Direct Sequence Spread Spectrum-Based Radio Steganography

Full text

Direct Sequence Spread Spectrum-based
Radio Steganography
Mateusz Wróbel
Institute of Communications Systems,
Faculty of Electronics,
Military University of Technology
Warsaw, Poland
mateusz.wrobel@wat.edu.pl
ORCID: 0000-0001-9099-7067
Zbigniew Piotrowski
Institute of Communications Systems,
Faculty of Electronics,
Military University of Technology
Warsaw, Poland
zbigniew.piotrowski@wat.edu.pl
ORCID: 0000-0003-3556-0297
Jan M. Kelner
Institute of Communications Systems,
Faculty of Electronics,
Military University of Technology
Warsaw, Poland
jan.kelner@wat.edu.pl
ORCID: 0000-0002-3902-0784
Abstract—Direct sequence spread spectrum (DSSS) is a
spread-spectrum modulation technique primarily used to
reduce overall signal interference. DSSS is one of the so-called
secondary modulation techniques, which makes the transmitted
signal wider in bandwidth than the information bandwidth. On
the other hand, this technique allows the transmitted
information signal to be hidden in the noise. Thus, DSSS signals
are classified as low probability of intercept/low probability of
detection (LPI/LPD). DSSS modulation is also used in
information hiding for both radio and audio signals. In this case,
the DSSS-based steganography algorithms used various
information-hiding mechanisms, e.g., modulation parameter
modification. In the novel solution proposed in this paper, we
use DSSS with variable processing gain for keying additional
information. In this case, the hidden message is transmitted in a
switching order of spreading sequences taken from a mapping
dictionary known only to the notified parties.
Keywords—radio communications, data hiding, information
hiding, steganography, direct sequence spread spectrum (DSSS).
I. INTRODUCTION
For military wireless communications, new waveform
types are being developed to ensure secure and reliable
transmission of voice, data, etc. According to the basic
approach adopted in electronics, acoustics, and related fields,
the signal waveform is the shape of its graph versus time,
regardless of its time and amplitude scales and any shift in
time [1]. In telecommunications, the waveform has a wider
meaning and can cover various aspects of the transmitted
signal, including modulation, coding, and the structure of
frames/packets.
As mentioned above, the creation of military waveforms
was intended to ensure secure communications. In this way, a
signal class was created that is characterized by a low
probability of intercept/low probability of detection
(LPI/LPD). LPI/LPD includes, i.a., frequency hopping spread
spectrum (FHSS) and direct sequence spread spectrum
(DSSS) [2]. FHSS and DSSS techniques are treated as so-
called secondary modulations of the signal which cause the
transmitted signal bandwidth is wider than the information
signal bandwidth.
DSSS is used, e.g., in global navigation satellite systems
(GNSS), IEEE 802.11b, g, and IEEE 802.15.4 standards, and
radio systems for control unmanned platforms.
Due to the LPI/LPD properties of DSSS signals, they are
difficult to detect by hiding the signal in the noise. So, this
technique is ideally suited for steganographic applications [3],
[4]. Data hiding in DSSS signals is difficult to detect because
these signals are difficult to detect. DSSS is used in
steganography and watermarking in relation to audio and
communication (i.e., wireless and wired) signals. A typical
approach to data hiding assumes hiding information in the
parameters or protocols of the transmitted DSSS signals as
covers. In others, the DSSS technique is used redundantly to
hide information in other types of broadcasts or audio signals.
The aim of this paper is to present the idea of a novel
steganography method consisting of hiding data in variable
processing gain of DSSS transmission. We also illustrate
simulation results for embedding and extracting hidden
information. In the extension of the paper, we will present the
effectiveness of extracting the developed steganography
method. In the next step, we want to implement this technique
using the Universal Software Radio Peripheral (USRP)
software-defined radio (SDR) [5].
The rest of the paper is organized as follows. In Section II,
a brief overview of data hiding solutions related to the DSSS
technique is presented. Section III describes the idea of the
new steganography technique in the DSSS signal by keying
spreading sequences. Embedding and extracting hidden data
based on simulation studies is shown in Section IV. Finally,
we present a summary.
