The OFDM system block diagram is shown in Fig.1. Using the IFFT/FFT cause an ISI-free channel, a circular convolution is providing by the channel [1,2]. Adding a guard band of at least v samples between OFDM symbols this cause each OFDM symbol is independent to others OFDM symbols which coming before and after it. If we assume T s is the symbol time, and T g is the CP time or time guard, the total duration of the symbol is T total [17]: The effect of the ISI can be minimized if the multipath delay or the channel impulse response is lower than the guard interval. This is occurring however, the inband fading or the inter carrier interference (ICI) is still found. The ratio between the usual symbol duration and the guard interval is dependent, as adding of a guard Ttotal =Tg+Ts. interval cause reducing the data throughput [14].The OFDM signal is transmitted through the wireless channel. A multipath fading channel and contaminated by additive white Gaussian noise (AWGN) is applied on the transmitted data.

Convolutional Codes
The encoder and decoding strategy for convolutional codes based on the Viterbi algorithm [18].The convolution code presents bits which are redundant into the stream of the data through the use of linear shift registers as shown in Fig. 2 [23]. The bits are inputted into shift registers and the output encoded bits are obtained by modulo-2 addition of the input bits and the contents of the shift registers. The code rate r for a convolutional code is defined as [23]: where k is the number of parallel input bits and n is the number of parallel output encoded bits at one time interval. K is the length for a convolutional code which is explained as:

Linear Equalization
To eliminate the ISI which caused by the multipath fading channel and improvement the performance of the communication system is made by using linear K=m+1 (10) equalization which is an efficient technique to made that. The linear equalization has different kinds in frequency domain such as the zero forcing (ZF) equalizer and the regularized zero forcing (RZF) equalizer and the linear minimum mean square (LMMSE) equalizer. The solution of ZF is defined as [4][5]: Where, H is the channel matrix. The advantages of the frequency domain ZF is source data of the equalizer are not required and statistics of the additive noise aren't required too. The disadvantages of the frequency domain ZF equalizer are that, ZF equalizer causes noise enhancement and the computations needed for matrix inversion are time consuming. A new regularization part is added into Eq.
(11) for the problem of noise enhancement in the ZF equalizer and become that [14][15][16]: Where,  is the parameter of regularization. The resulting equalizer is called RZF equalizer. So in the RZF equalizer the additive noise and the statistics of the transmitted data are not required. This causes a better equalizer which eliminates the mean square error (MSE) and removes the ISI. When =1/SNR, this achieve equalizer which is called the LMMSE equalizer. LMMSE equalizer is better than the ZF linear equalizer, as its better treatment to noise. The LMMSE is defined by [7]:

Chaotic Interleaving
A chaotic map type which used to make interleaving is called Baker map which generates a square matrix in a permuted version. The Baker map is used to randomize a square matrix of data in its discretized form which represent an M  M matrix as follows [8][9][10][11][12][13]: The perform of the chaotic interleaving steps explain as  Fig. (3-a); where, the right one is put at the top, and the left one is put at the bottom. 4. Divided each output ni  M vertical rectangle into ni boxes which its dimensions are M=ni  ni and exactly containing M points. Fig. (3-b).

The Proposed Model
The proposed model is based on chaotic map with OFDM, equalization and convolution code. There are 4 stages in this proposed model to transmit SVD watermarked video frames. The first stage is SVD watermarked frame of video data formatting, the second stage is a chaotic interleaving on the binary video data, the third stage is OFDM modulation and the final stage is equalization and convolution code. The proposed system model block diagram given in Fig.4.The proposed model based on OFDM system are explained as follows: 1. The SVD watermarked video frame is converted to the binary matrix form.
2. Reshape the binary matrix which is non-square to an M×M square binary matrix format.
3. Apply baker chaotic map randomization to square binary matrix format.
4. Reshape the square binary matrix to its original dimensions. 5. Perform the convolution codding and OFDM modulation on the binary data.
6. Apply the OFDM demodulation at the receiver this process is reversed to obtain channel estimation, assume that the receiver has the secret key of the chaotic map which the same secret key at the transmitter. After that, the received signal is equalized and made convolution decoding process to suppress the ISI, equalizer is called the LMMSE equalizer as follows: [4][5][6][7]  7. Reshape binary data to a square matrix and apply a chaotic map derandomization to binary matrix. 8. Reshape the square binary matrix to its original dimensions. 9. Retrieve the SVD watermarked video frame from the binary data.

10.Metric Parameters for System Performance Evaluation
The performance of the SVD watermark scheme under consideration is determined by using correlation coefficient between blocks and length vector containing histogram counts for correlation coefficient of blocks. It is counts the number of values in correlation coefficient of blocks that fall between the elements in the correlation coefficient vector. The normalized correlation is estimated by the following equation, where W is the original watermark and W * is the extracted corrupted watermark. Maximum of correlation coefficient is the maximum of the normalized correlation. To measure the quality of the reconstructed SVD watermarked video frames at the receiver use PSNR the peak signal to noise ratio.
PSNR is the ratio between the maximum power of a signal and the power of corrupting noise. The MSE and PSNR are defined as follows [18].  (18) where the original SVD watermarked video frames is I(x ,y), the reconstruct SVD watermarked video frames is I'(x ,y) and the dimensions of the SVD watermarked video frames are M,N.

Simulation Results
To evaluate the performance and efficiency of the proposed scheme system, various monochrome video frames are used for more accuracy such as sat.avi video 256*256*3 * 922 having 256*256 *3 pixels. These video frames are used as input to the simulation framework. So we use a chaotic map of size of 256×256 pixel for video frame of size 256×256 and we use a chaotic map of size of 256×256 pixel for video frames of size 256×256.
To test the performance of the proposed scheme carried out many simulation experiments to achieve accuracy. The model of wireless channel which is used is The SUI3 channel. The SUI3 channel is one model of six channel models adopted by the IEEE 802.16a standard. It is used for the evaluating the performance of broadband wireless systems in the 2-11 GHz band [23] from fesal.
SUI3 channel has three Rayleigh fading taps at delays of 0, 0.5 and 1ms and relative powers of 0 dB, -5 dB, and -10 dB, respectively [23]. The simulation parameters are summarized in Table (1).
We will explain the important simulation results in two cases. The first case is the transmission of SVD watermarked video frame using OFDM with only equalization and convolution code. In the second case, SVD watermarked video frame is transmitted using OFDM with equalization, convolution code and with chaotic interleaving. The simulation result is explained in the following figures:  It is found from the simulation result that, PSNR and c r max are increased when the SNR is increased. It is found from the simulation result that the case of using OFDM with equalization, chaotic interleaving and convolution code is better than the case of using OFDM with only equalization and convolution code. The better case gives high PSNR and high maximum correlation coefficient than the other case so that the proposed case is more efficiency than other case. The chaotic map for this case reduces the noise and increase the maximum correlation coefficient between extracted watermarked blocks. Table 2 is explaining the simulation results in all cases. Table (2) The received SVD watermarked Frame 189 and the correlation coefficient between blocks using OFDM in the cases. SUI-3 Raleigh fading channel influenced by AWGN is used.

Conclusion
This work introduced an efficient watermarked video frame communication