Open Access
Add to BibSonomy
Abstract
We report a real-time 150 kbps stealth transmission within public optical communication of 10 Gbps dual polarization QPSK. The stealth data is modulated onto the frequency tuning signals of a fast-tuning laser source in the transmitter, which causes slight frequency dithering for the transmitted optical signal. In the receiver, the stealth receiver recovers the stealth data from the estimated frequency offset by the QPSK DSP algorithm. The experiments show the stealth transmission has few impacts on the public channel over a 300 km distance. The proposed method is fully compatible with existing optical transmission systems, and the only hardware change is to upgrade the transmitter laser to support frequency tuning through an external analog port for receiving stealth signal. The proposed stealth scheme can combine with cryptographic protocols to improve the integrated security of the system, and can be used as signaling transport for low level network control to reduce the communication overhead.
© 2024 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
With the widespread application of mobile internet, the enormous data traffic in the cyberspace poses challenges to information security and privacy. At the same time, the scientific community and the general public are increasingly concerned about data security and privacy issues. Security schemes at the physical layer are promising technologies against security threats. Steganography and cryptography and are commonly used techniques to achieve security at the physical layer of the network. In optical networks through fiber or free space, optical domain encryption is an effective strategy for secure transmission, which protects the content of data from eavesdroppers and enables high-speed and secure data transmission over fiber optic networks. Various optical encryption methods, such as quantum cryptography [1,2], chaotic cryptography [3–8] and quantum noise stream cipher (QNSC) [9, have been proposed based on physical mechanism.
In practice, a secure communication system needs overall security countermeasures in all aspects. Steganography or stealth transmission in optical domain can be integrated as a complementary measure to provide high-level security for confidential information in optical secure communication system. Unlike the encryption scheme for protecting the plain data away from the unauthorized access or modification, steganography focuses on concealing the existence and transmission traces of data. In such strategy, the stealth data is usually concealed in the normal transmitted signal by an unnoticed method, therefore, an eavesdropper will not even pay attention to the presence of the stealth data before attempting to access it.
Up to now, several optical steganographic schemes have been proposed. The concept of optical domain spread-spectrum is used for covering the stealth data with low SNR by different methods, for instance, chirped fiber Bragg gratings (CFBG) [10,11], optical thresholder [12] and spectral notched temporal phase code [13]. Other proposed schemes use wideband optical signal (e.g., ASE noise) [14–16] and gain switched frequency comb as laser source [17,18]. In these steganography schemes, good steganography effects have been achieved, however, the stealth and public signals need to be combined in transmitter and split in receiver in a certain domain, and therefore the stealth channel needs to be constructed with extra dedicated hardware, such as separate optical sources, grating and modulator, which increases the overall cost and lowers the compatibility to existing network infrastructure and devices.
In our previous work, the slight and unobtrusive distortion is artificially made in the optical output signal, and it can be considered as an artificial unique feature, i.e., hardware fingerprints [19]. Similar principles have already been used to achieve physical layer authentication [20], watermarking [21], chaotic signal feature extraction [22] and stealth transmission [23,24]. In [24], we report 2 kbps stealth transmission via dither-remodulation in bias controller of Mach-Zehnder modulator (MZM). In this stealth scheme, the stealth data is embedded in the bias state of MZM in the transmitter, and recovered from the demodulated public baseband signal in the receiver. Thus, the stealth performance is dependent on the status of public QPSK receiver. Those tests show that the stealth bit error rate (BER) relegates rapidly while public BER degrading around the forward error correction (FEC) limit (∼${10^{ - 2}}$) due to the frequent phase jump in the phase compensation of QPSK digital signal processing (DSP).
In this work, we propose a novel optical steganography scheme based on frequency tuning of the laser. Different from the conventional schemes based on spectrum-spreading or noise modulation, the stealth data here is modulated into the frequency tuning signals of a fast-tuning laser source, which causes slight frequency dithering in the transmitted optical signal. Thus, the stealth information can be recovered from the estimated frequency offset in the receiver. In the coherent DSP structure [25], the frequency offset correction algorithm [26] is the intermediate step, which is the subsequent step after channel equalization (e.g., constant modulus algorithm, CMA) [27], and followed by the phase compensation algorithm. The estimated frequency offset is often averaged across a second for receiver laser wavelength feedback. The instantaneous value of the frequency offset used for the proposed stealth scheme is always encapsulated in the DSP algorithm, hence the existence of stealth signal is not susceptible to suspicion. The security of the proposed optical steganography scheme counts on constructing a stealth receiver for low SNR, for which the power is -40 dB lower than those of the inherent frequency instability of the laser. Lower SNR ensures stealthier transmission but inevitably brings higher difficulty. The experiments demonstrated 150 kbps real-time stealth transmission over 300 km fiber accompanied by [10] Gbps public transmission. The results show that the stealth receiver acts as a high sensitivity demodulator towards the small frequency dither signal in the proposed optical steganography strategy. Benefits from the tiny and deeply embedded dithering signal, the stealth information can be barely observed through the normal transmission.
