Penerbit Akademia Baru Journal of Advanced Research in Computing and Applications ISSN (online): 2462-1927 | Vol. 2, No. 1. Pages 39-54, 2016 Enhanced Security Audio Steganography by Using Higher Least Significant Bit S. Krishnan*,a, M. S. Abdullah b Advanced Information School, Universiti Teknologi Malaysia, Jalan Sultan Yahya Petra (Jalan Semarak), 54100 Kuala Lumpur, Malaysia a,*[email protected], [email protected] Abstract – Data hiding is an approach that conceals secret data into a carrier. Steganography is the art and science of storing information so that its existence is hidden. The goal of steganography is to communicate securely in a completely undetectable manner and hiding data efficiently to avoid unintended recipients to suspect the existence of steganographic medium that contains hidden data. The study had found that that the audio steganography using LSB technique can be easily broken to retrieve the data and vulnerable in security due to its low robustness. In this research, the scheme of Audio Steganography will be investigated to enhance LSB scheme with AES encryption and evaluate the robustness and quality of the scheme. This research presents Enhanced Security in Audio Steganography by using Higher Least Significant Bit to improve security and robustness by embedding bits of secret message in higher LSB of a cover audio. The stego key and AES encryption are applied to increase the security of the stego audio. Encoding process embeds the secret message while decoding process retrieves the secret message. In audio steganography, the secret message is embedded into a digitized audio signal which results in significant altering of binary sequences of the corresponding audio file. The result shows the scheme is robust and able to withstand steganography unintentional attack which is compression till 17% at maximum payload of 953 bps at 11th LSB. This new scheme model is proven to increase the robustness using higher LSB and will be useful as a basis for Audio Steganography.Copyright © 2016 Penerbit Akademia Baru - All rights reserved. Keywords: Audio Steganography, LSB, Payload, Imperceptibility, Robustness 1.0 INTRODUCTION Steganography is one of the mean used for safe and protected transmission of confidential information [1]. Hiding information in audio is safe and less doubtful than communicating an encrypted file. The main purpose of steganography is to convey the information secretly by concealing the very existence of information in some other medium such as image, audio or video [2, 3]. These objects are called cover objects or carrier objects of the steganographic method. A review on list of journals show that LSB method of audio steganography is vulnerable for data leakage of secret message on a steganography medium [4, 5, 6]. LSB technique is simple and modest that allows hiding secret message on least significant bits. This technique can be easily broken to retrieve the secret message due to its low robustness. Therefore, the hidden 39 Penerbit Akademia Baru Journal of Advanced Research in Computing and Applications ISSN (online): 2462-1927 | Vol. 2, No. 1. Pages 39-54, 2016 data should be retrieved by authenticate user without any tolerance. This shows that there is an immediate need to protect the secret message by enhancing the security of the LSB technique. Many researches have been carried out to enhance the quality and performance of audio steganography, but yet the robustness of LSB technique still can be compromised. Most of the techniques are less effective to provide a robust and strong LSB method. In contrast, the proposed research is an enhancement of the security of the LSB scheme. The term Steganography derives from the Greek steganos which is protected or secret and graphy which is writing or drawing [7, 8]. In other words, Steganography can be well-defined as the art of storing information to be hidden and concealed [9, 10]. The goal of steganography is to establish a secure communication in an absolutely untraceable way and to hide data well enough that unintentional recipients do not doubtful about the steganography object which comprehends secret data [11]. Cryptography and steganography are inimitable ways to defend information from intruders nonetheless technology and skills are perfect to lead those ways to be compromised [9, 12]. The purpose of steganography can be considered partially defeated if the hidden information is revealed or even suspected, [9, 13]. Cryptography supports in keeping the message as a secret while steganography supports in keeping the existence of the message as a secret. They can be considered as defeated if the data hidden is discovered [13]. Table 1: Summary of Audio Steganography Techniques [4] Method Strength Drawbacks Least Significant Simple, Modest Easy to extract and Bits retrieve data Parity coding More robust than Easy to extract and LSB retrieve data Echo hiding Evades problem Low capacity with additive noise Tone insertion Exploits