Researchers have been making progress toward practical optical encryption systems that exploit laser dynamics. In two recent papers, researchers in France and Japan have both shown how feedback-loop-based systems can be best used, and have introduced a new kind of modulation scheme. The separate results demonstrate both the security and ease of decoding of one class of the emerging chaotic-shift keying (CSK) systems, and the applicability of the other to a wide range of systems.
Chaotic-shift keying is a way of using fluctuations in wavelength to both encode and hide a communications signal. In an optoelectronic implementation, a laser is configured so that its output fluctuates chaotically—that is, in a deterministic way that nevertheless looks random.To change from one bit value to another (1 to 0 or vice versa) the chaotic mechanism is altered slightly. Because the output is still chaotic, an eavesdropper should not see any change in the transmission. However, the receiver—which is matched to one of the two configurations (or both, for enhanced reception)—detects that the chaos is sometimes synchronized, sometimes not, allowing the signal to be extracted.FIGURE 1. In a chaotic-shift keying scheme, the output of a tunable distributed Bragg reflector laser is fed back via a nonlinear spectral filter and time delay, causing a chaotic shift in wavelength. The chaos is modulated by the input signal, which varies the delay between time intervals T0 (511.5 µs) and T1 (543.2 µs). This modulation is then picked up by a receiver matched to T0, and the error signal (where the signal and receiver are unsynchronized) shows the location of the nonmatching bits (ones).
In a collaboration between researchers at Georgia Tech Lorraine (Metz, France) and the Laboratoire d'Optique at the Université de Franche-Compté (Besançon, France), researchers have demonstrated that this system can achieve high-fidelity, high-security recordings. In their system, the researchers used a feedback loop with a nonlinear spectral filter to provide the mechanism for chaotic emission from distributed Bragg reflector lasers (see Fig. 1).1
The French team wanted to quantify the advantages of changing different parameters to shift from bit to bit. Initially they considered three options: the mean wavelength (around which the chaos was fluctuating), the bifurcation parameter related to the dynamics of the photodetector and laser diode, and the time delay implemented in the feedback loop.The first of these options is the least sophisticated and is known as masking: essentially a small wavelength modulation signal is hidden in a large chaotic one. This makes it both difficult to decode and somewhat insecure, since the modulation could be detected by an eavesdropper. Tweaking the bifurcation parameter was also inadequate for the encryption application: they showed that, for more than 50% of bits to be correctly recognized, the mismatch between the bifurcation parameter had to be less than 0.03. For better than 98% masking (high security), the mismatch had to be more than 0.035.
Changing the time delay in the feedback loop, however, was successful. Not only were the researchers able to show that the bit sequence was both well-hidden and easily received (see Fig. 2), they also made sure to demonstrate that the signal could not be decrypted by looking at the signal spectrum or through autocorrelation.
In another project, a collaboration between engineers at Takushoku and Keio Universities (both of Tokyo, Japan), researchers chose to use an acousto-optic modulator (AOM) to degrade CSK synchronization for the nonmatched states.2 Though less elegant in some ways, this approach has the advantage that it is not dependent on the type of chaotically emitting laser used and so could, for example, be used with semiconductor lasers that have very fast oscillations. For this scheme to work at high speeds, however, the transient time for synchronization—currently 10 times the AOM frequency—will have to be shortened.
Sunny Bains
REFERENCES
- J.-B. Cuenot et al., IEEE J. Quant. Elect. 37 (7), 849 (July 2001).
- A. Uchida et al., Opt. Lett. 26 (12), 866 (June 15, 2001).