Direct Spread Spectrum Technology For Data Hiding in Audio PDF

Article Direct Spread Spectrum Technology for Data Hiding in Audio Alexandr Kuznetsov ^1,2,^\*, Alexander Onikiychuk ^1^, Olga Peshkova ^1^ Tomasz Gancarczyk ^3^, Kornel Warwas ^3^, Ruslana Ziubina ^3,^\* ^1^ V. N. Karazin Kharkiv National University, Svobody sq., 4, Kharkiv, 61022, Ukraine, [[ku...

Article Direct Spread Spectrum Technology for Data Hiding in Audio Alexandr Kuznetsov ^1,2,^\*, Alexander Onikiychuk ^1^, Olga Peshkova ^1^ Tomasz Gancarczyk ^3^, Kornel Warwas ^3^, Ruslana Ziubina ^3,^\* ^1^ V. N. Karazin Kharkiv National University, Svobody sq., 4, Kharkiv, 61022, Ukraine, [[kuznetsov\@karazin.ua]](mailto:[email protected]) ^2^ JSC "Institute of Information Technologies", Bakulin St., 12, Kharkiv, 61166, Ukraine, [[kuznetsov\@karazin.ua]](mailto:[email protected]) ^3^ University of Bielsko-Biala, Willowa st., 2, Bielsko-Biala, 43309, Poland, (tgan-, kwarwas\@ath.bielsko.pl) **\*** Correspondence: [[kuznetsov\@karazin.ua]](mailto:[email protected]); [[rziubina\@ath.bielsko.pl]](mailto:[email protected]) **Abstract:** Direct spread spectrum technology is traditionally used in radio communication systems with multiple access, for example, in CDMA standards, in global satellite navigation systems, in Wi-Fi network wireless protocols, etc. It ensures high security and reliability of information transfer. In addition, spread spectrum technology provides the transmitted signals with a noise-like appearance, thus hiding the semantic content of the messages. We are researching this technology for other implementations. The purpose of our study is new technologies for hiding data in multimedia files. In particular, we are investigating direct spread spectrum in the context of the development of methods for hiding data in audio containers. We consider various spreading sequences (chip codes) and also explore their use for hiding information in audio files. We have conducted experimental studies and estimated the bit error rate (BER) in the recovered data. The article also evaluates the distortion of an audio-container by peak signal-to-noise ratio (PSNR). The results of our research enable us to find out which method of forming chip codes gives a lower BER with equal PSNR. We are providing recommendations on the formation of spreading sequences to reliably and safely hide informational messages in audio files. **Keywords:** Spread Spectrum Technology; Data Hiding; Steganography; Chip Codes; Audio +-----------------------------------------------------------------------+ | **Citation:** Kuznetsov, A.; Onikiychuk, A.; Peshkova, O.; | | Gancarczyk, T.; Warwas, K.; Ruslana Ziubina, R. Direct Spread | | Spectrum Technology for Data Hiding in Audio. *Sensors* **2022**, | | *22*, x. https://doi.org/10.3390/xxxxx | | | | Academic Editor: Firstname Lastname | | | | Received: date | | | | Accepted: date | | | | Published: date | | | | **Publisher's Note:** MDPI stays neutral with regard to | | jurisdictional claims in published maps and institutional | | affiliations. | | | | **Copyright: © 2022 by the authors. Submitted for possible open | | access publication under the terms and conditions of the Creative | | Commons Attribution (CC BY) license | | (https://creativecommons.org/licenses/by/4.0/).** | +-----------------------------------------------------------------------+ 1\. Introduction Various techniques are used to conceal information \[1--3\]. The most common are cryptography techniques \[4--6\]. In this case, meaningful messages are converted into meaningless, noise-like form. I. e. this becomes useless data for an unauthorized user (who does not know the secret cryptographic key). Thus, cryptography hides the semantic content of the transmitted data, the very existence of messages is not hidden \[3,7\]. Another approach is to use steganographic techniques \[3,8\]. In this case, informational messages are hidden inside other data (also called containers, covers, carriers, etc.). Carriers can be presented, for example, in the form of various multimedia files (images, video, audio, texts, etc.) \[9\]. There are many applications with multimedia data, the transfer of video or audio files on the Internet does not cause any suspicion. At the same time, only the one who has the secret steganographic key knows that an informational message is hidden in these covers. It is like a suitcase with a double bottom, and the second bottom is reliably disguised and it is difficult to detect a hidden message without a secret key. Thus, steganography hides the fact that messages exist \[2,8\]. The present paper discusses steganographic information hiding techniques, in which audio files act as multimedia containers. In our experiments, we used the simplest audio signals recorded in the digital WAV format \[10,11\]. However, all our results and suggestions can be interpreted for a more general case with other audio container formats. To conceal information, this work takes a special technology traditionally used in radio communication systems \[12--15\]. This is direct spreading of the spectrum \[16\], which implements wideband modulation with pseudo-random sequences (chip codes) \[17\]. Correlation properties of chip codes guarantee the absence of mutual interference in the communication channel \[18\]. As a result, the radio communication system acquires