Research on the performance of page test method under the hypothesis of typical non-Gaussian distribution

The page test is an effective method for transient signal detection according to the change of probability distribution function (PDF) of the detected sequences when the sonar target signal emerges or disappears, and the PDF usually has non-Gaussian characteristics. A typical non-Gaussian PDF model, the t distribution, is taken to replace the traditional Gaussian PDF model in this paper. Under the hypothesis of t distribution, the expression of page test statistic and the numerical calculation method for the updated PDF are derived. The parameters essential for detection including threshold and degree of freedom are calculated by FFT approach. Simulated falling water signal, ship acceleration signal and the signal obtained in the anechoic tank are used to verify the algorithm of page test with t distribution. The results show that compared with the algorithm with traditional Gaussian PDF, the new algorithm proposed in this paper can detect underwater transient signals more effectively, respond to signal change more quickly and can better suppress background interference.