ASA 126th Meeting Denver 1993 October 4-8

3aEA6. Application of nonlinear dynamics for noise reduction in measured signatures containing broadband information.

Ronald Hughes

Radix Systems, Inc., 6 Taft Ct., Rockville, MD 20850

Nonlinear dynamics has been of great academic interest in recent years. Its concepts have been employed as a foundation for modeling nonlinear processes in diverse fields extending from economic trends to medical dysfunctions. Although the developments of nonlinear dynamics concepts in acoustics are sparse, their application in acoustics has advantages over linear techniques in extending broadband signatures in high-noise environments. Linear signal analysis approaches are significantly limited when the noise environment is nonstationary or when the signal duration is short. Furthermore, the application of filters alters the information content of the original broadband signals. Nonlinear dynamics methods do not suffer these limitations. Two such methods will be described for effecting noise reduction where the signal-to-noise ratio is zero or negative and where there is no a priori knowledge of either the signal or the noise. Signals used for demonstration include sinusoids and chaotic sequences; additive noise includes uniform and Gaussian random noise and noise that produces a power spectrum equivalent to that of the ``clean'' sequences. Noise reduction of 15--18 dB using nonlinear dynamics will be shown, and fidelity reproduction of the original signal will be demonstrated.