In contrast to our previous work in [24], the stealth signal in this work is split before phase compensation algorithm, and will not be affected by the issue of phase jump. Thus, it achieves higher stealth sensitivity comparing to the public transmission, resulting in stronger robustness. In addition, higher stealth bit rate is achieved and the transmission experiment with longer fiber is demonstrated in this work.
In the proposed steganography strategy, the stealth transmission and the public transmission share the common physical resources, including laser source, MZM, coherent receiver, analog-digital convertors (ADC), FPGA, etc. There is no need to adopt the technology for precise split of the public and stealth channels. The only hardware change is to upgrade the transmitter laser to support frequency tuning through an external analog port for receiving the stealth signal, which laser is essential for the normal data transmission. In the receiver, the stealth data is extracted along with the public data demodulation DSP algorithm. Meanwhile the stealth algorithm utilizes a very small amount of FPGA resources. Therefore, the changes for the existing optical coherent system are minimized, and it is possible to support the proposed stealth transmission on the existing system by upgrading the transmitter laser and updating the FPGA firmware. It constructs the stealth channel without changing the existing optical coherent transmission framework, and has little impact to the public channel. The excellent compatibility indicates that the proposed strategy is an alternative for transferring information stealthily and privately in the physical layer of optical networks.
2. Principle and experimental setup
2.1 Stealth transmission structure
The stealth transmission system consists of the transmitter, the transmission medium (e.g., fiber, free space) and the receiver, as shown in Fig. 1. The stealth channel and the public channel share the common physical resources and most of the components, such as two laser sources, MZM, coherent receiver, ADC and DSP algorithms.
In the transmitter, we employ a fast-tuning laser component for the common laser source for the public and stealth channels. Benefits from the piezoelectric transducer (PZT) tuning technology [28], the laser component with PZT of high modulation bandwidth achieves fast-tuning for the tuning frequency above 100 kHz, and provides an analog input tuning port with the tuning factor of approximately 400 MHz/V. For optimal performance, a stealth signal generator is preferred, which generates a small-amplitude frequency dither signal that is applied to the tuning port of the laser. In this condition, the dither signal is designed in form of triangle-like waveform, which minimizes the slope of frequency change comparing to other waveforms such as square and sine wave, to minimize the impact to the public performance. The amplitude of dither signal is in the range of 5∼20 mVpp, which generates approximately frequency dithering of 2∼8 MHz in the optical signal. The stealth signal generator consists of a signal transformer and a low-speed DAC, and it can be physically integrated into the transmitter laser due to its simple structure, as shown in Fig. 2. The signal transformer runs in FPGA and utilizes a few resources. It converts the stealth binary data to triangle-like waveform in digital domain, by modifying the rising and falling edges of stealth bit transition to linear DAC code change, which produces triangle-like waveform corresponding to the stealth data. The triangle-like waveform is then sent to DAC to produce the analog signal of the frequency dither signal. As a result, the stealth information is carried by the frequency dithering in the laser signal, and then delivered to the laser port of MZM. On the other hand, the public signal is amplified by the RF amplifier, and then sent to the RF port of MZM for electronic-optical conversion as the normal method. Therefore, the output optical signal of the transmitter carries the public data combined with tiny frequency dithering which carries the stealth data.
In the receiver, a coherent receiver, which consists of optical hybrid and balanced photodetector, converts the optical signal to electronic signal. The optical hybrid accomplishes the frequency mixing for the incoming optical signal and local laser signal, and the balanced photodetector converts the beat frequency signal to electronic signal. ADC digitalizes the electronic signal to digital domain and then sends it to DSP algorithm in FPGA for demodulation. DSP algorithm for QPSK demodulation mainly includes chromatic dispersion compensation (CDC), CMA, frequency offset compensation algorithm and phase compensation algorithm. In normal transmission, CDC algorithm is utilized to compensate the chromatic dispersion of the fiber, CMA algorithm is used for channel equalization, frequency compensation algorithm is employed to compensate the frequency offset to near 0, phase compensation algorithm is deployed to compensate the phase difference to 0, and then the baseband signal is recovered and then decided to public data.