numerous useful properties: noise immunity, high bandwidth and subscriber capacity, absence of mutual interference, environmental friendliness of communication, and many others \[19\]. Direct spread spectrum technology can also be used in steganography \[20--22\]. In particular, in one of the first works \[21\] in this area, it was proposed to interpret the carrier as noise in the communication channel. In this case, the task of hiding information in multimedia files is equivalent to transmitting useful signals over communication channels with natural additive noise. Today, a large number of studies on steganography and direct spreading of the spectrum have already been conducted. Various researchers have studied this issue in relation to various computer applications. It turns out that the main characteristics of reliability and safety depend on the properties of the spreading sequences. In particular, in \[23,24\], we investigated various ways of generating chip codes in relation to cover images. In this paper, we continue these studies in relation to another type of multimedia data - to audio carriers. We study the basic characteristics of an audio stegosystem for various spreading sequences. In our experiments, we estimate the bit error rate (BER) of hidden information data. This is the most important indicator of the quality of recovered messages. In addition, we evaluate the distortions of the audio container, which characterize the quality of multimedia data after hiding information messages in them. To do this, we use the mean squared error (MSE) and the peak signal-to-noise ratio (PSNR). In this article, we demonstrate that the variety of ways of generating chip codes have different results of BER, MSE, and PSNR. It becomes an area for possible optimization, as the choice of the method of forming the spreading sequences is very important for various computer applications. Thus, the main goal of our work is to find the best chip codes that would minimize the BER for comparable PSNR. 3\. Related works Hiding data in multimedia files using direct spread spectrum has been studied in many related papers. The first works \[20,21,25\] were devoted to the substantiation of the general concepts and structure of the steganosystem using direct spread spectrum. Images were used to hide the messages. Already in those articles, the authors note several conflicting requirements \[26\]. In particular, increasing the size of messages leads to the increasing the corruption of container; small BER values can be achieved only with a large \"energy\" of the embedded message, which significantly distorts the container. To eliminate some of these contradictions, the authors suggested using special filtering and correcting codes \[26\]. These and other ideas were developed in subsequent works. For example, in \[27\] correction codes were investigated, in \[28,29\] a discrete cosine transform (DCT) was implemented together with direct spreading of the spectrum, etc. Papers \[30,31\] are devoted to the study of chaos. In works \[32,33\] hiding information in video data was studied, and in \[27,34\] audio signals were used for hiding information. Indeed, hiding messages in audio signals is a promising direction \[35\], these methods can be used, for example, to protect the property rights of digital audio signals \[36\]. Note that in these numerous works, the properties of spreading sequences (chip codes) were almost not explored. The exceptions are our two recent works \[23,24\], in which we analyze various methods of forming spreading sequences, as well as their influence on the characteristics of Spread-Spectrum Image Steganography. In this work, we continue our research and expand their field to audio files. First, we look at different ways to form spreading sequences and explore BER, MSE and PSNR. Afterwards, we explore possible compromises, such as how to minimize carrier distortion with low bit error rates. It turns out that the Walsh-Hadamard expansion sequences are best suited for this, and our experiments clearly demonstrate this. 3\. Materials and Methods Direct spread spectrum technology is a wideband modulation in which the original bit sequence is converted into a pseudo-random spreading sequence \[18\]. As a result, we have the transmission of a significantly larger number of extended (direct) sequence bits in the same time interval. Consequently, the frequency range for transmitting signal increases in proportion to the length of the spreading sequence \[19\]. 3.1. Direct spread spectrum technology in telecommunications The fundamental Shannon-Hartley theorem sets a constraint on the channel capacity, i.e. to the upper limit of the maximum amount of information that can be transmitted in a communication system with a given frequency band and power of additive white Gaussian noise \[37\]: -- ----- (1) -- ----- - -- channel throughput, bit / s; - -- frequency band, Hz; - -- desired signal power; - -- additive white Gaussian noise power; - ­-- signal power to noise ratio (SNR). From formula (1) we see that at low SNRs, the throughput can be increased only by expanding the frequency band, and direct spread spectrum technology is well suited for this. Indeed, using very long spreading sequences, even at a very low SNR value, high-speed information transmission can be realized. This circumstance can also be