On non-Gaussian characterization of shipping traffic underwater noise

This paper is originated from the desire to obtain classification and characterization of a vast category of signals which have been ignored in conventional classical applications, like the ones concerning shipping traffic underwater noise. Current interest in the field of communications, the falling price of personal computers (based on 80486 or pentium), and the availability of extremely high level languages, have lead the authors to develop this work. The paper is documented in several stages: initially the non parametric and parametric approach, for the nongaussian and nonlinear signal processing are described, and proposed by system which divides interferences sources into classes A, B and C is introduced. In order to achieve to the main goal of the present paper various mathematical techniques have been adopted such as example the bispectrum approach (H.O.S. Theory), and Gabor expansion. After the nonlinearity identification due to transient signals, the processing system is applied to class B interferences. Although the heavy computational complexity in various stages, i.e. the calculation of the coefficients used for the Gabor processing and the routine for the bispectrum estimation, the results confirm the wide extendibility of the D. Middleton approach. This work has been developed in the contest of the MAST-I-SNECOW Project ( Shipping Noise Evaluation in Coastal Waters) of the European Community Marine Science and Technology Program. >