On the other side, the stealth transmission shares the CDC algorithm, CMA algorithm and frequency compensation algorithm. Chromatic dispersion is often compensated by FIR filters with a few to hundreds of taps, and the number of taps depends on the baud rate and fiber distance [29]. However, the stealth frequency (∼100 KHz) is 1/10000 lower than the normal data frequency (beyond 1 GHz), thus the stealth signal can be pass through the chromatic dispersion FIR filter with hundreds of taps. The signal path in the CMA algorithm is a 5-tap finite impulse response (FIR) filter, thus the frequency dither signal can also pass through the CMA algorithm. In the frequency compensation algorithm, the frequency offset $\Delta \omega $ between the two laser sources, including the frequency difference, the frequency instability and the stealth frequency dithering, is estimated first. Considering that Integrable Tunable Laser Assembly (ITLA) is usually employed in the coherent transmission system, and the inherent frequency instability of ITLA can reach 200 MHz in 1 ms. Thus, the frequency offset curve is dominated by the frequency instability of ITLA rather than the tiny frequency dithering, and the SNR of frequency dithering is low. For this reason, the eavesdropper is barely aware of the existence of the stealth signal, meanwhile, the stealth receiver should handle the low SNR signal processing. The stealth receiver accepts the frequency offset signal to extract the frequency dithering and then restore the stealth data. The stealth receiver consists of high-pass filter (HPF), DC cancellation, data recovery and data decision. Due to the low SNR for frequency dithering, HPF is firstly performed for improving the stealth SNR at the first time. Secondly, DC cancellation removes the residual DC level. After that, the data recovery finds the best phase for sampling. Finally, the stealth data can be simply decided by its sign bit.
For the employment of commercial ITLA, the spectrum of the receiver laser frequency instability will overlap with the stealth signal of raw PRBS data, thereby it causes signal distortion and error decision. One feasible method is encoding the stealth binary data to Manchester code, which moves the dominated power to higher frequency, for minimizing the overlap with the frequency instability of the receiver laser. However, Manchester coding lowers the effective data rate to a half of raw code rate. Another practicable method is utilizing a receiver laser with narrower linewidth, which accompanied by better frequency stability.
In the proposed stealth transmission strategy, the stealth information is hidden by frequency dithering for transmitter laser that is commonly used in optical transmission systems. Thus, it achieves good compatibility with existing optical coherent system and low cost for hardware upgrade. Besides, the stealth processing algorithm requires low FPGA resource utilization due to the low complexity.
2.2 Experimental setup
We demonstrate a real-time 150 kbps stealth transmission over a dual polarization QPSK (DP-QPSK) transmission with the public data rate of 10 Gbps, as shown in Fig. 3. In the transmitter, a PZT featured fast-tuning laser component is employed for transmitter laser source. The stealth data is sent to the frequency tuning port provided by the fast-tuning laser. Meanwhile, the dual polarization IQ modulator (Fujitsu FTM7977) converts the public data to optical signal. In the setup, we utilize a Xilinx Kintex 7 FPGA (XC7K325 T) as the public and also the stealth data source. In another case of stand-alone stealth data source, no additional operation is needed due to the independence between the public and stealth transmission. The wavelength used in the experiment is C band beyond 1550 nm. Single mode fiber (SMF) with different length of 0 km, 20 km, 100 km and 300 km is used for the transmission medium. In the receiver, we employ commercial ITLA (Emcore TTX1994) for receiver laser source. The input optical signal is received by coherent receiver (Fujitsu FIM24706), and then sampled by four 5 GSa/s ADCs. The real-time DP-QPSK DSP algorithms runs in a Xilinx Ultrascale + FPGA (XCVU9P). The demodulated public data is sent to Bit Error Rate Tester (BERT) for error statistics. The stealth receiver and the stealth BERT also run in the Ultrascale + FPGA.