used, e.g. to build environmentally friendliness of communication systems, i.e. when. Then, e.g. using spreading sequences with a bit length or more, it is possible to realize information transfer at rates of tens and hundreds of Mb / s. Let us explain the technology of spreading the spectrum with direct sequence in the following simple way. Consider a set of pseudo-random bit sequences (vectors): , ,..., , Vectors are designed to spread the spectrum of the original message, so they are called chip codes, or chipping codes. Each value is called a chip or sequence element. The vectors are formed so that their cross-correlation is negligible, i.e. --- -----. (2) --- ----- For example, if the vectors are taken equal to the rows (or columns) of the Hadamard matrix, we obtain orthogonal chip codes, i.e.. There are other ways to generate chip codes, for example, when the elements take a random value over an interval. However, condition (2) for the generation of spreading sequences is decisive. Suppose the message is made up of bits, i.e.. E.g. let us write the information bits also in polar form, i.e. Modulation of the information bits can be achieved in different ways, e.g. with polar notation according to the expression: --- -----. (3) --- ----- In this way, instead of one binary element a vector (sequence) of binary elements will be transmitted. Let us assume that the durations of an element and its extended chip version are the same. Suppose, for example, they are equal to seconds. Then the duration of one chip will be times less:. Therefore, the chips are transmitted at a frequency Hertz. This frequency is times greater than the frequency of information bits. Thus, we have obtained the spread of the frequency of the original message by times. The multiplying of the signal duration by its frequency is called the base. If then the signals are called complex. For our case, we have , Several information bits can be simultaneously transmitted to the communication channel, i.e. the receiver gets an additive mixture: --- ----- , (4) --- ----- The receiver has synchronized copies of the chip codes. To restore the information bits, it is necessary to calculate the correlation between the received signal and the corresponding chip codes. For example, to restore the value in (3), the receiver computes. Formula (2) is linear, i.e. write down --- -----. (5) --- ----- For random noise we have: --- -----. (6) --- ----- Moreover, , therefore --- -----. (7) --- ----- In this way, we can write the information bit recovery rule --- -----. (8) --- ----- Note that in the above reasoning, individual bits may belong to different senders. In this case, code division multiple access is implemented \[18\]. Let us note the following advantages of using direct spread spectrum technology: - Resistance to unintended or intended jamming; - Sharing a single channel among multiple users; - Reduced signal/background-noise level hampers interception; - Determination of relative timing between transmitter and receiver Direct spread spectrum technology is implemented in many industry standards and applications, for example: - Code division multiple access communication systems (IS-95, CDMA2000, WCDMA, DS-CDMA, TD-CDMA, TD-SCDMA etc.); - Global satellite navigation systems (American GPS, European Galileo and Russian GLONASS); - family of wireless network protocols Wi-Fi (family of standards IEEE 802.11); We are considering the use of this technology in steganography, i.e. to hide informational messages in multimedia files. 3.2. Direct spread spectrum in steganography Direct spread spectrum technology has long been used in steganography. In the first works \[20,21,25\] a general scheme of a steganosystem with direct spreading of the spectrum was proposed. The authors used cover images, but all their reasoning can be extended to other multimedia data. The general idea of such a stegosystem is as follows. Multimedia data (images, audio, video, etc.) is interpreted as noise in the communication channel. Using our notation, this is in (4). The information message is presented in polar form and multiplied bit by bit by the spreading sequence as in (3). Further, this data is hidden in the cover data according to the rule (4). However, we currently denote: - \- multimedia data (container, cover), inside which the message is hidden; - \- modified multimedia data (filled container), after hiding messages in them. For information recovery, formula (5) is also used. If assumptions (6) and (7) are true, then the value of the recovered bit is calculated by the formula (8). It should be noted that even in the first works, the authors encountered significant difficulties. For example, in \[38\] was shown that the error rate in the extracted messages is very high (see, for example, Table 2 on page 12 of \[38\]). BER can be reduced in different ways, for example, using filtering, correction codes, etc. The simplest is to increase the power of chip codes. For this, expression (4) can be rewritten as --- -----. (9) --- ----- By increasing the power it is actually possible to achieve a decrease in BER. However, this inevitably leads to distortion of the container. For example, for and each value of the cover file may be skewed by. This may present a real problem because BER is still high. For example, in \[38\] it is shown that even with an error rate is more