Acknowledgment First of all great thanks are due to Allah who helped me and gave me the ability to achieve this research from the first to the last step. I would like to express my deep gratitude and sincere thanks to my supervisor Dr. Abeer M. Yousif for guidance, assistance and encouragement during the course of this project. Grateful thanks for Dr. Loay E. George the Head of Department of Computer Science of Baghdad University for the continuous support during the period of my studies. Deep gratitude and special thanks to my family: my parent, brothers, and Husband for their encouragements and supporting to succeed in doing this work. Deep thanks to Dr. Haitham Abdul Lateef and Dr. Ban Nadeem Thannoon for their support, interest and generosity. Special thanks to all my friends for giving me advises. Sarah Appendix A RIFF WAVE (.WAV) File Format A.1. Waveform Audio File Format (WAVE): The canonical WAVE format starts with the RIFF header: Offset Field name Length Contents "RIFF" file 0 description 4 bytes Contains the letters "RIFF" in ASCII header The file size which is less the size of the "RIFF" description (4 bytes) and the size of 4 Size of file 4 bytes file description (4 bytes).this is usually: file size-8 "WAVE" 8 description 4 bytes Contains the letters "WAVE" header Next to the RIFF header comes first chunk 'fmt chunk' which describes the sample format: Offset Field name Length Contents 12 "fmt" id 4 bytes Contains the letters "fmt " The size of the WAVE type format (2bytes) + number of channels (2bytes) + sample 16 Size of file 4 bytes rate (4 bytes) + Byterate (4bytes) + Block alignment (2bytes) + Bitspersample (2 bytes). This is usually 16 Type of WAVE format. if 1=PCM (Pulse 20 Audio Format 2 bytes Code Modulation), values other than 1 indicate some form of compression Number of 22 2 bytes Channels :mono=1, stereo=2 Channels 24 SampleRate 4 bytes Sample per Second e.g.8000, 44100 Byte per Second or ==SampleRate*number 28 ByteRate 4 bytes of channels*bitspersample/8 Block alignment: the number of bytes for 32 BlockAlign 2 bytes one sample including all channels. Or ==number of channels*bitspersample/8 34 BitsPerSample 2 bytes Bits per sample 8 bits=8, 16 bits=16, etc. i Finally, the data chunk contains the sample data: Field Offset Length Contents name "data" 36 description 4 bytes Contains the letters "data" header Number of bytes of data is included in the Size of 40 4 bytes data section == number of samples * number data chunk of channels * bitspersample/8 Unspecified 44 Data The actual sound data data buffer A.2 Data Packing for PCM WAVE Files: In single channel WAVE files, samples are stored consecutively. For stereo WAVE files, channel 0 represents the left channel and channel 1 represents the right channel. In multiple channel WAVE files, samples are interleaved. The following diagrams show the data packing for some common WAVE file formats. Data packing for 8-bit mono PCM: Sample1 Sample2 Sample3 Sample4 Channel 0 Channel 0 Channel 0 Channel 0 Data packing for 8-bit stereo PCM: Sample1 Sample2 Channel 0 Channel 1 Channel 0 Channel 1 (left) (right) (left) (right) Data packing for 16-bit mono PCM: Sample1 Sample2 Channel 0 Channel 1 Channel 0 Channel 1 Low-order byte High-order byte Low-order byte High-order byte ii Data packing for 16-bit stereo PCM: Sample1 Channel 0 Channel 0 Channel 1 Channel 1 (left) (left) (right) (right) Low-order byte High-order byte Low-order byte High-order byte A.3 Data Format of the Samples: Each audio sample is contained in an integer i. The size of i is the smallest number of bytes required to contain the specified sample size. The least significant byte is stored first. The data format and maximum and minimum values for PCM waveform samples of various sizes are shown in the following table: Sample Size Data Format Maximum Value Minimum Value One to eight bits Unsigned integer 255 0 Nine or more bits Signed integer 32767 -32768 iii Appendix اﻟﻤﻠﺨﺺ ا ا وا ت ا ھ ر ا ا ! ات أو أ ا()'& ا%$# ، !& ا* ي .- ت ھ 0ن 2م ا$3 5$4 ت. و (4 ا6 و 7 ات وا;: ا(دة، و0> 2ة )==ت ا< ا6 ا)(- ( SSS ). م ھ@ه اn) 6,2) ط6- ( ا و 65- ، وا @ي ($6 ع & أ اع ا< ا6- ا)(- و ع Dص & م (k,n) ا ا6ي . ! ذج )== ف اJ ا6ي H I#%5< 5 4& ا%ر4K& 2د n، وأي ا4L& & ا%ر4K& #& أن %(K6 2Nدة 5ء ا J 6ي ، وإ !& ( #& ا %ر4K& & ا(P6ع اي ﺷء . 