II. DSSS DATA HIDING
Steganography is the practice of hiding secret information
within seemingly innocuous carriers to ensure covert
communication [4], [6]. Wireless steganography focuses on
concealing information within radio and acoustic signals,
exploiting the vast spectrum of frequencies available for
communication. In radio steganography, data can be
concealed within the modulation parameters or noise levels of
radio signals [6]–[10], while acoustic steganography involves
embedding information in audio signals [11]. On the other
hand, network steganography deals with concealing data
within network protocols, such as TCP/IP packets, exploiting
various techniques like protocol manipulation, timing
variations, or payload modification [12]. These forms of
steganography play a crucial role in ensuring covert
communication channels, enabling secure transmission and
exchange of sensitive information in a digital age.
DSSS is one of the telecommunication technologies used
also in steganography in addition to its original functionality,
which is used to transmit unclassified (i.e., cover) data. A
survey on network flow watermarking, including DSSS-based
solutions, is depicted in [13]. In [14], the general idea of using
This work was co-financed by the Military University of Technology
under Research Project no. UGB/23-864/2023/WAT on “Watermark
embedding and extraction methods as well as aggregation and spectral
analysis methods using neural networks”.
978-83-956020-7-8 ©2023
Warsaw University of Technology
203 SPSympo 2023
Authorized licensed use limited to: Wojskowa Akademia Techniczna. Downloaded on December 18,2023 at 09:19:05 UTC from IEEE Xplore. Restrictions apply.

FHSS and DSSS techniques for information hiding is
presented.
A physical layer watermarking technique called
watermarked DSSS, which embeds authentication
information in the pseudo-noise (PN) codes (i.e., spreading
sequences) of the DSSS system, is presented in [15]. The
author of [16] proposed the use of DSSS-based steganography
for slow-scan television using radio amateur channels. In [17],
a new class of flow marking technique for invisible traceback
based on DSSS, utilizing a PN code, is shown. A special case
of this method with long PNs is included in [18].
DSSS is also used to hide information in audio signals.
The authors of [19] proposed a method of steganalysis, i.e.,
detection of hidden data using DSSS, in audio signals, which
is based on statistical moments of histogram in discrete cosine
transform, frequency, and wavelet transform domains
III. DSSS STEGANOGRAPHY BASED ON VARIABLE
PROCESSING GAIN
In this paper, we propose a new steganography method
that uses variable processing gain in DSSS technique. The
scheme of hidden data embedding and extraction is shown in
Fig. 1. The presented scheme uses an exemplary binary phase
shift keying (BPSK) modulator, utilized, e.g., in satellite
communications [20].
Fig. 1. Embedding and extracting data hiding in DSSS system using
variable processing gain.
In a typical DSSS system, the PN code as a spreading
sequence is used to hide the cover signal in the noise. In
developed approach, covert bits are transmitted by keying the
used spreading sequences (i.e., in the used PN code order),
which are taken from a PN dictionary. This dictionary is
known to each user of the system. While the method of
mapping the PNs to individual bits or words is known only to
trusted parties performing covert transmission.
The hidden information extraction is done by checking the
individual spreading sequences from the dictionary, for which
received maximum processing gain. The received covert bit or
symbol (word) corresponds to the PN code with maximum
gain. To ensure the effective decoding of hidden data in the
receiver, individual PN codes from the dictionary may be
checked in parallel.
The difference between the proposed solution and the
typical DSSS steganography technique is worth highlighting.
In our approach, keying processing gain is a hidden message.
Finally, the lengths of the PNs should be selected in such a
way that the difference in their gains is subtle, i.e. undetectable
to an outside observer. It should ensure the transparency of the
method. On the other hand, this gain difference must be large
enough for the covert demodulator to decode secret messages
correctly. It should ensure robustness for this technique. In the
proposed approach, hidden information depends on the info-
mapper of processing gains of used spreading sequences.
Formally, a dictionary consisting of two PN codes is
sufficient to transmit each 0 or 1 bit. We may transmit N-bit
symbols (words), if the dictionary is based on 2N spreading
sequences. Hidden bits or symbols may correspond to specific
PN codes from the dictionary or a specific change between
spreading sequences - the so-called differential method.
The transmission rate of covert information depends on the
transmission rate of cover data and the length of the used PN
codes.
