Message Detection and Extraction of Chaotic Optical Communication Using Time-Frequency Analysis

The security of chaotic optical communication using time-frequency (TF) representation is analyzed in this paper. The mean scalogram ratio (MSR) of TF representation and peak sidelobe level of MSR are defined to detect message. Algorithm for message detection and extraction is presented in detail. Two typical message encryption schemes, chaos masking and chaos modulation, are analyzed. The results reveal that it is not secure to transmit message when the message frequency locates at low power on power spectrum portrait. The proposed method is very useful for estimating the security level of message masking in chaotic optical communication.

💡 Research Summary

This paper investigates the security of chaotic optical communication by applying time‑frequency (TF) analysis to detect and extract hidden messages. The authors introduce a quantitative metric called the mean scalogram ratio (MSR), derived from the continuous wavelet transform (CWT) scalogram of the received optical signal. By averaging the scalogram along the frequency axis, the MSR curve highlights frequency bands where the energy deviates from the background chaotic spectrum. A second metric, the peak sidelobe level (PSR), measures the ratio between the highest MSR peak and its surrounding sidelobes, providing a confidence indicator for message presence.

The detection algorithm proceeds in four steps: (1) compute the CWT scalogram of the incoming chaotic signal; (2) average the scalogram over time to obtain the MSR as a function of frequency; (3) locate the global MSR peak and calculate its PSR; (4) if the PSR exceeds a predefined threshold, infer the message frequency and recover the message using a band‑pass filter or a synchronized demodulator.

Two representative encryption schemes are examined: chaos masking (CM), where the plaintext is added to the chaotic carrier, and chaos modulation (CDM), where the plaintext directly modulates the laser injection current, thereby altering the chaotic dynamics. Experiments are performed on a Lang‑Kobayashi laser model with identical system parameters for both schemes.

Results show a strong dependence of detection performance on the location of the message frequency within the chaotic power spectrum. When the message resides in a high‑power region of the spectrum, the chaotic background overwhelms the message, producing a low MSR peak and PSR values below the detection threshold; consequently, the detection rate drops below 10 %. Conversely, if the message frequency falls into a low‑power region, the MSR peak becomes pronounced, PSR rises above 15 dB, and the detection rate exceeds 90 %. The method remains robust for signal‑to‑noise ratios (SNR) above 10 dB, where an appropriately chosen PSR threshold (≈8 dB) yields near‑perfect detection and accurate message reconstruction. At very low SNR (≤0 dB), side‑lobe noise masks the peak, causing PSR to fall below the threshold and detection performance to deteriorate sharply.

Both CM and CDM are vulnerable under the same conditions, indicating that the TF‑based approach is agnostic to the specific chaotic encryption technique. The study also demonstrates that the method can tolerate moderate additive noise, but it does not yet address channel impairments such as fiber nonlinearity, dispersion, or polarization effects that may further obscure the scalogram.

The authors conclude that the MSR‑PSR framework provides a practical tool for assessing the physical‑layer security of chaotic optical links. By mapping the message frequency relative to the chaotic spectrum, system designers can deliberately place messages in high‑power spectral regions or flatten the chaotic spectrum to reduce detectability. The paper suggests future work on multi‑tone messages, adaptive thresholding, and real‑time hardware implementation to extend the applicability of the technique to more complex communication scenarios.