Statistical model of time-varying underwater acoustic channel in a shallow water using local scattering function

Abstract

The wide-sense stationary uncorrelated scattering assumption is not applicable to underwater acoustic channels. For the local time-frequency (TF) features of underwater acoustic channels, this paper studied a refined statistical model for the local TF root mean square (RMS) delay and Doppler shift at different frequencies based on the local scattering function (LSF). We presented a discrete form of the LSF with time- and frequency-domain sliding, which was then used to analyze the local characteristics of underwater acoustic channels and obtain the TF RMS delay and Doppler shift. Based on the measured channels under shallow-water conditions, the TF RMS delay and Doppler shift were characterized using the Gaussian mixture (GM) and Burr distributions, respectively. When deep or shallow fading occurs in a certain frequency range, the expectation, variance, and weighting factor of the GM distribution fluctuate notably, indicating complex multipath scattering. As the frequency increases, the scale parameter and first shape parameter of the Burr distribution reflect the gradual dispersion of the TF RMS Doppler shift.