IV. EMBEDDING AND EXTRACTING DATA HIDING.
SIMULATION RESULTS.
A. Classic DSSS Method
In DSSS, the digital signal is spread by a code sequence
determined by the spreading factor, otherwise known as
processing gain. This parameter is expressed as:
bit
chip
T
PG T
=
()
where Tbit and Tchip are the one-bit duration of the transmitted
signal and one-chip duration, expressed by spreading the
original signal, respectively.
Figure 2 shows an example of the so-called "chip bank",
i.e., spreading sequences with different Tchip. The set of signals
present in the bank can be used to spread the original digital
signal, in relation to the duration of Tbit. To simplify further
analysis, we assume Tbit = 1 s, then for the presented chip
bank, the processing gain is equal to
PG = {1, 2, 3, 4, 5, 6, 8, 10, 12} [s], where the smallest and
largest PG values correspond to Tchip = 1 s and Tchip = 0.083 s,
respectively. In the presented graphs, differences in the
amplitude of the chips have been used only for a more clear
representation.
Fig. 2. Exemplary chip bank in time domain.
Figure 2 depicts the representation of the chip bank in
frequency domain. The graphs show that the energy of the
transmitted signal is spread in a wider band, when the duration
of the pulse (i.e., chip) is shorter.
204 SPSympo 2023
Authorized licensed use limited to: Wojskowa Akademia Techniczna. Downloaded on December 18,2023 at 09:19:05 UTC from IEEE Xplore. Restrictions apply.

Figure 4 illustrates the spreading process of an example
digital signal with binary values s(t) = [1, 1, 0, 1, 1, 1, 0, 1]
for two chips with processing gains equal to PG1 = 3 and
PG2 = 4, respectively.
As a result of the operation shown in Fig. 4, a new
modulated digital signal is created. The chip duration of this
signal is defined by the processing gain, and its energy takes
the characteristic shapes shown in Fig. 2. The discussed
process is a classic representation of the DSSS method.
Fig. 3. Analyzed chip bank in frequency domain.
Fig. 4. Exemplary binary signal (graph above), speading sequences (chips)
with PG1 = 3 and PG2 = 4 (graph inside), and spreading analyzed binary
signal using the selected chips (graph below).
B. Embedding and Extracting Data Hiding
According to the proposed steganography method, the
processing gain parameter can be used to hide information in
the DSSS signal. The idea of this solution is shown in Fig. 5
in relation to the already specified signals and parameters
Fig. 5. Analyzed binary signal (graph above), speading sequences (chips)
with PG1 = 3 and PG2 = 4 (graph inside), and spreading digital signal using
these chips keyed with respect to successive bit (graph below).
The presented concept of hiding information in DSSS is
based on using the chips (i.e., spreading signals) with variable
processing gains (i.e., different lengths) to spreading the
transmitted cover bits (i.e., the digital signal), in accordance
with the adopted keying sequence relating to the covert data.
For the signals shown in Fig. 4, the hidden information is
represented by the 4-bit sequence h(t) = [1, 0, 1, 1],
represented by PG2 = 4 and PG1 = 3 as ‘0’ and ‘1’ symbols,
respectively.
By using a larger number of spreading sequences (i.e., PN
codes/chips) with a variable processing gain, it is possible to
send a larger amount of hidden data at the same time, e.g., for
PG1, PG2, PG3, and PG4, the ‘00’, ‘01’, ‘10’ and “11’ symbols
can be encoded. It should be noted that the above example is
illustrative, and the interpretation of symbols can be
represented by various combinations of spreading codes.
The digital signal spread in this way is modulated on the
modulator. Based on example presented in Fig. 1, we use
simple BPSK modulator. So, the transmitted cover signal
spreading by chips keying by covert bits is modulated by
harmonic signal at carrier frequency (see Fig. 6).
Fig. 6. Transmitted signal at input and output of BPSK modulator.
Proper operation of the proposed steganographic method
requires ensuring synchronous and correct detection of BPSK-
DSSS signals. For this purpose, the received signal is detected
using a Costas loop receiver. It is a receiver with quadrature
detection and a phase lock loop (PLL) system with a voltage-
controlled oscillator (VCO). Adjusting the phase of the local-
generated harmonic signal to the received-signal phase is the
primary task of the receiver. The detection of the received
signal depending on the different values of the phase shiftinh
is shown in Fig. 7.