2.4 Principle of the stealth transmission
The angular frequency of the transmitter laser can be described as
where ${\omega _{T0}}$ is the nominal frequency, ${\omega _{Tinst}}$ is the frequency instability, and ${\omega _{stealth}}$ is the external applied stealth frequency dithering.Similarly, the angular frequency of the receiver laser can be described as
where ${\omega _{R0}}$ is the nominal frequency, and ${\omega _{Rinst}}$ is the frequency instability.In the receiver, the coherent receiver achieves optical frequency mixing and optical-electronic conversation. In the DSP algorithm, the frequency compensation algorithm estimates the frequency offset $\Delta \omega $ between the two lasers, as
We use the conventional Viterbi-Viterbi (V-V) algorithm [30] for QPSK frequency offset estimation. In addition, the stealth receiver relies on the frequency offset estimation value output by any type of frequency compensation algorithm, as well as the FFT-based frequency compensation algorithm. Regardless of the different modulation formats, it is necessary to estimate the frequency offset value and generate the IF carrier for frequency compensation in the digital coherent DSP structure. The proposed stealth scheme relies on the frequency estimation value, therefore the proposed stealth scheme is applicable for various formats including BPSK, QPSK and QAM.
In our DSP configuration, the average time for frequency offset estimation is 1.6384 µs, indicating that the sampling rate of $\Delta \omega $ is nearly 610 kSa/s. Thus, the stealth frequency dithering ${\omega _{stealth}}$ of 150 kbps can be fully transferred to $\Delta \omega $. In the normal transmission, $\Delta \omega $ is used for generating intermediate frequency (IF) carrier to compensated the frequency offset, and the averaged value of $\Delta \omega $ in typical 1 second is calculated for receiver laser wavelength feedback control. After this, the instantaneous value of $\Delta \omega $ is always discarded. However, the proposed stealth receiver further processes the instantaneous signal of $\Delta \omega $ to extract ${\omega _{stealth}}$.
The characteristics of the three components of $\Delta \omega $ in Eq. (3) are:
- 1) The nominal frequency difference ${\omega _{diff}}$ can be considered as a fix value, and presents as a DC level.
- 2) The frequency instability difference ${\omega _{inst}}$ depends on the physical characteristics of the laser, is typically 200 MHz in 1 ms for ITLA. The time domain waveform is similar to sine wave, and its frequency is beyond 900 Hz.
- 3) The stealth frequency dithering ${\omega _{stealth}}$ is 2 to 8 MHz (applies 5 to 20 mVpp with factor of 400 MHz/V), in which applies lower amplitude for stealthier, and higher amplitude for better stealth SNR.
It can be clearly observed that, the amplitude of ${\omega _{stealth}}$ is 1/100 to 1/25 of ${\omega _{inst}}$, the SNR of ${\omega _{stealth}}$ reaches $SN{R_{stealth}} = $ -40 to -28 dB.
The diagram of stealth receiver is shown as Fig. 4. At first, $\Delta \omega $ of two polarizations are summed up to low down the noise because they are from the same source. As a consequence, the signal is sent to HPF. For the best split for the spectrums of ${\omega _{inst}}$ and ${\omega _{stealth}}$, HPF is designed for the stop frequency of 1 kHz.
The HPF output is
where ${\omega _{residual}}$ is the residual DC level and frequency instability passed through the HPF.The following DC cancellation process calculates the mean value of ${\omega _{HPF}}$, and subtracts the mean value from ${\omega _{HPF}}$. Thus, the stealth signal ${\omega _{stealth}}$ is restored.
In the experiment, we use the common clock source for public normal data and stealth data in the transmitter, so that the stealth clock shares the common clock source with the public clock that derived from the normal real-time CDR clock recovery loop in the receiver. Due to the low-speed stealth bit rate, multi-phase sampling is employed for data recovery. Stealth clock generator is utilized to generate 4 times sampling clock to perform 4x sampling. SNR calculation runs for each phase and operates on demand (e.g., stealth data link initialization), thus it can be performed by either hardware or software, and the software implementation is preferred due to no occupation of the dedicated hardware resource. The phase with best SNR is selected for final data decision to recover the stealth data.
Because of the independence between the public and stealth transmission, the public and stealth clocks can be synchronous or asynchronous. In the case of stealth data with a different clock source, an asynchronous data receiver, such as a Universal Asynchronous Receiver/Transmitter (UART), can be used to replace the data recovery module in Fig. 4.
