Spectral Analysis of Encrypted Chaotic Signals Using Fast Fourier Transforms and Laboratory Spectral Analyzers

The use of acousto-optic chaos, as manifested via first-order feedback in an acousto-optic Bragg cell, in encrypting a message wave and subsequently recovering the message in the receiver using a chaotic heterodyne strategy, has been reported recently [1-3]. In examining the dynamical system analytically using computer simulation, (expected) modulated chaos waveforms are obtained within specified observation windows. Because of the relatively random nature inherent in chaos waveforms, it is essentially impossible to ascertain from the visual display of the chaotic wave whether a given message signal has in fact modulated the chaotic carrier . In fact, it has been observed from earlier work that by appropriately controlling the chaos parameters, one may hide the silhouette of the message from the envelope of the modulated chaos [1]. This was found to be especially true for low-frequency chaos (in the KHz range). For chaos in the mid-RF (up to 10s of MHz) range, it is seen that the silhouette is more difficult to suppress (even though this does not affect the robustness of the encryption). To adequately determine whether modulation has in fact occurred by passing the AC signal through the sound cell bias input, one needs to examine the spectral content of the chaos wave. In this paper, we discuss the results of such spectral analyses using two different approaches, (i) fast Fourier transforms applied to the displayed waveform; and (ii) transferring the intensity-vs-time data to an Excel spreadsheet, and then applying this information to a laboratory spectrum analyzer with adequate bandwidth. The results are mutually compared and interpreted in terms of encryption and decryption properties