Open Access
Add to BibSonomyAbstract
Temporal phase modulation of spread stealth signals is proposed and demonstrated to improve optical steganography transmission privacy. After phase modulation, the temporally spread stealth signal has a more complex spectral-phase-temporal relationship, such that the original temporal profile cannot be restored when only dispersion compensation is applied to the temporally spread stealth signals. Therefore, it increases the difficulty for the eavesdropper to detect and intercept the stealth channel that is hidden under a public transmission, even with a correct dispersion compensation device. The experimental results demonstrate the feasibility of this approach and display insignificant degradation in transmission performance, compared to the conventional stealth transmission without temporal phase modulation. The proposed system can also work without a clock transmission for signal synchronization. Our analysis and simulation results show that it is difficult for the adversary to detect the existence of the stealth transmission, or find the correct phase mask to recover the stealth signals.
©2010 Optical Society of America
1. Introduction
As an important component of photonic layer security, optical steganography enables transmission of a secrete data channel (or “stealth channel”), which can be hidden in plain sight in public optical networks [1], keeping observers unaware of its existence. Implementing stealth transmission has been considered for different optical networks, such as WDM and optical code-division multiple access (OCDMA) networks [2, 3]. These results indicate that stealth transmission is compatible with several common optical multiplexing techniques. Optical steganography is particularly suitable where the signals are not filtered or digitally regenerated at the nodes, as in the case of many of today’s passive optical networks (e.g., FIOS).
The principle of optical steganography is to temporally stretch stealth data pulses using high-dispersion elements. Due to the strong chromatic dispersion, the stealth data pulses are stretched and their peak amplitudes are reduced to a very low level such that the entire stealth signal is buried under the system noise in the optical network. Therefore, the stealth transmission is hidden under public signals which are simultaneously transmitted in the network. At the receiver, the stealth signal pulses are recovered using matched dispersion compensation. Previous research demonstrated that it is difficult to detect the existence of the stealth signal in the presence of public signals by observing either the temporal or the spectral profiles of the transmitted signals [2, 3], or even by analyzing their statistical features [4]. However, since the pulse spreading is realized linearly through chromatic dispersion, once the eavesdropper suspects the existence of a stealth channel, she can fully recover the stealth pulses using a tunable dispersion compensating device. Therefore the privacy of the stealth transmission is not ensured under such kind of attacks.
In this paper, we propose and experimentally demonstrate the use of temporal phase modulation of the spread stealth signal before sending it into the network, to improve the privacy of stealth transmission. As shown in Fig. 1 , after a temporal phase mask is imposed on the spread stealth signal, different portions of the spread pulse experience different phase shifts. To recover the stealth pulses, the corresponding phase recovery is required at the receiver along with the matched dispersion compensation, as shown in the stealth channel receiver in Fig. 1. Without phase recovery, as illustrated in the eavesdropper case in Fig. 1, the eavesdropper cannot fully recover the stealth signal even with the right dispersion compensation, and will obtain a worse signal than the stealth channel receiver does. Our experiment demonstrates the feasibility of the temporal phase modulation approach in the presence of public signal transmission, when the received stealth channel experiences only <0.1dB performance degradation resulting from the temporal phase modulation, compared with the approach without phase modulation. This approach also works without a clock transmission for signal synchronization, which further increases the adversary’s difficulty of detecting the stealth signal. Our simulation results and discussion prove that the adversary’s attacks can be inhibited by our proposed system.
2. Principle description
Temporal phase modulation of stretched pulses was first introduced to create spectral phase encoding (SPE) codes for OCDMA [5]. The principle of SPE code generation is shown in Fig. 2 . A wide-bandwidth optical pulse is temporally stretched by a dispersive element, which provides large chromatic dispersion. As a result, different spectral components are spread to different positions in the time domain. When a temporal phase mask (containing n chips) is applied to each stretched pulse, the phase modulation of each chip is imposed on each individual spectral component. An SPE code is created when the phased-modulated spread pulses are compressed back using the inverse dispersion of the stretcher. Figure 3 shows the experimental temporal profiles when different phase masks are imposed. The original pulse is distorted because of the phase modulation on individual spectral components. SPE codes can also be generated using spatial phase masks [6], yielding less interference in optical CDMA applications than temporal phase modulation [5]. However, temporal SPE code generation has the advantage of being more rapidly reconfigurable and is more compatible with fiber optic technology.
Fig. 3 Temporal profiles of various SPE codes generated using different phase masks (16 chips in each bit). 0 means 0 phase shift, and 1 means a π phase shift. (a) no phase mask; (b) phase mask = 1010101010101010; (c) phase mask = 0100101101010111; (d) phase mask = 1110100101000110; (e) phase mask = 1001101001011010.