than 10% (see Table 2 from \[38\]). We have carried out detailed studies of this problem in \[23,24\]. In particular, we repeated the experiments with images from \[21,25,38,39\]. We studied the effect of different ways of forming spreading sequences on the BER value. The general conclusions about the high BER were confirmed and the explanation for this phenomenon is, in our opinion, the following. The first term in (4) presents certain implementation of random additive white noise. This is indeed the case for spread spectrum communications systems. Those assumption (6) is fulfilled, which gives an almost error-free recovery of the message according to rule (8). However, in the stagnosystem, in (4) multimedia data is meant. This is usually highly redundant, highly correlated data. This is not a random realization of Gaussian noise, and assumption (6) is often not met. Obviously, this leads to errors in the recovered bits according to rule (8). In works \[40,41\] an effective way to combat this phenomenon was proposed. We proposed to form chip sequences in a special, adaptive way. Indeed, if we take into account the statistical properties of multimedia data (in \[40,41\] we used cover images), then it is possible to form chip codes , in such a way that condition (6) is guaranteed to be satisfied. In this case, it is possible to ensure almost error-free extraction of information messages according to rule (8). In this article, we continue our experiments. We are now using audio covers and experimenting with different ways to generate chip codes. We show that the variety of ways of generating spreading sequences produce different BER values. This is an area for possible optimization of the stegosystem and we show which chip codes are really best used to hide information messages in audio containers. 3.3. Initial Data for Hiding Messages We used various audio carriers and various families of chip codes as the initial data for our experiments. We were guided by the works \[21,25\] and our papers \[23,24\], when choosing the initial data and research methodology. An audio signal recorded in the digital WAV format \[10,11\] was used as an audio container. In our experiments, we used an audio file with a student anthem (Gaudeamus Igitur) downloaded from an Internet resource \[42\] and converted to WAV format. WAV audio is most commonly encoded using linear pulse code modulation (LPCM). Such a cover file contains uncompressed audio signal data; its main characteristics are: - the number of channels (streams) of audio data. We used stereo signals, i.e.; - sampling frequency. Our audio signals from ; - the number of bits allocated to encode each discrete sample of the audio signal. We used covers from bit. As input information messages, we used plain text files (steganography textbooks), processed and presented in binary polar form. To hide information, various spreading sequences were used, studied by many authors, for example, in \[21,25\] and also in our previous works \[23,24\]. Let us consider these chip codes in more detail. 3.4. Used chip codes In this article, various methods for generating chip codes were considered and their influence on the characteristics of a steganosystem was studied. Specifically, we explored five different ways to generate spreading sequences. ***1. Spread sequences from*** \[21,25\] In the basic works \[21,25\] on Spread-Spectrum Steganography, a nonlinear rule for the formation of chip codes was used. In particular, each chip was formed as a realization of a random variable distributed according to the standard normal law. For this, the rule was proposed: --- ------. (10) --- ------ - \- implementation of a random variable uniformly distributed over the interval ; - \- inverse cumulative distribution function for a standard Gaussian random variable. In our experiments, we used chip codes of length generated according to rule (10). Spreading sequences with a Gaussian chip distribution (10) were also studied by us earlier in relation to cover images in \[23,24\]. We are expanding our research to the case of audio containers now. ***2. Chip codes from random numbers uniformly distributed over the interval \[-1,1\]*** We have also implemented the formation of spreading sequences of length from random numbers uniformly distributed over the interval \[-1,1\]. Each chip is formed randomly with equal probability and independently of other values. ***3. Chip codes from normally distributed random numbers*** This method of forming spreading sequences of length consists in generating normally distributed random numbers. In fact, the second and third methods are very similar. Each chip is randomly formed. The second method uses a uniform distribution on the interval \[-1,1\], the third one uses a standard normal distribution. The third method is also similar to the first, but the rule for generating chip codes is somewhat different. ***4. Binary chip codes generated by a pseudo-random bit generator*** The simplest, in our opinion, is the formation of individual chips using a pseudo-random bit generator. We converted the generated bits to polar form and formed spreading sequences of length. ***5. Walsh-Hadamard spreading sequences*** These are discrete sequences of length , , which can be formed as rows (or columns) of the Hadamard matrix of order : ----- ------ ,. (11) ----- ------ For example, for we have four chip codes: , , ,. In our experiments, we used spreading sequences with a length of. To do this, we recursively generated according to rule (11) and formed as a set of strings. The Walsh-Hadamard sequences form an orthogonal system, i.e. , Nevertheless, the main disadvantage of such chip codes should be pointed out. The number of different orthogonal sequences of length cannot exceed. That is, to hide a large number of bits in the same container, this generation method may not be sufficient. 