masking Low embedding property capacity Phase coding Robust, Higher resistance Low capacity Spread spectrum Increases transparency and Occupies more imperceptibility bandwidth Wavelet domain Capable to hide large amount Lossy data retrieval of data, higher hiding A portion of data might be capacity and lost due to compression. transparency The basic model of Audio Steganography consists of Carrier (Audio file), Message, Password, Stego file. The carrier is also known as a cover file which conceals the secret information [5, 14]. Audio steganography is an apprehensive with a data hiding in cover audio signal in an imperceptible way. The secret message is embedded into a digitized audio signal which results into altering binary sequence of the corresponding audio file [7, 15]. Table 1 shows a summary of audio steganography techniques. Least Significant Bit is one of the most simple and modest technique of data hiding compared to other domain with its different techniques studied. This simplicity allows for frequent modification and alteration of 40 Penerbit Akademia Baru Journal of Advanced Research in Computing and Applications ISSN (online): 2462-1927 | Vol. 2, No. 1. Pages 39-54, 2016 the techniques which is desired by the users. Unfortunately, Least Significant Bit shows the tolerance to extract and retrieve data and which is vulnerable for data hiding compared to other domain with its different techniques studied and become the drawbacks for this method compared to other domain. In this section, there are many ways to implement steganography especially in audio which can be combined with encryption to enhance its security. Eventually, the LSB technique needs a thorough alteration to form Audio Steganography with good robustness and security. This leads to the next stage of proposing the methodology, which is discussed in the next section of this paper. 2.0 METHODOLOGY 2.1 Proposed Model The Proposed Model for audio steganography to enhance the security of the LSB method has been designed as presented in Figure 1. The proposed model output will be an encrypted stego- audio which is similar as cover audio. The extracted secret message during the decoding process and AES decryption would have a trade-off between capacity and robustness. The capacity will be a trade-off with robustness which uses higher LSB and another layer of security which is AES encryption. Figure 1: Proposed Model Table 2: Process in Each Phase at Sender and Receiver Phase Sender Receiver Phase 1 Secret Message & Cover Stego Audio Receive Audio Conversion to Binary Phase 2 Identification and AES decryption authentication Phase 3 LSB Embed Secret Identification and Message with Stego Key authentication Phase 4 AES encryption Inverse LSB Extract Secret Message with Stego Key Phase 5 Stego Audio Transmit Binary Conversion to retrieve Secret Message 41 Penerbit Akademia Baru Journal of Advanced Research in Computing and Applications ISSN (online): 2462-1927 | Vol. 2, No. 1. Pages 39-54, 2016 This proposed model consists of two main parts and these are the sender and the receiver. The sender involves encoding process while the receiver functions in decoding process. Each part has been divided into five phases for simplicity and tolerance during the execution of the proposed model as shown in the Table 2. Figure 2: Encoding process for AES encryption and LSB embedding A secret message in the form of ASCII characters is embedded within a carrier of an audio file (.wav) with LSB function. After the secret message insertion into cover audio with LSB bits selection, the Cover Audio will be encrypted again. The output will be similar to the cover audio with a secret message embedded inside it which also known as Stego Audio. The Proposed scheme consists of the following two main parts as below: (i) Sender side (Encoding process) (ii) Receiver side (Decoding process) 42 Penerbit Akademia Baru Journal of Advanced Research in Computing and Applications ISSN (online): 2462-1927 | Vol. 2, No. 1. Pages 39-54, 2016 2.2 Encoding Process Encoding Process on Sender Side (Secret Data Hiding) takes place at the sender side to hide the secret information. The carrier audio file and a secret message will be converted to binary formats. Bits of secret information will be hidden using modified LSB scheme in carrier audio file. The steps of encoding process in Figure 2 are as follows: (i) Convert the cover audio file and secret message to binary format. (ii) Stego key configured in 8 bits for identification and authentication. (iii) Hide the secret information into a cover audio file using LSB scheme. (iv) Generate the stego key. (v) AES encryption applied. (vi) Key and the stego audio are sent to the receiver. 