In our stealth signal transmission, we propose applying a temporal phase mask to the spread pulses after pulse stretching, in order to improve its transmission privacy. After phase modulation, the stealth signals are combined with public transmission signals and are sent through an optical network. If the eavesdropper taps the signals in the network and suspects the existence of the stealth signal, both the correct dispersion compensation and conjugate phase modulation with matched alignment must be applied to recover the stealth pulses. By applying only the dispersion compensation without correct phase recovery, the adversary will obtain the stealth signal with SPE code profiles shown in Fig. 3 (b)-(e). Since the stealth channel usually has lower power than the public channel, the incorrectly received stealth signal is easily mixed with the public channels, making it difficult for the eavesdropper to detect. Note that our applied phase mask aims at randomizing the temporal profiles of the stealth pulses, rather than creating SPE codes. Therefore the code orthogonality is not considered, and the phase shift at each chip theoretically can be any value between 0 and 2π. In this case, a large number of possible phase masks can be used, and it is difficult for the adversary to find the correct phase mask to recover the stealth signal.
3. Experimental setup and results
Chirped fiber Bragg gratings (FBG) are compact devices which provide large chromatic dispersion and are suitable for stealth transmission [7]. Chirped FBGs can have customized dispersion features other than the typical linear chromatic dispersion of single mode fibers, making it more complicated to recover the spread stealth signal through dispersion compensation. Figure 4 shows the dispersion measurement of the chirped FBG in our experiment at its transmission and reflection ports at different wavelengths.
The schematic of our experimental setup is shown in Fig. 5 . A mode-locked laser (MLL) generates a pulse train at 2.48832 GHz. The repetition rate is down-converted to 622.08 MHz before the pulse train is launched into a piece of dispersion decreasing fiber (DDF) to generate the initial wide-band stealth pulse train. The center wavelength of the pulses is 1554.6nm, and the 3dB spectral width is 1.3 nm. The stealth pulse train is modulated at the intensity modulator by the stealth data. To provide enough dispersion for achieving the required pulse spreading, a piece of dispersion compensating fiber (DCF) and two chirped FBGs with the same sign of dispersion are cascaded as the stretcher, producing a total dispersion of −840ps/nm. In principle, a single piece of chirped FBG can be made to provide this amount of dispersion.
Fig. 5 Schematic diagram of the stealth transmission experiment setup. MLL: mode-locked laser; DDF: dispersion decreasing fiber; IM: intensity modulator; PM: phase modulator; PC: polarization controller; D: tunable delay line; EDFA: erbium-doped fiber amplifier.
The stealth signal profiles before and after spreading are shown in Fig. 6 (a) and (b) , respectively. The phase modulator after the stretcher operates at 9.95328 Gs/s. Hence the phase mask pattern on each stretched pulse consists of 16 chips, each of the chips being either 0 or π. We assume that the pulse with 1.3nm bandwidth is spread to the entire bit interval, and the spectrum resolution of each chip is about 10GHz.
Fig. 6 Temporal profiles of the original stealth pulses (a), and spread pulses after stretching and amplification (b)
Another copy of the MLL pulse train at 2.48832 GHz is used to generate a pulse train for public transmission. The pulse train is modulated by an intensity modulator using a pseudo-random binary sequence (PRBS) to emulate a public data stream. The erbium-doped fiber amplifier (EDFA) is used to amplify the public channel and to generate the system noise as found in real optical networks. The power of the stealth channel and the public channel is −18.5dBm and 1.6dBm, respectively.
Figure 7 displays the temporal profiles of the public channel and stealth channel at different points in the experiment. Figure 7(a) displays the original stealth pulses and Fig. 7(b) shows the stealth signal after stretching and phase modulation. To receive the stealth data, another phase modulator is used to restore the phase of each chip, using the conjugate phase mask as at the transmitter. A tunable delay line is used to align the spread pulse with correct position of the phase mask. A polarization controller and a polarizer are added before the phase modulator to ensure the best phase recovery. After the phase recovery, two cascaded chirped FBGs and a 18-km single-mode fiber (SMF) are used to compensate the dispersion at the transmitter. Figure 7(c) shows the eye diagram of the recovered stealth signal, where the public data pulses are highly dispersed due to the compressor. The public channel signal is received using a conventional receiver. Figure 7(d) and 7(e) are the eye diagrams of the public transmission signals with and without the stealth channel, from which it is difficult to discern the existence of stealth channel. Figure 7(f) is the obtained recovered stealth signal with only the appropriate dispersion compensation. This indicates that without phase recovery, the compressed stealth signals are distorted and mixed with the public transmission signals.
Fig. 7 Temporal profiles (a) original stealth pulses; (b) spread stealth signal after phase modulation; (c) recovered stealth signal; (d) eye diagram of modulated public data (e) eye diagram of public data and stealth data. (f) recovered stealth signals with only dispersion compensation.