3. Experiment results and analysis
3.1 Steganography effects
Figure 5 shows the comparisons of the optical spectrums, ADC digitalized signal spectrums, the baseband waveforms and spectrums for the signals in the conditions of stealth transmission on and off. The optical spectrums are acquired by an optical spectrum analyzer (OSA). The snapshots for real-time digital signals are captured by the integrated logic analyzer (ILA) in FPGA and exported to the computer, and then FFT is preformed to present the digital spectrum. Figure 5(a)–5(d) show the optical spectrum, ADC digitalized signal spectrum, the baseband signal waveform and the baseband spectrum, respectively, when stealth transmission is on. As comparisons, Fig. 5(e)–5(h) show the respectively drawing when stealth transmission is off. These figures show the similar drawings between stealth on and off in both time domain and frequency domain. Since the stealth signal is embed in the public signal by the tiny frequency dithering, it is deeply covered in the normal signals and the low SNR of the dither signals is also verified. Furthermore, comparisons between these featured plots when stealth transmission is on and off also show that the stealth signal is deeply hided and cannot be observed, and no clues leak about the stealth data. Therefore, the test indicates the good steganography effects of the proposed stealth scheme.
Fig. 5. Comparisons of signals for stealth transmission on and off. (a)-(d) Optical spectrum, ADC digitalized signal spectrum, baseband signal and baseband spectrum, respectively, when stealth is on. (e)-(h) Optical spectrum, ADC digitalized signal spectrum, baseband signal and baseband spectrum, respectively, when stealth is off.
3.2 Performances of the stealth receiver
Figure 6 shows the signals in the stealth receiver. The waveforms are drawn from the data snapshotted and exported by FPGA ILA. Figure 6(a)–6(c) shows the waveforms of the frequency offset signal, HPF output and recovered stealth signal. Figure 6(d)–6(f) show the detail drawing of Fig. 6(a)–6(c). Figure 6(g)–6(i) show the spectrums of Fig. 6(a)–6(c). Figure 6(a) shows the frequency offset signal is dominated by the frequency instability of the laser, besides, Fig. 6(g) shows the stealth SNR is below -40 dB, where the signal component is the wideband stealth signal and the noise component is the narrowband spur caused by the laser frequency instability. After HPF filtering, the stealth SNR is significantly improved to near 0 dB and the stealth signal can be clearly found. However, a slight DC offset remains as shown in Fig. 6(e), and will affect the stealth decision when SNR is low. The stealth signal is recovered after DC cancellation as shown in Fig. 6(c), (f) and (i). The stealth data decision can be performed simply by the threshold value of zero, i.e., by the sign bit. The test proves that, the stealth signals can be extracted from the frequency offset signal, and no additional hardware component (e.g., photo-detector or ADC) is needed to deploy for the stealth receiver.
Fig. 6. The results for the stealth receiver. (a)-(c) The waveforms of frequency offset, HPF output and restored stealth signal, respectively. (d)-(f) The detail drawing of (a)-(c). (g)-(i) The spectrums of (a)-(c).
3.3 Performances of the stealth transmission
Figure 7 shows the tests for the SNR and bit error rate (BER) for both public and stealth transmission. In these tests, the SNR is defined as the ratio of the stealth signal to the noise floor, and is calculated in the time domain. Figure 7(a) shows the test setup. In the test, a variable optical attenuator (VOA) performs attenuation to emulate the insertion loss of the transmission medium, and a 1:1 coupler splits the received signal to two half branches, one half for receiver and another half for power record. An erbium-doped fiber amplifier (EDFA) boosts the weak optical signal to the appropriate power for the coherent receiver, meanwhile, a power meter records the optical power of received signal. We record the SNR and BER measurements, and export ILA data in the conditions of different power while changing the attenuation value. Figure 7(b) shows the BER and SNR results under different receiving power. The constellation of public data, the waveform and eye diagram of stealth signal at -46 dBm are shown as Fig. 7(c)–7(e), respectively. The respective drawings at -34 dBm are shown as Fig. 7(f)–7(h). The public BER is near ${10^{ - 1}}$ below -47 dBm due to the frequent reversal of data polarity caused by the phase jump of the V-V algorithm for phase compensation in the QPSK DSP when SNR is low. The BER tests show no stealth error bit is detected when the optical power is ≥ -43 dBm, and thus a dash line segment of stealth BER is added in Fig. 7(b), indicating the estimated trend while the optical power is higher. It is clearly observed that, the public and stealth BER curves cross the LDPC FEC limit (∼${10^{ - 2}}$) between -46 dBm and -47 dBm, and the stealth sensitivity is ∼0.4 dB higher than the public transmission. Furthermore, the stealth BER is approximately equal to and even lower than the public BER in all conditions, that presents strong robustness of the stealth transmission. The tests prove that the stealth transmission with strong robustness achieves higher sensitivities than the public transmission.
