Statistical Analyses of Measured Forward-Looking Sonar Echo Data in a Shallow Water Environment

The performance of underwater target detection significantly depends on the statistical distribution of reverberation and knowledge of data correlation for active sonar. To reasonably design constant false alarm rate (CFAR) detectors that exploit knowledge of reverberation characteristics, a preliminary statistical analysis of active sonar echo data is necessary. In this study, we analyze the amplitude statistics of the measured forward-looking sonar echo data after applying the beamforming processing with a variety of distributions including log-normal, Weibull, Rayleigh, K, generalized Pareto (GP) and inverse Gaussian distributions. The results show that for the high-frequency sonar working in shallow water, both the K and the GP distributions can provide a good data fitting, while the GP distribution performs better especially in the tail region of the distribution. At last, the spatio-temporal correlation of data analysis and concluding remarks complete this paper.