2.2 Decoding Process Decoding process will take place in the receiver side to extract the encrypted secret information and decrypt it to obtain the secret message the inverse of the modified scheme. The stego key is used to gain access to the secret information. Secret message can be retrieved with appropriate and correct parameter values which corresponding from the encoding process. The steps of decoding process are as shown in Figure 3 and as follows: (i) The receiver will decrypt the AES encryption of stego audio. (ii) The decrypted Stego Audio will be authenticate with its stego key. (iii) The Cover Audio will be removed from the extracted Stego Audio. (iv) The hidden bits will be extracted accordingly to its hidden scheme or pattern of LSB. Header’s bits, length of secret message’s bits and the secret message’s bits will be extracted accordingly. (v) The binary formats of secret message will be converted to decimal and coupling with ASCII code to retrieve its Secret Message. (vi) The secret message will be displayed immediately. 3.0 RESULT AND DISCUSSION 3.1 Hiding in Higher LSB Initially, a cover audio file which is ‘.wav’ file is chosen for the data hiding. This is because Matlab R2014a works easily for modification and analysis purpose in ‘.wav’ file. Other formats show much more error due to smaller size or compressed features. The composition of key embedding elements in an audio sample in cover audio is visualized in Figure 4. The first 40 bytes are used to make as a header for the cover audio. The header function as a reference point for other bits modification. The next 3 bytes which is from 41st byte to 43rd byte is used as a length of wav data samples of the cover audio. The next 16th bit will be the space to embed the secret message, which is from 44th to 60th byte. The higher LSB for embedding the bits from the secret message will be determined in the encoding process. 43 Penerbit Akademia Baru Journal of Advanced Research in Computing and Applications ISSN (online): 2462-1927 | Vol. 2, No. 1. Pages 39-54, 2016 Figure 3: Decoding process for AES encryption and LSB extraction Figure 4: Diagram of Composition of an Audio Sample in Cover Audio for Embedding Pseudocode for Encoding in Matlab R2014a as shown below: 1. Set total to zero 2. Open a wav file for hidding text Call uigetfile 44 Penerbit Akademia Baru Journal of Advanced Research in Computing and Applications ISSN (online): 2462-1927 | Vol. 2, No. 1. Pages 39-54, 2016 Call wavread 3. Store the header and the length of wav data samples Call uint (unsigned integer) 4. Copy the 16 bit wav data samples 5. Select LSB to embed Call lsb 6. Insert secret message 7. Binary conversion and reshape msg_double=double(msg); msg_bin=de2bi(msg_double,8) [m,n]=size(msg_bin); 8. Insert stego key 9. Embed the bits of stego key and secret message 10. Generate new .wav file – stego audio Call fopen 11. Compare cover audio with stego audio for PSNR for j = 1:size_host s = s+(stegoaudio_double() - coveraudio_double())^2 12. Plotted Normalized Cross Correlation (NCC) for both audio 13. Apply AES encryption While Pseudocode for Decoding as shown below: 1. Set total to zero 2. Apply AES decryption 3. Open a wav file for extracting text Call uigetfile Call wavread 4. Open stego audio if stego key Call fopen Else end 5. Binary conversion Call uint (unsigned integer) 6. Extract the header and the length of wav data samples Call fread 7. Extract 16 bit wav data samples Call fread 8. Select LSB to extract Call lsb 9. Secret message Size Bits retrieval m_bin( )=bitget(dta( ),lsb) n_bin( )=bitget(dta( ),lsb) 10. Secret message bits retrieval secmsg_bin( :len)=bitget(dta(:+len),lsb) secmsg_bin_re=reshape(secmsg_bin,len/8,8) 11. Convert to ASCII code secmsg_double=bi2de(secmsg_bin_re) 12. Display secret message 45 Penerbit Akademia Baru Journal of Advanced Research in Computing and Applications ISSN (online): 2462-1927 | Vol. 2, No. 1. Pages 39-54, 2016 3.2 Steganography Attack Compression is one of the unintentional attacks which will be performed on the audio steganography scheme to measure the ability of embedded data to withstand against the attacks. This attack refers to common data manipulation which is compression and decompression. Different level of compression will be performed on the stego audio to validate the robustness of this scheme. This section presented the proposed model to enhance security in Audio Steganography by using the higher LSB scheme and AES encryption. The encoding and decoding processes are developed accordingly to meet the requirements of higher LSB to fulfill a better robustness for this scheme. Therefore, this research will implement the proposed scheme and the outcome will be analyzed and validated in the next section. 3.3 Analysis Embedding a secret message which is ‘Surenthiran Krishnan’ within a cover audio and the process of decoding the hidden message from stego audio, and finally retrieval of the secret message as shown in Figure 5. Binary conversion and reshape in the vector form plays an important role to carry the bit, channel the bits for embedding and for extraction as well. A good vector matrix should be an appropriate size to fit the possible secret message size. A suitable vector determines the performance and stability of a scheme in Matlab environment. Figure 5: Flow Chart of Encoding and Decoding in Higher LSB Scheme in Matlab R2014a This proposed higher LSB scheme does not stand for high compression due to the stronger congestion of bits. However, it may stand compression at a level of 17% and below. The maximum compression that can be applied is 17% as shown in Table 3. 