Figure 8 shows the optical spectra of the public signal, the stealth signal, and the combined signal, respectively. The spectra of the public signal and the combined signal have a insignificant difference, and will become indistinguishable in real optical networks when the public signal power varies with users randomly accessing of the networks.
We measured the bit error rates (BER) of both the public channel and the stealth channel transmission, as shown in Fig. 9(a) and (b) , respectively. The power penalty of the public channel induced by the presence of stealth channel is only 0.1dB. For stealth transmission, the use of phase modulation for privacy enhancement results in <0.1 dB power penalty to the stealth signal. Compared to back-to-back stealth transmission without public pulses and pulse spreading and compression, 8.2dB power penalty is observed.
4. Discussion
In our experimental demonstration, the stealth pulse is received using a synchronous clock from the transmitter. However, this synchronous operation would not be practical if the transmitter and recevier were not co-located. In this case, the stealth data can still be received asynchronously if a commercial clock and data recovery module (CDR) is utilized. In order to have the CDR succesfully receive the recovered stealth pulses, the spread public signal needs to be removed, which can be achieved by adding an all-optical thresholder [8]. A passive all-optical thresholder is able to suppress the low peak-amplitude public signal, and pass the stealth signal with high peak-amplitude, as shown in Fig. 10 .
Fig. 10 Function demonstration of an all-optical thresholder: (a) thresholder inputs; (b) thresholder outputs
The schematic of the receiver system for asynchronous stealth transmission is shown in Fig. 11 . An all-optical thresholder and a CDR are added after the phase and dispersation receovery. The delay line is tuned to align the conjugate phase mask in a range of one bit interval. When the conjugate phase mask matches the original phase modulation, high-peak stealth pulses will be generated, and be recognized by the CDR. A phase mask misalignment will result in random-profile optical signals (shown in Fig. 12 ), which will be suppressed after passing through the thresholder. The stealth data therefore cannot be detected in this misalignment case. In practice, this alignment can be implemented in an automatic fashion by continously adjusting the optical delay until the CDR successfully recovers a training sequence.
Fig. 11 Schematic diagram of the stealth receiver module without a synchronous clock. PM: phase modulator; PC: polarization controller; D: tunable delay line
Fig. 12 The recovered stealth signal waveforms (a) perfectly aligned phase mask (b) misaligned phase mask
Now we discuss our system’s robustness against the adversary’s attacks. We first look at the adversary’s capability of detecting the existence of the stealth channel, without using a conjugate phase mask. In Fig. 7(f), we assumes that the adversary applies correct dispersion compensation, because the adversary can keep tuning the dispersion to get the best stealth signal waveform. However, this dispersion tuning will be difficult and unavailable under the following conditions. First, further randomization to stealth signal’s temporal profile can be applied, by increasing the chip number of the phase mask (e.g., increasing phase modulation speed or performing more pulse spreading) and using arbitrary phase shift at each chip. Figure 13 shows the simulation results of the signal’s temporal profiles using phase masks with different chip numbers and arbitrary phase shifts. The stealth signal with a 32-chip phase mask generates lower peak amplitude (10 percent of the original pulse peak) and has more temporal spreading, compared to that of a 16-chip phase mask. Secondly, the power of the stealth channel should be operated at a even lower level, so that the stealth signal is completely mixed with the public signal and cannot be identified. For example, when the public signal has a power 22dB larger than the stealth signal, the amplitude of the spread public signal can be up to 20 precent of the recovered stealth pulse’s peak amplitude, after the stealth signal recovery. The stealth channel may not have an optimal reception performance, but forward error correction (FEC) can be utilized to improve it. Furthermore, since there is no clock transmission, the adversary is not able to observe the stealth signal’s waveform without the synchronized clock. In summary, it is difficult for the adversary to detect the stealth channel by only tuning the dispersion.
Fig. 13 Simulation results of the stealth signals’ waveforms with different chip numbers and arbitrary phase shifts. The simulation is based on a 1.3nm-bandwidth MLL at a rate of 2.5Gs/s. The 16-chip phase mask uses pattern (π/8)* [1,-2,6, −8, 1,-3,1,0, 1,-2,5,-4, 3,-4,1,-7]. The 32-chip phase mask uses pattern (π/8)*[5,-6,-3,0,7,-4,4,-2,1,5,-2,6,-3,1,-3,1,6, 1,-2,5, −4, 3,-4,1,-7,-5,2,-2,1,-2,6,-6]. Figure (b) is the enlarged version of the rectangular part in Figure (a).
