Anda di halaman 1dari 6

Watermarking with Fast and Highly Secured Encryption for Real-time Speech Signals

Hussain Mohammed Dipu Kabir, Saeed Anwar, Syed Bahauddin Alam, K. M. Sabidur Rahman, Md. Abdul Matin
Bangladesh University of Engineering & Technology, Dhaka, Bangladesh dipu0506@yahoo.com

Imranul Kabir Chowdhury


American International University Bangladesh, Dhaka, Bangladesh

AbstractThis paper presents a watermarking technique with fast encryption and compression. Recently a number of algorithms exist for watermarking and encryption. Their drawbacks are size, security and time for execution. Compression is necessary to maintain size small where bandwidth and storage capacity is limited. A loss-less compression technique is used to compress speech signal. Due to compression, vacant places are created. These vacant places are used for watermark. In the proposed encryption technique size of encrypted data is increased by only 12.5%. Execution of encryption is faster than present unbreakable algorithms. The proposed encryption is also highly secured and can be used for real-time application and saving signal. Watermark will exist even after signal cropping. KeywordsCompression; Watermark; Random

Traditional digital watermarking schemes are mainly based on spatial-domain or transform-domain, such as discrete cosine transform and discrete wavelet transform. In [1] Z. M. Lu, et al. proposed a novel VQ-based watermarking algorithm. After that, various VQ-based watermarking algorithms emerged, such as the algorithms based on codebook partition [2], code- book expansion [3], code words labeling [4], indices statistical characteristics [5], predictive VQ [6], lattice VQ [7], finite-state VQ [8], and MSVQ etc. The compression technique used is discussed in details in [9]. Robust, fragile and semi-fragile approaches [10], [11] are the classification of recent watermarking techniques. The robust watermarking approach protects the copyright identifier of data in which watermarks are not easily removed by attacks. The fragile watermarking approach confirms content integrity. However, speech, a one-dimension signal, faces replacement attacks, such as copy-and-paste, counterfeiting and transplantation attacks, and deletion and insertion attacks that influence speech length. The proposed algorithm can overcome these problems. As encryption system is added with it watermarking attacks are not yet powerful. In case of copy-paste, cropping, shifting original signal and watermark becomes protected. In case of counterfeiting original signal is damaged, if it is done after encryption; so none will do it. Recently watermark is also used for quality evaluation of speech [12].

number; Encryption; Real-time Application; Eavesdropping

I.

INTRODUCTION

Digital watermarking is a technique for inserting security information (the watermark) into information like image, speech, video and data frame. The watermark can be used for identification; authentication purposes and protects the integrity of data. The watermark is extracted or detected at any point where identification or integrity is concerned. In this paper, a better real-time watermarking process of speech signal is represented. The proposed scheme can be applied on the speech signal without the need of any additional space for watermarking. The scheme is secured as encryption is followed by watermarking.
___________________________________ 978-1-4244-6943-7/10/$26.00 2010 IEEE

II.

MATHEMATICAL BACKGROUND

III.

COMPRESSION AND CREATING SPACES FOR WATERM ARK

An approach for compressing speech signal is described here. In practical case a signal contains large number of same valued samples. Suppose samples are taken as 8-bit PCM; value of samples will be -128 to +127 according to 2s compliment number. The value of a sample 120 or more is rare. Most of the sample contains a value, near zero. The values of samples have normal distribution of = 6 to 25; depending on signal source. Formula of normal distribution is given in eqn. (1).

A theory of watermarking with encryption is proposed in this paper. A lossless compression technique is used to create space for watermark.

Here, sample.