Fig. 7. SNR and BER tests. (a) The test setup. (b) BER and SNR vs. receiver input optical power. (c)-(e) Constellations of public data, the stealth signal and eye diagram at -45 dBm, respectively. (e)(f) Constellations of public data, the stealth signal and eye diagram at -34 dBm, respectively.
3.4 Impacts of stealth transmission to the public transmission
Figure 8 shows the test of stealth transmission over fiber. Figure 8(a) shows the test setup for the transmission over long-haul SMF with two EDFAs of one for power amplification on TX side and another for low noise amplification on RX side. The stealth and public transmission operate normally over fiber length from 0 km to 300 km, indicating that it can effectively confront the EDFA produced noise. We record the corresponding error vector magnitude (EVM) measurements while changing the fiber length to 0 km, 20 km, 100 km or 300 km in the condition of stealth on or off. EVM measurement is calculated real-time and then averaged in FPGA. It can be observed that the differences of EVM results are very small as the comparison histogram shown in Fig. 8(b). Besides, the public data constellation shown in Fig. 8(c) and 8(d) are similar. Figure 8(e) shows that the demodulated stealth signal is clear when stealth is on. By contrast, Fig. 8(f) shows no stealth signal is recovered when stealth is off. Furthermore, the EVM differences are counted for various fiber lengths, as shown in Fig. 8(g). We find the EVM differences are in the range of ±0.2%, accordingly, the small differences can be considered as statistical error and partially caused by the DC bias drift of the MZM modulator. Therefore, the result proves that the stealth transmission has little impact on public transmission.
Fig. 8. The tests of stealth transmission over fiber. (a) The test setup. (b) The EVM comparison when stealth on and off. (c)(d) The public data constellations over 300 km fiber when stealth is on and off, respectively. (e)(f) The stealth signals over 300 km fiber when stealth is on and off, respectively. (e) The EVM differences between stealth is on and off.
3.5 Stealth transmission resource utilization
The proposed stealth transmission method can use existing coherent transmitter and receiver hardware except for the following two upgrades. On one hand, we upgrade the transmitter laser to support an analog port for frequency tuning. On the other hand, we upgrade the FPGA firmware in the transmitter and receiver to support stealth transmission, which will occupy a few FPGA resources. These configurations ensure real-time implementation of stealth transmission, as well as compatibility with the existing coherent system. In the proposed scheme, stealth data is transferred without any buffer or storage, thus the space complexity of the stealth algorithm is O(1). For the real-time transmission, the number of operations for each sample is a constant, thus the time complexity of the stealth algorithm is O(n). The FPGA resource utilization for the stealth transmitter and receiver are shown in Table 1 and 2, respectively. “LUTs”, “Block RAM Tiles” and “DSP slices” indicate the utilizations of logic resource, RAM and hardware multiplier, respectively.
Table 1. The FPGA resource utilization for the stealth transmitter
Table 2. The FPGA resource utilization for the stealth receiver
We can find the stealth receiver utilizes very small number of resources. The reason is that the stealth receiver is mainly implemented by a HPF and DC cancellation algorithm. However, HPF can be constructed with only a multiplier for series multiplexed structure due to the low sampling rate of the stealth transmission. On the other side, the SNR calculation for data recovery runs during the procedure of transmission initialization, thus it can be implemented by software in Microblaze soft CPU, and occupies none of the dedicated hardware resource.
4. Conclusion
The concept of this work is to create a stealth transmission channel on the existing optical transmission system with minimal modification to the existing devices and protocols, that brings better compatibility and lower cost. Due to the simple system structure, the proposed stealth transmission scheme is compatible with cryptographic methods. The proposed stealth scheme can combine with cryptographic protocols to achieve complete security of the system, and can be used as signaling transport for low level network control to reduce the communication overhead. More specifically, some confidential data, critical parameters or network signaling can be encrypted with algorithms, and then transmitted stealthily via the proposed scheme. However, the use of additional encryption strategies increases implementation costs and system complexity, so the use of additional encryption methods depends on the actual requirements of different scenarios.