46 Penerbit Akademia Baru Journal of Advanced Research in Computing and Applications ISSN (online): 2462-1927 | Vol. 2, No. 1. Pages 39-54, 2016 Table 3: Robustness Validation through Compression Compression Ser LSB Selection Cover Audio (kB) Secret Message Size (kB) Percentage of Payload (%) Stego Audio Size after Compression (kB) Percentage of Compression (%) Audio file format mrAeseetbsrtciioslreai evtgtye e 1. 4 1820 0.035 0.0357 390 80 .acc No 2. 4 1820 0.035 0.0357 203 90 .ogg No 3. 4 1820 0.035 0.0357 728 61 .wma No 4. 4 1820 0.035 0.0357 392 79 .m4a No 5. 4 1820 0.035 0.0357 501 74 .ape No 6. 4 1820 0.035 0.0357 1220 33 .wv No 7. 4 1820 0.035 0.0357 529 72 .flac No 8. 4 1820 0.035 0.0357 39 98 .mmf No 9. 4 1820 0.035 0.0357 1596 17 winrar Yes 10. 8 1820 0.035 0.0357 390 80 .acc No 11. 8 1820 0.035 0.0357 203 90 .ogg No 12. 8 1820 0.035 0.0357 728 61 .wma No 13. 8 1820 0.035 0.0357 392 79 .m4a No 14. 8 1820 0.035 0.0357 501 74 .ape No 15. 8 1820 0.035 0.0357 1220 33 .wv No 16. 8 1820 0.035 0.0357 529 72 .flac No 17. 8 1820 0.035 0.0357 39 98 .mmf No 18. 8 1820 0.035 0.0357 1596 17 winrar Yes 19. 11 1820 0.035 0.0357 390 80 .acc No 20. 11 1820 0.035 0.0357 203 90 .ogg No 21. 11 1820 0.035 0.0357 728 61 .wma No 22. 11 1820 0.035 0.0357 392 79 .m4a No 23. 11 1820 0.035 0.0357 501 74 .ape No 24. 11 1820 0.035 0.0357 1220 33 .wv No 25. 11 1820 0.035 0.0357 529 72 .flac No 26. 11 1820 0.035 0.0357 39 98 .mmf No 27. 11 1820 0.035 0.0357 1596 17 winrar Yes 28. 16 1820 0.035 0.0357 390 80 .acc No 29. 16 1820 0.035 0.0357 203 90 .ogg No 30. 16 1820 0.035 0.0357 728 61 .wma No 31. 16 1820 0.035 0.0357 392 79 .m4a No 32. 16 1820 0.035 0.0357 501 74 .ape No 33. 16 1820 0.035 0.0357 1220 33 .wv No 34. 16 1820 0.035 0.0357 529 72 .flac No 35. 16 1820 0.035 0.0357 39 98 .mmf No 36. 16 1820 0.035 0.0357 1596 17 winrar Yes (Partially) 47 Penerbit Akademia Baru Journal of Advanced Research in Computing and Applications ISSN (online): 2462-1927 | Vol. 2, No. 1. Pages 39-54, 2016 1 2 3 4 LSB Figure 6: Stego Audio Size after Different Type of Compression The comparison of the cover audio and the decompressed stego audio from 17% compression which resulted in 83% size of the cover audio. This compression bowdlerized 320kB, which is 257,519 bps. The compression has expurgated most of the lost data. During the decompression process, the scheme will only able to retrieve those data which are strongly attached to the cover audio. So, the bits at higher LSB are safely located in their respective position and sequence with less distortion. In other words, the compression of the stego audio only holds the higher LSB able to sustain. This section presented the analysis of the proposed scheme to enhance security in Audio Steganography by using LSB and AES encryption. This proposed scheme prevents the intruders or third party to be suspicious about the existence of secret information in the cover audio. The cover audio has same audio performance with good quality and high PSNR which present no detection of audio distortion. A file format that used is .wav format which consists of large bytes compare to other bytes. The scheme shows the trade-off relation between the payload and the robustness of the stego audio. This scheme provides a solution to embed a secret message on other than LSB only where the capacity is traded off with the robustness. The higher LSB is robust, but capable to carry the little secret message. Enhancement of Security in Audio Steganography by Using Higher Least Significant Bits (LSB) and Advanced Encryption Standard (AES) provides a better data hiding specific in the robustness and the security. AES encryption and stego key have provided double layer security before accessible, partially to the stego audio. As in the higher LSB scheme, the secret message will be embedded at selective positions of higher LSB within the audio carrier where the positions to be determined during the encoding process. It can be considered as a better and efficient method for hiding the data. This proposed scheme is embedding secret bits in higher LSB and increases the robustness and withstands audio compression. AES encryption has increased the level of security before the secret information is being shifted to steganography environment where higher LSB embedding is embarked. This scheme is a combination of steganography and cryptography, which combines secured communication with protected of information. Therefore, there is a need for future study on the combination of steganography and cryptography for the robustness, capacity and quality with a greater data hiding. 48