= variance, = median. x = value of a


Figure 1 line Sampling of an analog signal (Horizontal grid

So the probability of a sample having a value 125 is very rare. According to eqn. (1) the probability is 3.75741096 (calculating using Matlab considering 6) to 5.962108 (considering = = 25). In a signal of

represents quantized values and vertical grid line represents samples)

A. Sampling and Compressing Speech Signal Lets consider a speech signal shown in Fig. 1. The signal is sampled after a given interval Ts s3,..., s9 = - 4, 0,0, 0, 0, 4, - 2, 3, 1. After applying compression, number of sample for water- mark is one. Samples are s1, s2, s3,..., s8 = - 4, 0, 125, 3, pw,4, - 2, 3, 1. Here,s3 = 125 means s2 is repeated.s4 = 3 means number of repetition is 3. pw = place for watermark. Only one sample is for watermark as number of repetition is only 3. When number of repetition is 100, (100 - 2) = 98 sample can be used for watermarking. B. Real-time Application In real-time system signal is sent frame by frame. Each frame may contain 10 to 60 samples, depending on transmis- sion technique. Compression will be performed inside of a frame. Such that a frame has 20 samples and its 10th to 15th samples are zero. So, after compression 10th sample is 0, 11th sample is 125 and 12th sample is 4, 13th to 15th samples can be used for watermarking. IV. ADDING WATERMARK WITH REPETITION (sampling period). After sampling values of samples will be s1, s2,

14,001 samples, no sample has value of 125. The value 127 is not so rare because when a sample have value 127 or more; the value of this sample is saved as 127. A speech of 1 second contains 8000 samples. When a speaker is telling something, 2 to 27 percent samples contain 0. So, at least 160 samples or 1280 bits for sending watermark is available. This value may be 2160 samples or 17280 bits, though speaker is telling something. The data encryption and decryption technique consists of multiple XORing of the signal with random variables. Random variables are also XORed using XORed signal. Say, a random variable R and two variables A, B are selected. To encrypt the watermarked signal, the performed steps are given in eqn. (2)-(4). A = A R (2) B = B A(3) R = R B (4) After executing eqn. (2)-(4), A and B are now like random number. A, B and R will be transmitted through the channel. To decrypt the data the steps are given in eqn. (5)-(7). R =R B =B A=A B A R (5) (6) (7)

Watermark signal will be added in places ready for watermark (pw). Watermark signal is small compared to places ready to watermark in case of human voice. So,

watermark signal should be repeated to avoid loss of watermark due to cropping. Watermark signal depends on application. The proposed watermark technique can be used in recording calls of an office. In this case phone number or terminal number and speakers ID will be saving. When it is used in voice chat using computer, watermark will save computers MAC address and IP address.

y(9) = r

(8)

Where, r =1-bit random number. Then XOR operation is performed with a random number and previously XORed bits. The sequential XORing process is shown in Fig. 2. This operation makes the signal like a noise where information is not detectable. Now, new operation is done which depends on frame size. For instance, a frame size is of 10 samples is assumed for simulation. Now, the frame is divided into two blocks A and B (Fig. 3). The XOR operation of A and B given in (9) is performed next. A=A B (9)

Figure 2

Encryption Process (

represents XOR operation)

V.

ENCRYPTION PROCESS

After watermarking, the signal is to be encrypted for secure transmission or storage. Here a fast encryption-decryption technique is followed. At present, a number of unbreakable encryption algorithms exist. These techniques are complex and take time and enlarge signal after encryption. So, these techniques cause delay and require more bandwidth. So they are not useful for real-time system. Conventional real-time encryption algorithms a r e not unbreakable, not even highly secured. For this reason, a new fast and highly secured algorithm is used. In the proposed encryption technique, size of encrypted data is increased by only 12.5% and requires less time. The encryption process is done step by step. A. Encryption Firstly, analog voice samples are converted into 8-bit PCM after sampling and quantization process. At the very beginning of encryption, samples are scrambled in any sequence. This provides additional security as without knowing the precise scrambling sequence, the original signal would never be constructed. The following scrambling sequence is chosen for simulation. y(1) = y(5); y(2) = y(2); y(3) = y(7); y(4) = y(1); y(5) = y(3); y(6) = y(4); y(7) = y(8); y(8) = y(6); Now, the 8-bit PCM is made 9-bit by concatenating one random bit with each sample. For instance, an 1-bit random number is declared in eqn (8).
Figure 4 Location of C & D in frame Figure 3 Location of A & B in frame

Next operation also depends on the frame size. Here two block C and D are selected (Fig. 4). The XOR operation of C and D given in (10) is performed next. C =C D (10)

The previous two steps ensure the security of the encryption as the sequence of the encryption can be retrieved easily by professional hackers. But the block-wise selection has far more probability than ordinary sequence. It will be very time consuming to retrieve which blocks are used for operation.