3.5. Indicators of Effectiveness of Hiding Messages To evaluate the efficiency of the steganosystem, it is necessary to evaluate the distortions in the recovered messages, as well as the distortions of the original container. To do this, we experimentally estimate the bit error rate (BER) in recovered messages. We also estimate audio cover distortions by mean squared error (MSE) and Peak signal-to-noise ratio (PSNR). These are the most common metrics used in many related works. ***Bit Error Rate, BER*** The bit error rate is calculated using the formula: --- ------ , (12) --- ------ ***Mean squared error, MSE*** To estimate the container distortion, the root mean square error (MSE) is used, i.e. the averaged square of the difference between the original multimedia data and the cover data obtained after the message was hidden. For an audio signal represented by a set of discrete samples, MSE is calculated by the formula: --- ------. (13) --- ------ ***Peak signal-to-noise ratio, PSNR*** A more visual characteristic of cover distortion is the ratio of the maximum possible signal power to the power of the distorting noise. Usually this characteristic is expressed on a logarithmic scale and calculated by the formula: -- ------ (14) -- ------ In our case, by we mean multimedia data, i.e. an audio signal represented by a set of discrete samples. If each discrete sample is coded in bits, then (in linear pulse code modulation, PCM) the maximum possible value of. In our experiments, we used the simplest audio signals with i.e.. Formula (14) for this value will take the form:. 4\. Results 4.1. Experimental Evaluation of Hidden Message Distortion In fig. 1-5 the results of our experiments on BER estimation are shown for different methods of generating chip codes of length. We changed the number of hidden information bits and the coefficient for amplifying the power of the spreading signals in (9). The obtained results of experimental studies correspond to the points in the given diagrams. Each point corresponds to a value averaged over 5000 tests. For better perception, these and all subsequent figures also show trend lines. **Figure 1.** dependencies for different values of , rule for generating chip codes №1 **Figure 2.** dependencies for different values of , rule for generating chip codes №2 **Figure 3.** dependencies for different values of , rule for generating chip codes №3 **Figure 4.** dependencies for different values of , rule for generating chip codes №4 **Figure 5.** dependencies for different values of , rule for generating chip codes №5 The analysis of the results obtained (see Fig. 1-5) enables us to make the following observations. First, an increase of the coefficient makes it possible to achieve low BER values for any of the considered classes of chip codes. At the same time, chip codes with generation rules №. 1,2 and 3,4 give similar BER values, i.e. for this indicator, such spreading sequences are equivalent to each other. Walsh-Hadamard sequences give significantly better results in terms of error rates. Even at low values of , these chip codes provide low BER. The second observation is that the BER increases as the number of bits in the hidden message increases. However, different chip codes give different results. For example, spreading sequences №1 and №2 at provide only at. For chip codes №3 and №4, the same is provided with a lower power gain, for example, for. The Walsh-Hadamard sequences even with and provide. To compare the BER dependences corresponding to different methods of forming chip codes, Fig. 6 shows diagrams. These are averaged over 5000 test results for bits. **Figure 6.** dependencies for different rules for generating chip codes, The analysis of the results shown in Fig. 6 demonstrates that the techniques for generating chip codes №1 and №2, as well as №3 and №4 do indeed give similar BER values. However, such a difference in the BER value is observed only for large values. With a small number of hidden bits, the BER values for chip codes №1-4 are practically the same. This is clearly seen in Fig. 7. Thus, for small values of , it is really possible to achieve a low error rate for practically any method of forming chip codes. **Figure 7.** dependencies for different rules for generating chip codes, Walsh-Hadamard sequences show significantly better BER results. For this family of chip codes, the error rate, other things being equal, is one to two orders of magnitude lower. At the same time, BER can be significantly reduced only for large values, and this will inevitably lead to container distortion. 4.2. Experimental evaluation of container distortion We used MSE and PSNR to assess container distortion. The obtained experimental results indicate that using chip codes generated according to rules 1, 3, 4 and 5 leads to approximately equal distortions of containers. The corresponding dependences of and for different values are shown in Fig. 8.9. Somewhat less distortion of the container is provided by using the rule for generating chip codes № 2, the corresponding dependencies are shown in Fig. 10.11. **Figure 8.