Fig. 7. Signal detection in Costas loop receiver with different phase
shifting.
The digital signal form recovered in this way must be
concentrated to its original form. For this aim, the receiver,
using the available spreading sequences, detects cover bits.
Implementation of the detection process allows the
demodulation of cover bits and hidden data. The covert bits
205 SPSympo 2023
Authorized licensed use limited to: Wojskowa Akademia Techniczna. Downloaded on December 18,2023 at 09:19:05 UTC from IEEE Xplore. Restrictions apply.

are included in the order of used chips. The idea of this process
is depicted in Fig. 8.
Fig. 8. Conception of detection spreading codes in received signal.
The upper graph shows the signal received detection by
the BPSK demodulator in the time window Tobserv = 8 s for
Tbit = 1 s. The remaining two graphs show individual detection
windows of the received signal spread by two different chips
with variable processing gains (PG1 and PG2), respectively, in
the duration of Tbit. In the given example, the transmitted
signal is spread by code sequences with PG1 = 3 and PG2 = 4.
The generated spreading codes, the current transmission of the
received digital signal, and the representation of the
reproduced value of the expected signal with the duration of
Tbit, after applying the XOR (i.e., eXclusive OR) function are
marked by dashed black, continuous red or black, and
continuous blue lines, respectively.
Figure 9 illustrates versus time, the energy obtained for the
applied spreading sequences with considering the reference
level, i.e., bringing the maximum energy for Tbit. Approaching
the limits of [0, 1] by any of the waveforms (i.e., red or blue
lines) allows the reconstruction of explicit as cover as covert
information.
Fig. 9. Received signal reconstruction based on detection energy versus
time for two analyzed spreading sequences.
Figure 10 shows the recovered useful signal (cover bits),
along with the secret information (covert bits) that was hidden
in it.
Fig. 10. Restored cover and covert data.
V. SUMMARY
This paper focuses on the novel technique of
steganography in DSSS signal. In this case, we proposed used
keying of spreading sequences (i.e., PN codes) for hiding
information in variable processing gain. Currently, this new
method of embedding and extracting hidden bits is
implemented in the MATLAB environment. In the paper, we
present embedding and extracting hidden data in DSSS signal
based on simulation studies. In the near future, we will present
the effectiveness of embedding and extraction of covert bits
based on the developed technique. In the next step, the authors
plan to implement this method on the USRP SDR platform.
REFERENCES
[1] H. He, J. Li, and P. Stoica, Waveform design for active sensing systems:
A computational approach. Cambridge ; New York: Cambridge
University Press, 2012.
[2] M. K. Simon and J. K. Omura, Spread spectrum communications
handbook, 1st edition. McGraw Hill, 2001.
[3] J. Fridrich, Steganography in digital media: Principles, algorithms,
and applications. Cambridge University Press, 2009.
[4] W. Mazurczyk, S. Wendzel, S. Zander, A. Houmansadr, and K.
Szczypiorski, Information hiding in communication networks:
Fundamentals, mechanisms, applications, and countermeasures, 1st
edition. Wiley-IEEE Press, 2016.
[5] R. Szczepanik and J. M. Kelner, “Software-based implementation of
SDF method using SDR USRP platform,” Telecommun. Rev.
Telecommun. News, (in Polish “Implementacja metody SDF na
platformie radia programowalnego,” Przegląd Telekomunikacyjny i
Wiadomości Telekomunikacyjne) vol. 92, no. 6, pp. 217–222, Jun.
2019, doi: 10.15199/59.2019.6.18.
[6] Z. Piotrowski and J. M. Kelner, “Radio steganography – Hazards and
challenges,” Telecommun. Rev. Telecommun. News, vol. 90, no. 6, pp.
165–172, Jun. 2017, doi: 10.15199/59.2017.6.2.
[7] K. Grzesiak, Z. Piotrowski, and J. M. Kelner, “A wireless covert
channel based on dirty constellation with phase drift,” Electronics, vol.
10, no. 6, Art. no. 6, Jan. 2021, doi: 10.3390/electronics10060647.