In this work, we report a real-time stealth data transmission system by applying fast laser frequency dithering on transmitter laser. It is compatible with the existing optical coherent transmission system, by upgrading the transmitter laser to support fast frequency tuning, and adding stealth demodulation algorithm, which utilizes very small amount of FPGA resources. Therefore, the total cost of the proposed stealth scheme is small. The real-time experiment demonstrates 150 kbps stealth data transmission embed in 10 Gbps DP-QPSK public transmission over 300 km fiber with EDFA amplification. The results prove that, the steganography effect of the stealth transmission is significant, moreover, the impact on public transmission is little. In addition, the stealth sensitivity is slightly higher than the public transmission. Hence the proposed stealth transmission scheme can be employed in practical scenarios. In these scenarios, the demand of transferring data rate is relatively low, and the level of 100 kbps is enough in practice. Besides, the proposed stealth transmission scheme can provide extra security enhancement with no extra laser source, wavelength and bandwidth occupation. Moreover, further theoretical deduction suggests that, spread spectrum technology can be applied together for significantly improving the security of the stealth transmission.
Funding
National Key Research and Development Program of China (2021YFB1808200); National Natural Science Foundation of China (62175077, 62205115, 62275091); Key Research and Development Program of Hubei province (2023BAB008).
Disclosures
The authors declare no conflicts of interest.
Data Availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. M. Minder, M. Pittaluga, G. L. Roberts, et al., “Experimental quantum key distribution beyond the repeaterless secret key capacity,” Nat. Photonics 13(5), 334–338 (2019). [CrossRef]
2. Y.-A. Chen, Q. Zhang, T.-Y. Chen, et al., “An integrated space-to-ground quantum communication network over 4,600 kilometres,” Nature 589(7841), 214–219 (2021). [CrossRef]
3. H. Gao, A. Wang, L. Wang, et al., “0.75 Gbit/s high-speed classical key distribution with mode-shift keying chaos synchronization of Fabry–Perot lasers,” Light: Sci. Appl. 10(1), 172 (2021). [CrossRef]
4. Z. Gao, Z. Ma, S. Wu, et al., “Physical secure key distribution based on chaotic self-carrier phase modulation and time-delayed shift keying of synchronized optical chaos,” Opt. Express 30(13), 23953–23966 (2022). [CrossRef]
5. Y. Huang, P. Zhou, and N. Li, “High-speed secure key distribution based on chaos synchronization in optically pumped QD spin-polarized VCSELs,” Opt. Express 29(13), 19675–19689 (2021). [CrossRef]
6. K. Wu, H. Wang, and Y. Ji, “Channel characteristics estimation based on a secure optical transmission system with deep neural networks,” Opt. Express 30(18), 32391–32410 (2022). [CrossRef]
7. A. Zhao, N. Jiang, Y. Zhang, et al., “Semiconductor Laser-Based Multi-Channel Wideband Chaos Generation Using Optoelectronic Hybrid Feedback and Parallel Filtering,” J. Lightwave Technol. 40(3), 751–761 (2022). [CrossRef]
8. X. Zhang, P. Li, Z. Jia, et al., “Leader-laggard synchronization of polarization chaos in mutually coupled free-running VCSELs,” Opt. Express 31(2), 2414–2425 (2023). [CrossRef]
9. F. Futami, K. Tanizawa, and K. Kato, “Y-00 Quantum-Noise Randomized Stream Cipher Using Intensity Modulation Signals for Physical Layer Security of Optical Communications,” J. Lightwave Technol. 38(10), 2774–2781 (2020). [CrossRef]
10. Z. Gao, X. Wang, N. Kataoka, et al., “Stealth Transmission of Time-Domain Spectral Phase Encoded OCDMA Signal Over WDM Network,” IEEE Photonics Technol. Lett. 22(13), 993–995 (2010). [CrossRef]
11. Z. Wang and P. R. Prucnal, “Optical Steganography Over a Public DPSK Channel with Asynchronous Detection,” IEEE Photonics Technol. Lett. 23(1), 48–50 (2011). [CrossRef]
12. K. Kravtsov, B. Wu, I. Glesk, et al., “Stealth Transmission over a WDM Network with Detection Based on an All-Optical Thresholder,” in LEOS 2007 - IEEE Lasers and Electro-Optics Society Annual Meeting Conference Proceedings, (IEEE, Lake Buena Vista, FL, 2007), pp. 480–481.