B. Sending and Receiving encrypted signal The encrypted signal will be send using a reliable media where bit error rate is small and bit error rate is detected. Check-sum bit, parity bit etc. are used for transmission error measurement [13]. If one bit is corrupted, 10 samples will be dropped. C. Decryption In decryption operation, same operations are done in reverse order. For instance, the following operation will be performed C = C D at first and y(1) = y(1) y(9) at last. After also that, 9th bit will be eliminated and scrambled bits of step 1 will be re-ordered. Decrypted signal is perfectly same to

D. Receiving Watermark and Constructing Original Signal In decryption part both watermark and speech signal is decrypted. Watermark is extracted from speech signal using the technique of compression. For example, if value of any sample is 125 and next samples value is 8, program will treat next 7 samples as watermark; as number of repetition is 8. VI. SIMULATION RESULT

This algorithm is verified by five male and five female voices. Here an example of, compression, watermarking, encryption and decryption is shown in Fig. 7-11. Some samples of compressed signal has value greater than 127, because number of repetition sample is 8-bit unsigned integer. Encrypted signal has value 0 to 511, because 1-bit is added with 8-bit, in encryption operation. Decrypted signal is same to compressed watermarked signal. Decrypted watermark and decrypted speech signal are also same to watermark and original signal. For this reason, they are not shown.

Following this entire steps sequentially decrypted signal is found. watermarked signal.

Figure 5

Block Diagram of operations before transmission

Figure 6

Block diagram of operations at receiving end Figure 9 Watermark in the created space

Figure 7

Original signal Figure 10 Compressed signal with watermark

Figure 8

Compressing and creating space for watermark where sample value-125 represents space for watermark

increased so much like AES, it requires low bandwidth for transmitting and it saves disk space for storing the data. Moreover, it is a mammoth task to decrypt if somehow the saved speech data is stolen from server where DES can be decrypted in a few milliseconds. So, it provides great security and reliability.
Figure 11 Encrypted signal of the original signal of Fig. 7

The proposed watermarking and encryption algorithm can be well used in business enterprises, telecommunication service providers, banks, call-centers and customer care centers. IX. CONCLUSIONS

VII. DISSCUSSION AND COMPARISON In the proposed watermark system, watermarked signal is also encrypted. In some watermark system encryption is absent [14]. In such systems designer have to consider various attacks on watermarking [15], [16]. A new technique of compression and encryption is used, this is making watermarked signal more secured. The watermark system described in [17] is similar to proposed watermark system. The major differences are shown below. x Proposed watermark system is for speech signal and [17] is for image. x Signal is Compressed before watermarking in proposed scheme. x Execution and security of proposed encryption system is better, as public key is not used. x In this system an user key is needed to extract watermark, in proposed system decryption and watermark code is needed for extraction. However bit error percentages of plane mode signal (not encrypted) are smaller than secure mode signal (encrypted) in wireless transmission [18]. VIII. APPLICATIONS

This paper represents a new watermarking scheme for identification, verification and authentication of real-time speech signal. The watermark is not easily detectable as the size of the frame is not increased. If the there is any attempt to change the content of the signal in the transmission channel, proposed scheme can detect it and return noise at the end. However, if any other decryption algorithm other than that is used, it also returns just noise at end. Thus it prevents eavesdropping, man-in-the-middle attack and provides security of the stored speech signal. It is only possible for someone who has the possession of watermarking algorithm scheme to get the speech signal meaningful. Finally, the user or administrator doesnt require additional memory space because of watermarking and encryption.
[1]

[2]

[3]

Watermarking with fast encryption provides secured voice communication. So it is very useful for sending speech signal from place to place. When an enterprise grows to include branch offices, e-commerce services, or global operations, general speech signal transmission longer sufficient to meet the requirement of confidentiality in business. It is also necessary to keep records of these speech signals within the enterprise. The proposed technique of watermarking with fast encryption for real-time speech can meet this need very efficiently. There is minimal delay in proposed technique for the watermarking and encryption process. As the size is not
[9] [6] [5] [4]