** dependencies for different values of , rules for generating chip codes №1, 3, 4 and 5 **Figure 9.** dependencies for different values of , rules for generating chip codes №1, 3, 4 and 5 **Figure 10.** dependencies for different values of , rule for generating chip codes №2 **Figure 11.** dependencies for different values of , rule for generating chip codes №2 The analysis of Fig. 8-11 enables us to make the following observations. First, an increasing in the number of hidden information bits leads to an inevitable increasing in container distortion. Usually, PSNR values of about 30..40 dB are considered acceptable quality. If you focus on these values, then you can effectively hide no more than 10..20 information bits. Secondly, an increase in the power gain also leads to an increase in container distortion. If we focus on PSNR of about 30..40 dB, then the values of are acceptable (with ). Thus, the obtained results of experimental studies show that by increasing the length of the hidden message, we inevitably distort the container. Raising in order to reduce BER also leads to an increase in container distortion. That is, there are several conflicting factors that directly affect the efficiency of the steganosystem. To find a compromise, it is necessary to study the influence of these factors on each other. 4.3. Correlation of error rate and container distortion, finding a compromise To find a compromise solution, let us study the relationship between BER and PSNR. In fig. 12-16 the dependences obtained as a result of averaging over 5000 tests are shown. **Figure 12.** dependencies for different values , rule for generating chip codes №1 **Figure 13.** dependencies for different values , rule for generating chip codes №2 **Figure 14.** dependencies for different values , rule for generating chip codes №3 **Figure 15.** dependencies for different values , rule for generating chip codes №4 **Figure 16.** dependencies for different values , rule for generating chip codes №5 5\. Discussion The analysis of the data shown in Fig. 1-16 results enables us to draw the following conclusions. First, it is difficult to implement information hiding without significant container distortion. The above dependences of clearly demonstrate this. By increasing the volume of the hidden message and / or the power amplification factor , we will certainly distort the container, i.e. decrease PSNR. For example, chip codes №1 and №2 allow hiding only 1-2 bits of the message when and. Spreading sequences №3 and №4 are somewhat better, they make it possible to hide 1-2 message bits when and. However, this is still not enough. Finally, the Walsh-Hadamard signals do perform well. If they are used, it is possible to hide 16 or more message bits and provide and. This confirms the earlier conclusion about the preference of such chip codes for hiding information. This is also consistent with our findings from Spread-Spectrum Image Steganography \[23,24\]. The second conclusion is a possible trade-off between the amount of hidden data and the introduced distortion of the audio container. For example, as follows from Fig. 16, halving the value leads to an increase in PSNR by about 5 dB. This can be used to configure the desired stegosystem parameters in a specific practical application. The third and probably the most important conclusion is that it is almost impossible to achieve high volumes of hidden information with the known techniques. For example, even for the best case shown in Fig. 16, with and , only bits of information can be hidden. It is possible to improve these values by traditional techniques only by increasing the length of the chip codes, which inevitably leads to an increase in the computational complexity of the transformations. Another way is implementation of adaptive techniques, for example, as in \[40,41\], or the using of new, advanced methods of hiding information, for example, based on addressing chip codes \[43\]. 6\. Conclusion In this article, we explored techniques for hiding information in audio containers using direct spread spectrum. Various spreading sequences (chip codes) were considered and their influence on the error rate in recovered messages was studied. We also investigated the distortion of the audio container in terms of MSE and PSNR. The results clearly demonstrate the promising characteristics of this direction. In each segment of the container, we managed to reliably hide a part of the information message. At the same time, the BER, MSE and PSNR values are within an acceptable range. The best characteristics were shown by Walsh-Hadamard chip codes, which showed the lowest the error rate (with comparable container distortions). At the same time, it should be noted that there exist objectively conflicting factors that negatively affect one another. For example, by increasing the volume of hidden messages, we inevitably increase the BER and distort the container. By increasing the power of the chip codes, it is possible to reduce the BER. However, this distorts the container even more. We managed to show the availability of compromise