4 ا)= ا(I H6 %# زدة H. ا< إS P T 0ن درU( -P & ا(4 . اذج ا(6ح (#ن & و4H& : و Hة ا(%64 ووHه !W ا (46' ، و#ن ادDل 2$ره 2& 45 ت Yت & ع ( 4H ! ( . WAV& أن اZ ھ n & ا< . H6 6- ا(H6 6$2 64%(4& : اU]: وا( 46' . و( ذW 2& ط6 إ6Pاء I 0]: ا .S4 ا(م ا(=> DCT ، [$4\ اI4[% ، ( Quantization ) 4 46' ا =ل Run Length ) ( Encoding و، I 6اIH ا(4H ! . ( Shifting Coding ) 4 4^6& ( ال T 2 ا(%64 2 & ط6 %6 ات وا2ده ز> ا^* ( Diffusing ) ، و 4 7ت 2%ا;Generate Random Coefficients ) 4 ) (4 دت )4< H< ا %رT2 ( Generate Share Functions ) -K ا(ا 4( 4 2د n & ا< . اD($ر أداء ا)= ا(H6 ، وذW 5()ام 5_ ا I] 64 (: ا)=` ( MSE ) و $ ا0Uء و ا@روة 7ﺷره ( PSNR ) 4س ل ا)=` ! 4H& ( ا()ام $ ا U]ط ( CR ) 4س Kءة اU]: . وK \ (4. اD($ر (;34H -.% Z # & 0]: ا7ت وU( -5 >H T2 H- 2& ا7ت ا4Y- و5ت واb0 وT2 در4P -Pه Pا & ا0ح 2 ا(ع P- 5 ا(2P6* & اي H(4& 2%ا;-4 . ر ااق وزارة ا ا و ا ا ا ام #" !م ا ب ر اوت ا دام ول اب م اط ر ا ام / ا ()ء &%ت #" !دة ا م اب "%* ﺳﺎرﻩ ﺳﻌﺪ ﻋﻠﻲ اﻟﺒﻐﺪادي ( ﺑﻜﺎﻟﻮرﻳﻮس ٢٠٠٣ ) ) إ%اف ا) ذة ا(رة ﻋﺒـﻴــﺮ ﻣﺘـــ ﻲ ﻳـﻮﺳــﻒ ﻧﻴﺴﺎن ٢٠١٤ ﺟﻤﺎدي اﻟﺜﺎﻧﻲ ١٤٣٥ Chapter Five Chapter Five Conclusions and Suggested Future Wor k Conclusions and Suggested Future Work 5.1 Conclusions In the previous chapters, sharing audio files using discrete cosine transform were established and its performance was tested. Various tests were performed to study the effects of the involved coding parameters on performance. The main goal of this work is to reduce the size of the shares for the purposes of saving storage media and to speed up transmit via low bandwidth network besides guaranteeing security at low complexity. In the following points some remarks related to the behavior and performance of the suggested Share Audio Cryptography Using DCT transform is indicated and some derived conclusions are presented too: 1. The security of proposed system is acquired by several layers which add more strength. At the first layer, the attacker should guess secret coefficients, a1 and a2 , for each share at share reconstruction phase, which made recovery secret data is time consuming for him (i.e. computationally secure method). At the second layer, is the use of diffuser to prune the existing bits significance in compressed data makes shares unbiased toward local significance; this will avoid the occurrence of localization problem. 2. If any block or blocks of one share is corrupted, we still can fully recover the whole secret file from other shares. 65 Chapter Five Conclusions and Suggested Future Wor k 3. Spatial correlation property of the secret audio is eliminated when compression is performed. 4. Since the linear function is unique to a specific share, even minor changes to that share result in a dramatically different function, thereby alerting a user to potential tampering. 5. The size of shares are significantly less than the size of the secret file. 6. Table (4.9) shows Best Results the PSNR values of the reconstructed secret audio range from 40.2457 to 39.9813 dB with high MSE values. 7. Better CR without high MSE were not enough to get best PSNR values as can be seen in comparison between table (4.9) and (4.10). 5.2 Suggested Future Work The following suggestions are recommended for future work: 1. The proposed system could be implemented on image or video instead of audio wave file. 2. Instead of using wave file, other audio formats could be used in the future. 3. In Share Process Instead of (2,n) scheme make it (k,n) scheme where k<n, this will lead to make number of random coefficients will be change according to k, where needed number of coefficients and functions will be the same as k, in order to retrieve the original file by k authorized subsets. 65 Chapter Four Experimental Results and System Evaluation Chapter Four Experimental Results and System Evaluation 4.1 Introduction This chapter is devoted to present definition for the proposed scheme with simple example and discuss the results of the conducted tests to test the performance of the established system. The test was conducted on wave files with 1 channel and 16 bits per sample. 4.2 Mathematical Definition for the Proposed SSS (2,n) The Goal is to distribute a Secret Compressed File (F) into n shares {Share 1, Share 2, …., Share n} in such a way that: – Knowledge of any 2 shares makes secret file easily computable. – None of the shares appear to reveal information about the secret file – Shares size will be smaller than the secret file.