Detection of ship echo signals in reverberation background based on sample entropy and multiscale sample entropy

Detecting target signals in a reverberant underwater environment has long been a prominent and enduring challenge in the field of underwater acoustic signal processing. Traditional frequency-based methods are only effective when dealing with high signal-to-reverberation ratios (SRR) and significant Doppler shifts. This paper introduces the sample entropy (SampEn) and multiscale sample entropy (MSE) algorithms to explore the complexity differences between target echoes and reverberation. By conducting simulations under various SRR and Doppler shifts, we analyze the performance of traditional frequency-energy-based detection method, as well as entropy-based detection methods. The results show that in situations with a minor Doppler shift but a high SRR, frequency-based methods fail, yet the SampEn remains effective in detecting the target echo. Moreover, even when the Doppler shift is insignificant and SRR decreases to 0 dB, MSE can still detect the target echo.