Signal Processing Techniques for Non-Stationary Chaotic Spread Waveforms

When communicating with chaos, the greatest challenge over the past 30 years has been achieving a robust form of signal synchronization that can survive the dynamics of a practical communications channel [1,2,3]. The ideal chaotic signal characteristic in AWGN channels for satisfying security, channel capacity [4] and anti-jam performance [5] requirements is that of a bandlimited white Gaussian noiselike waveform, yet this causes the signal energy content to vary in time. Therefore, a secondary challenge in chaotic communication system design is to optimize the channel usage given a priori knowledge of the ‘deterministically random’ signal evolution. Recent techniques have been published for the efficient generation of digital chaotic sequences achieving this Gaussian noise-like characteristic [6], their extension to a family of secure waveforms [7], and their generalization to featureless chaotic spreading of arbitrary data constellations [8]. The signal processing flow of the prototype hardware systems incorporating these digital chaotic waveforms follow a similar construction to that of traditional direct sequence spread spectrum (DSSS) systems with key modifications for the non-stationary signal characteristic. The seemingly random waveform does still satisfy the definition for stationarity from the perspective of an external observer, yet its determinism at a synchronized receiver provides advance knowledge of the varying signal energy content, expanding the range of signal processing capabilities for short-term signal observations like soft symbol estimates. This paper discusses a series of DSSS signal processing modifications including baseband symbol controls, data modulation controls, peak-toaverage power ratio (PAPR) limitations, and selective noise cancellation that improve not only the channel utilization, but also the security of a chaotic communications signal. The net improvement is up to 3 dB for communications performance and 30 dB for detectability.