Underwater Preamble Detection Neural Network Based on Second-Order Synchrosqueezing Transform

Preamble detection is critical in underwater acoustic systems because it directly affects communication reliability and operational coexistence. However, the underwater environment poses significant challenges to preamble detection due to noise interference, multipath propagation, severe signal attenuation, and the Doppler effect. This paper investigates preamble detection under interference that overlaps with the desired signal in both the time and frequency domains. We incorporate the second-order synchrosqueezing transform (2SST) into an end-to-end complex-valued synchrosqueezed wavelet neural network, aiming to enhance the detection performance of the network. Simulational results show that the proposed network exhibits superior performance in distinguishing between preamble signals and interference, especially in scenarios where interference overlaps with the preamble signal in both the time and frequency domains.