[8] K. Grzesiak, Z. Piotrowski, and J. M. Kelner, “Covert channel based
on quasi-orthogonal coding,” Electronics, vol. 12, no. 10, Art. no. 10,
Jan. 2023, doi: 10.3390/electronics12102249.
[9] K. Grzesiak and Z. Piotrowski, “NN-based 8FSK demodulator for the
covert channel,” Sensors, vol. 22, no. 19, Art. no. 19, Jan. 2022, doi:
10.3390/s22197181.
[10] K. Grzesiak and Z. Piotrowski, “From constellation dithering to
NOMA multiple access: security in wireless systems,” Sensors, vol. 21,
no. 8, Art. no. 8, Jan. 2021, doi: 10.3390/s21082752.
[11] J. Wojtuń and Z. Piotrowski, “Synchronization of acoustic signals for
steganographic transmission,” Sensors, vol. 21, no. 10, Art. no. 10, Jan.
2021, doi: 10.3390/s21103379.
[12] K. Sawicki, G. Bieszczad, and Z. Piotrowski, “StegoFrameOrder—
MAC Layer Covert Network Channel for Wireless IEEE 802.11
Networks,” Sensors, vol. 21, no. 18, Art. no. 18, Jan. 2021, doi:
10.3390/s21186268.
206 SPSympo 2023
Authorized licensed use limited to: Wojskowa Akademia Techniczna. Downloaded on December 18,2023 at 09:19:05 UTC from IEEE Xplore. Restrictions apply.

[13] L. Zhang, Y. Kong, Y. Guo, J. Yan, and Z. Wang, “Survey on network
flow watermarking: model, interferences, applications, technologies
and security,” IET Commun., vol. 12, no. 14, pp. 1639–1648, 2018, doi:
10.1049/iet-com.2017.1053.
[14] B. A. Bash, D. Goeckel, D. Towsley, and S. Guha, “Hiding information
in noise: Fundamental limits of covert wireless communication,” IEEE
Commun. Mag., vol. 53, no. 12, pp. 26–31, Dec. 2015, doi:
10.1109/MCOM.2015.7355562.
[15] X. Li, C. Yu, M. Hizlan, W.-T. Kim, and S. Park, “Physical layer
watermarking of direct sequence spread spectrum signals,” in 2013
IEEE Military Communications Conference (MILCOM), San Diego,
CA, USA, Nov. 2013, pp. 476–481. doi: 10.1109/MILCOM.2013.88.
[16] A. Westfeld, “Steganography for Radio Amateurs— A DSSS Based
Approach for Slow Scan Television,” in Information Hiding, Springer,
Berlin, Heidelberg, Jul. 2006, pp. 201–215. doi: 10.1007/978-3-540-
74124-4_14.
[17] W. Yu, X. Fu, S. Graham, D. Xuan, and W. Zhao, “DSSS-based flow
marking technique for invisible traceback,” in 2007 IEEE Symposium
on Security and Privacy (SP), Berkeley, CA, USA, May 2007, pp. 18–
32. doi: 10.1109/SP.2007.14.
[18] J. Huang, X. Pan, X. Fu, and J. Wang, “Long PN code based DSSS
watermarking,” in 2011 IEEE Conference on Computer
Communications (IEEE INFOCOM), Shanghai, China, Apr. 2011, pp.
2426–2434. doi: 10.1109/INFCOM.2011.5935064.
[19] C. Wang, L. Guo, and Y. Wang, “Audio steganalysis of DSSS based
on statistical moments of histogram,” J. Electron. China, vol. 26, no.
5, pp. 659–665, Sep. 2009, doi: 10.1007/s11767-009-0050-2.
[20] R. Lu, G. Ye, J. Ma, Y. Li, and W. Huang, “A numerical comparison
between FHSS and DSSS in satellite communication systems with on-
board processing,” in 2009 2nd International Congress on Image and
Signal Processing (CISP), Tianjin, China, Oct. 2009, pp. 1–4. doi:
10.1109/CISP.2009.5304577.
207 SPSympo 2023
Authorized licensed use limited to: Wojskowa Akademia Techniczna. Downloaded on December 18,2023 at 09:19:05 UTC from IEEE Xplore. Restrictions apply.