solutions. This is especially true for Walsh-Hadamard chip codes, for which the range of possible solutions is much wider than for other spreading sequences. Nevertheless, it should be noted that when using traditional data hiding techniques, the volume of hidden messages cannot be large. The search for new ways of hiding information is promising e.g. based on the adaptive formation of spreading sequences \[41\], advanced techniques for addressing chip codes \[43\], etc. **Supplementary Materials:** https://github.com/AlexanderOnikiychuk/spread-spectrum-technology-researches. **Author Contributions:** Conceptualization and methodology, A. K.; software and validation, A. O., T.G.; writing---review and editing, O. P., R.Z.; visualization, K.W.. All authors have read and agreed to the published version of the manuscript. **Funding:** The research work reported in this paper was in part supported by the National Centre for Research and Development, Poland under the project no. POIR.04.01.04-00-0048/20. **Institutional Review Board Statement:** Not applicable. **Informed Consent Statement:** Not applicable. **Data Availability Statement:** Not applicable. **Acknowledgments:** Not applicable. **Conflicts of Interest:** The authors declare no conflict of interest. References 1\. Li, R.; Xu, S.; Rong, B.; Yang, H. Host Cancelation-Based Spread Spectrum Watermarking for Audio Anti-Piracy over Internet. *Security and Communication Networks* **2016**, *9*, 4691--4702, doi:10.1002/sec.1673.2. Manoj, I.V.S. Cryptography and Steganography. *IJCA* **2010**, *1*, 63--68, doi:10.5120/257-414.3. Moulin, P.; O'Sullivan, J.A. Information-Theoretic Analysis of Information Hiding. *IEEE Transactions on Information Theory* **2003**, *49*, 563--593, doi:10.1109/TIT.2002.808134.4. Aroua, S.; Champagnat, R.; Coustaty, M.; Falquet, G.; Ghadfi, S.; Ghamri-Doudane, Y.; Gomez-Kramer, P.; Howells, G.; McDonald-Maier, K.D.; Murphy, J.; et al. Security and PrIvacy foR the Internet of Things: An Overview of the Project. In Proceedings of the 2019 IEEE International Conference on Systems, Man and Cybernetics (SMC); October 2019; pp. 3993--3998.5. Fazzat, A.; Khatoun, R.; Labiod, H.; Dubois, R. A Comparative Performance Study of Cryptographic Algorithms for Connected Vehicles. In Proceedings of the 2020 4th Cyber Security in Networking Conference (CSNet); October 2020; pp. 1--8.6. Levi, I.; Rudin, Y.; Fish, A.; Keren, O. Embedded Randomness and Data Dependencies Design Paradigm: Advantages and Challenges. In Proceedings of the 2018 Design, Automation Test in Europe Conference Exhibition (DATE); March 2018; pp. 395--400.7. Menezes, A.J.; Oorschot, P.C. van; Vanstone, S.A.; Oorschot, P.C. van; Vanstone, S.A. *Handbook of Applied Cryptography*; CRC Press, 2018; ISBN 978-0-429-46633-5.8. Yahya, A. Introduction to Steganography. In *Steganography Techniques for Digital Images*; Yahya, A., Ed.; Springer International Publishing: Cham, 2019; pp. 1--7 ISBN 978-3-319-78597-4.9. Yahya, A. Steganography Techniques. In *Steganography Techniques for Digital Images*; Yahya, A., Ed.; Springer International Publishing: Cham, 2019; pp. 9--42 ISBN 978-3-319-78597-4.10. WAV File Extension - What Is It? How to Open a WAV File? Available online: https://filext.com/file-extension/WAV (accessed on 25 April 2021).11. Fleischman \, E. WAVE and AVI Codec Registries Available online: https://tools.ietf.org/html/rfc2361 (accessed on 25 April 2021).12. Song, T.; Zhou, K.; Li, T. CDMA System Design and Capacity Analysis Under Disguised Jamming. *IEEE Transactions on Information Forensics and Security* **2016**, *11*, 2487--2498, doi:10.1109/TIFS.2016.2585089.13. Wang, X.; Liu, X.; Chen, H.-H.; Meng, W. Complementary Coded CDMA Systems With CP-Free OFDM. *IEEE Transactions on Vehicular Technology* **2020**, *69*, 11515--11528, doi:10.1109/TVT.2020.3013155.14. Sedaghat, M.A.; Müller, R.R.; Marvasti, F. On Optimum Asymptotic Multiuser Efficiency of Randomly Spread CDMA. *IEEE Transactions on Information Theory* **2015**, *61*, 6635--6642, doi:10.1109/TIT.2015.2482483.15. Khalife, J.; Kassas, Z.M. Navigation With Cellular CDMA Signals---Part II: Performance Analysis and Experimental Results. *IEEE Transactions on Signal Processing* **2018**, *66*, 2204--2218, doi:10.1109/TSP.2018.2799166.16. Stüber, G.L. Spread Spectrum Techniques. In *Principles of Mobile Communication*; Stüber, G.L., Ed.; Springer International Publishing: Cham, 2017; pp. 449--499 ISBN 978-3-319-55615-4.17. Torrieri, D. Chapter 2 Direct-Sequence Systems. In *Principles of Spread-Spectrum Communication Systems*; Torrieri, D., Ed.; Springer International Publishing: Cham, 2015; pp. 79--145 ISBN 978-3-319-14096-4.18. Ipatov, V.P. *Spread Spectrum and CDMA: Principles and Applications*; John Wiley & Sons, Ltd: Chichester, UK, 2005; ISBN 978-0-470-09180-7.19. Torrieri, D. *Principles of Spread-Spectrum Communication Systems*; Springer International Publishing, 2018;20. Smith, J.R.; Comiskey, B.O. Modulation and Information Hiding in Images. In Proceedings of the Information Hiding; Anderson, R., Ed.; Springer: Berlin, Heidelberg, 1996; pp. 207--226.21. Marvel, L.M.; Boncelet, C.G.; Retter, C.T. *Methodology of Spread-Spectrum Image Steganography*; ARMY RESEARCH LAB ABERDEEN PROVING GROUND MD, 1998;22. Wang, Y.-G.; Zhu, G.; Kwong, S.; Shi, Y.