Secondly, we focus on the adversary’s ability to recover the stealth signal when the stealth channel’s existence is known. When the adversary also uses temporal phase modulation to recover the stealth pulses, both the dispersion compensation and the conjugate phase mask are required to be set correctly. The phase mask’s temporal alignment is also an issue for the adversary. Other phase modulation approaches, such as spatial phase modulation and spectral phase modulation, can also be used to recover the stealth pulses. With spatial phase modulation and spectral phase modulation, the phase recovery should be applied after the dispersion compensation, without the need of temporal alignment. In this case the adversary only needs to find the right dispersion compensation and the conjugate phase mask. We assume that the adversary finds the matched dispersion compensation (though this is not easy), and concentrate the stealth signal recovery with a partially matched phase mask. Figure 14 shows the simulation results of the recovered stealth signals’ waveforms with two partially matched phase masks. A 32-chip phase mask with 8 matched chips only results in a signal with peak amplitude similar to the original stealth signal. A phase mask with 12 matched chips yields the stealth pulse with doubled peak amplitude, but may still not be significantly large, considering the existence of public signal and the function of optical thresholder. Here we simply treat 12-chip matched phase mask as a detectable boundary. Since we have 16 possible phase shift values at each chip, there are 1612 (~1014) possible combinations. And if the dispersion is not correctly compensated, the peak amplitude will be further distorted. In summary, we can conclude that our approach is robust against the adversary’s attack.
Fig. 14 Simulation results of the stealth signals’ waveforms when different conjugate phase masks are applied The original 32-chip phase mask uses pattern (π/8)* [-5,0,-2,1, 5,-6,4,-2, 7,-6,3,-2, 1,5,-2,-6,3,-2,6,-8, 1,-8,1,0, 5,-2,5,-4, 3,-4,5,-2]; Figure (b) is the enlarged version of the rectangular part in Figure (a).
5. Conclusion
We propose and demonstrate imposing temporal phase modulation on temporally spread stealth signals to improve the transmission privacy. The phase mask added on the stealth signal changes the spectral-phase-temporal relationship of the pulse. Thus, when dispersion compensation is applied, the temporal profile of the compressed stealth signal is changed and it is difficult to detect when public transmission is present. Our approach is also applicable without a clock for synchronization, by use of an all-optical thresholder and an CDR module. Our simulation results show that faster phase modulation and more complex phase mask patterns can be applied to further modify the characteristics of the stealth signal and increase the adversary’s difficulty of finding and intercepting the stealth signal, inhibiting the unauthorized detection of the stealth signal, and therefore increasing its privacy. Insignificant signal quality degradation is induced by temporal phase modulation, compared to conventional stealth transmission just using the temporal spreading.
Acknowledgment
This work was supported in part by the U.S. Defense Advance Research Projects Agency under Grant MDA972-03-1-0006 and SSC Pacific Grant N66001-07-1-2010. The authors would like to thank R. Menendez of Telcordia Technologies for his help in the simulation.
References and links
1. B. B. Wu and E. E. Narimanov, “A method for secure communications over a public fiber-optical network,” Opt. Express 14(9), 3738–3751 (2006). [CrossRef] [PubMed]
2. Y.-K. Huang, B. Wu, I. Glesk, E. E. Narimanov, T. Wang, and P. R. Prucnal, “Combining cryptographic and steganographic security with self-wrapped optical code division multiplexing techniques,” Electron. Lett. 43(25), 1449–1451 (2007). [CrossRef]
3. K. Kravtsov, B. Wu, I. Glesk, P. R. Prucnal, and E. Narimanov, “Stealth transmission over a WDM network with detection based on an all-optical thresholder”, IEEE/LEOS Annual Meet., 480–481 (2007).
4. B. Wu, P. R. Prucnal, and E. E. Narimanov, “Secure Transmission Over an Existing Public WDM Lightwave Network,” IEEE Photon. Technol. Lett. 18(17), 1870–1872 (2006). [CrossRef]
5. X. Wang and N. Wada, “Spectral phase encoding of ultra-short optical pulse in time domain for OCDMA application,” Opt. Express 15(12), 7319–7326 (2007). [CrossRef] [PubMed]
6. R. P. Scott, W. Cong, K. Li, V. J. Hernandez, B. H. Kolner, J. P. Heritage, and S. J. Ben Yoo, “Demonstration of an error-free 4×10 Gb/s multiuser SPECTS O-CDMA network testbed,” IEEE Photon. Technol. Lett. 16(9), 2186–2188 (2004). [CrossRef]
7. M. P. Fok and P. R. Prucnal, “A Compact and Low-Latency Scheme for Optical Steganography Using Chirped Fiber Bragg Gratings,” Electron. Lett. 45(3), 179–180 (2009). [CrossRef]
8. K. Kravtsov, P. R. Prucnal, and M. M. Bubnov, “Simple nonlinear interferometer-based all-optical thresholder and its applications for optical CDMA,” Opt. Express 15(20), 13114–13122 (2007). [CrossRef] [PubMed]