[7]

[8]

REFERENCES Z.M. Lu and S.H. Sun, Digital Image Watermarking Technique ol. Based on V ector Quantization, Electronic Letters, V 36, No. 4, pp. 303-305,2000. C. Lin and J. Pan, Robust VQ-based Digital Image Watermarking forNoisy Channel, Proc. ICICIC06, pp. 673-676, 2006. C. C. Chang, C. S. Shieh, J. S. Pan, B. Y. Liao and L. Hong, An Improved Digital Watermarking Scheme in the Domain of Vector Quantization, Journal of Computers, Vol. 17, No.2, pp. 19-26, 2006. Zhe-Ming Lu, et al., Digital Image Watermarking Method Based on Vector Quantization with Labeled Codewords, IEICE Trans. Inf. & Syst., Vol. E86-D, no. 12, pp. 2786-2789, 2003. H. C. Huang, F. H. Wang, and J. S. Pan, A VQ-based Robust Multi- watermarking Algorithm, IEICE Trans. Fundamentals Vol. E85-A, No.7, pp. 1719-1726, 2002. Y. N. Li, C. H. Liu, Z. M. Lu, J. S. Pan, Robust Image Watermarking Algorithm Based on Predictive Vector Quantization, Proc. ICICIC06, Vol. 3, pp. 491-494, 2006. L. Guillemot, J. M. Moureaux, Indexing Lattice Vectors in a Joint Watermarking and Compression Scheme, Proc. ICASSP06, Vol. 2, pp.11-329- 11-332, 2006. Y. C. Lin, Z. K. Huang, R. T. Pong, C. C. Wang, A Robust Watermarking Scheme Combined with the FSVQ for Images, Proc. ICITA05, Vol. 2, pp. 597-602, 2005. H. M. D. Kabir, S. Bahauddin, M. I. Azam, et. al.,"A Theory of Loss-less Compression of High Quality Speech Signals with

[10]

[11]

[12]

[13] [14]

Comparison" in European Modelling Symposeum (EMS 2010), 17-19 Nov. 2010. C. I. Podilchuk, E. J. Delp, Digital Watermarking: Algorithms and Applications, in IEEE Signal Process. Mag., Jul. 2001, Vol. 18, no 4, pp. 33-46. E. Izquierdo and V. Guerra, An Ill-posed Operator for Secure Image Authentication, in IEEE Trans. Circuits Syst. Video Technol, Vol. 13, No.8, pp. 842852, Aug. 2003. L. CAI, R. Tu, J. Zhao and Y. Mao, Speech Quality Evaluation: A New Application of Digital Watermarking, IEEE Transaction On Instrumentation And Measurement, Vol. 56, No. 1,Page: 45-55 February 2007. D. V. Hall- Microprocessor and Interfacing, Programming and Hard- ware, 2nd ed. L.V. Jiuining, L. Jingqing, and Y. Xuehua, Digital Watermark Technique Of Speech Based On Wavelet Transform, in ICSPO4 Proceedings, IEEE, pp. 605-608, 2004.

[15] O.T.C. Chen, Member, IEEE, and C.H. Liu, Content-Dependent Wa- termarking Scheme in Compressed Speech With Identifying Manner and Location of Attacks, in IEEE Transactions on audio, speech, and language processing, Vol. 15, No. 5, pp. 1605-1616, July 2007. [16] M. W. Fakhr, Member, IEEE, and C. H. Liu, A Novel Data Hiding Technique for Speech Signals with High Robustness, in IEEE Inter-national Symposium on Signal Processing and Information Technology, Cairo, Egypt, pp. 379-384, 2007. [17] P. W. Wong, and N. Memon, Secret and Public Key Image Watermark- ing Schemes for Image Authentication and Ownership Verification, in IEEE TRANSACTIONS ON IMAGE PROCESSING, vol. 10, no. 10, pp. 1593-1601, October 2001. [18] K. Kim, and J. Hong, Evaluation of Transmission and Quality Perfor- mance of Digital Secure Voice Communications in an HF Network, in Fourth International Conference on Digital Telecommunications, IEEE Proceedings, pp. 89-92, 2009.

Anda mungkin juga menyukai