-Q. A Study on the Security Levels of Spread-Spectrum Embedding Schemes in the WOA Framework. *IEEE Transactions on Cybernetics* **2018**, *48*, 2307--2320, doi:10.1109/TCYB.2017.2735989.23. Kuznetsov, A.; Smirnov, O.; Arischenko, A.; Chepurko, I.; Onikiychuk, A.; Kuznetsova, T. Pseudorandom Sequences for Spread Spectrum Image Steganography. In Proceedings of the Proceedings of the International Workshop on Cyber Hygiene (CybHyg-2019) co-located with 1st International Conference on Cyber Hygiene and Conflict Management in Global Information Networks (CyberConf 2019), Kyiv, Ukraine, November 30, 2019.; CEUR-WS.org, 2020; Vol. 2654, pp. 122--131.24. Kuznetsov, A.; Smirnov, A.; Gorbacheva, L.; Babenko, V. Hiding Data in Cover Images Using a Pseudo-Random Sequences. In Proceedings of the Proceedings of The Third International Workshop on Computer Modeling and Intelligent Systems (CMIS-2020), Zaporizhzhia, Ukraine, April 27-May 1, 2020; Subbotin, S., Ed.; CEUR-WS.org, 2020; Vol. 2608, pp. 646--660.25. Marvel, L.M.; Boncelet, C.G.; Retter, C.T. Spread Spectrum Image Steganography. *IEEE Transactions on Image Processing* **1999**, *8*, 1075--1083, doi:10.1109/83.777088.26. Marvel, L.M. *Image Steganography for Hidden Communication*; ARMY RESEARCH LAB ABERDEEN PROVING GROUND MD, 2000;27. Youail, R.S.; Samawi, V.W.; Kadhim, A.-K.A.-R. Combining a Spread Spectrum Technique with Error-Correction Code to Design an Immune Stegosystem. In Proceedings of the Security and Identification 2008 2nd International Conference on Anti-counterfeiting; August 2008; pp. 245--248.28. Agrawal, N.; Gupta, A. DCT Domain Message Embedding in Spread-Spectrum Steganography System. In Proceedings of the 2009 Data Compression Conference; March 2009; pp. 433--433.29. Ling Lu; Xinde Sun; Leiting Cai A Robust Image Watermarking Based on DCT by Arnold Transform and Spread Spectrum. In Proceedings of the 2010 3rd International Conference on Advanced Computer Theory and Engineering(ICACTE); August 2010; Vol. 1, pp. V1-198-V1-201.30. Satish, K.; Jayakar, T.; Tobin, C.; Madhavi, K.; Murali, K. Chaos Based Spread Spectrum Image Steganography. *IEEE Transactions on Consumer Electronics* **2004**, *50*, 587--590, doi:10.1109/TCE.2004.1309431.31. Kumar, A.; Albagul, A.; Ghose, M.; Negrat, K. Improved Chaos Based Spread Spectrum Image Steganography. *Proceedings of the 10th IASTED International Conference August* **2008**, 18--20.32. Weihua, X.; Yongbing, W.; Shuiyuan, Y. H.264 Video Watermark Algorithm Using DCT Spread Spectrum. In Proceedings of the 2015 3rd International Conference on Applied Computing and Information Technology/2nd International Conference on Computational Science and Intelligence; July 2015; pp. 447--450.33. Zarmehi, N.; Akhaee, M.A. Video Steganalysis of Multiplicative Spread Spectrum Steganography. In Proceedings of the 2014 22nd European Signal Processing Conference (EUSIPCO); September 2014; pp. 2440--2444.34. Nugraha, R.M. Implementation of Direct Sequence Spread Spectrum Steganography on Audio Data. In Proceedings of the Proceedings of the 2011 International Conference on Electrical Engineering and Informatics; July 2011; pp. 1--6.35. Li, X.; Yu, H.H. Transparent and Robust Audio Data Hiding in Subband Domain. In Proceedings of the Proceedings International Conference on Information Technology: Coding and Computing (Cat. No.PR00540); March 2000; pp. 74--79.36. Hernandez-Garay, S.; Vazquez-Medina, R.; Nino de Rivera, L.; Ponomaryov, V. Steganographic Communication Channel Using Audio Signals. In Proceedings of the 2008 12th International Conference on Mathematical Methods in Electromagnetic Theory; June 2008; pp. 427--429.37. Sklar, B. *Digital Communications: Fundamentals and Applications*; Edición: 2.; Prentice Hall: Upper Saddle River, NJ, 2017; ISBN 978-0-13-472405-8.38. Brundick, F.S.; Marvel, L.M. *Implementation of Spread Spectrum Image Steganography:*; Defense Technical Information Center: Fort Belvoir, VA, 2001;39. Boncelet, J.C.G.; Marvel, L.M.; Retter, C.T. Spread Spectrum Image Steganography \| Unified Patents 2003.40. Kuznetsov, A.; Smirnov, O.; Kovalchuk, D.; Kuznetsova, T. New Technique for Data Hiding in Cover Images Using Adaptively Generated Pseudorandom Sequences. In Proceedings of the Proceedings of the International Workshop on Cyber Hygiene (CybHyg-2019) co-located with 1st International Conference on Cyber Hygiene and Conflict Management in Global Information Networks (CyberConf 2019), Kyiv, Ukraine, November 30, 2019.; CEUR-WS.org, 2020; Vol. 2654, pp. 1--14.41. Kuznetsov, A.; Smirnov, O.; Onikiychuk, A.; Makushenko, T.; Anisimova, O.; Arischenko, A. Adaptive Pseudo-Random Sequence Generation for Spread Spectrum Image Steganography. In Proceedings of the 2020 IEEE 11th International Conference on Dependable Systems, Services and Technologies (DESSERT); May 2020; pp. 161--165.42. Gaudeamus Igitur Available online: http://graecolatini.bsu.by/texts/gaudeamus/gaudeamus-data/gaudeamus03.mp3 (accessed on 7 May 2021).43. Kuznetsov, A.; Kiian, A.; Kuznetsova, K.; Smirnov, A. Data Hiding Scheme Based on Spread Sequence Addressing. In Proceedings of the Proceedings of the 1st International Workshop on Computational & Information Technologies for Risk-Informed Systems (CITRisk 2020) co-located with XX International scientific and technical conference on Information Technologies in Education and Management (ІТЕМ 2020), Kherson, Ukraine, October 15-16, 2020.; 2020; pp. 44--58.