13. X. Hong, D. Wang, L. Xu, et al., “Demonstration of optical steganography transmission using temporal phase coded optical signals with spectral notch filtering,” Opt. Express 18(12), 12415 (2010). [CrossRef]
14. B. Wu, Z. Wang, Y. Tian, et al., “Optical steganography based on amplified spontaneous emission noise,” Opt. Express 21(2), 2065 (2013). [CrossRef]
15. P. Y. Ma, B. Wu, B. J. Shastri, et al., “Steganographic Communication via Spread Optical Noise: A Link-Level Eavesdropping Resilient System,” J. Lightwave Technol. 36(23), 5344–5357 (2018). [CrossRef]
16. H. Zhu, Z. Liu, S. Chen, et al., “Optical stealth communication based on quantum noise stream ciphered amplified spontaneous emission light,” Opt. Express 31(3), 3595–3605 (2023). [CrossRef]
17. E. Wohlgemuth, Y. Yoffe, P. N. Goki, et al., “Stealth and secured optical coherent transmission using a gain switched frequency comb and multi-homodyne coherent detection,” Opt. Express 29(24), 40462–40480 (2021). [CrossRef]
18. E. Wohlgemuth, Y. Yoffe, I. Attia, et al., “A Field Trial of Multi-Homodyne Coherent Detection Over Multi-Core Fiber for Encryption and Steganography,” J. Lightwave Technol. 41(9), 2723–2735 (2023). [CrossRef]
19. J. Hall, M. Barbeau, and E. Kranakis, “Detection of transient in radio frequency fingerprinting using signal phase,” Wirel. Opt. Commun. pp. 13–18 (2003).
20. S. Li, M. Cheng, Y. Chen, et al., “Enhancing the Physical Layer Security of OFDM-PONs with Hardware Fingerprint Authentication: A Machine Learning Approach,” J. Lightwave Technol. 38(12), 3238–3245 (2020). [CrossRef]
21. W. Shao, D. Huang, H. Li, et al., “Machine Learning-Assisted Watermarking via Bias Control of Electro-Optical Modulator,” J. Lightwave Technol. 41(15), 4941–4948 (2023). [CrossRef]
22. Y. Chen, R. Xin, M. Cheng, et al., “Unveil the time delay signature of optical chaos systems with a convolutional neural network,” Opt. Express 28(10), 15221 (2020). [CrossRef]
23. Y. Wang, W. Shao, L. Zhong, et al., “Real-time stealth transmission via dither-based bias control,” in CLEO 2023, Technical Digest Series (Optica Publishing Group, 2023), paper JTu2A.87.
24. Yuanxiang Wang, Weidong Shao, Jinyang Wu, et al., “Real-time stealth optical transmission via dither-remodulation in a bias controller of a Mach-Zehnder modulator,” Opt. Express 31(12), 18773–18789 (2023). [CrossRef]
25. S. J. Savory, “Digital Coherent Optical Receivers: Algorithms and Subsystems,” IEEE J. Sel. Top. Quantum Electron. 16(5), 1164–1179 (2010). [CrossRef]
26. J. Tabares, S. Ghasemi, V. Polo, et al., “Simplified Carrier Recovery for Intradyne Optical PSK Receivers in udWDM-PON,” J. Lightwave Technol. 36(14), 2941–2947 (2018). [CrossRef]
27. Hai Zhu, Xue Chen, Weiqin Zhou, et al., “A modified CMA for blind equalization and phase recovery in optical coherent receivers,” 2009 Asia Communications and Photonics conference and Exhibition (ACP), Shanghai, China, 2009, pp. 1–6.
28. A. T. Schremer and C. L. Tang, “External-cavity semiconductor laser with 1000 GHz continuous piezoelectric tuning range,” IEEE Photonics Technol. Lett. 2(1), 3–5 (1990). [CrossRef]
29. Seb J. Savory, “Digital filters for coherent optical receivers,” Opt. Express 16(2), 804–817 (2008). [CrossRef]
30. A. J. Viterbi and A. M. Viterbi, “Nonlinear estimation of PSK-modulated carrier phase with application to burst digital transmission,” IEEE Trans. Inf. Theory 29(4), 543–551 (1983). [CrossRef]
