Abstract:In order to solve the problem of communication signal recognition in underwater acoustic channel, the modulation recognition technology of underwater acoustic CW, LFM, 2FSK, 4FSK, BPSK, QPSK, DSSS-BPSK and OFDM signals is studied, aiming to propose a modulation recognition method suitable for these eight signals. By analyzing time-frequency characteristics, quadratic spectrum characteristics and autocorrelation characteristics of signals, and using ResNet network based on migration learning and linear support vector machine for classification, a multi feature joint underwater acoustic communication signal modulation recognition method is finally obtained. The simulation shows that when the SNR is greater than 9dB in the Qiandao Lake multipath channel, the recognition rate of the eight signals studied is above 99%. Finally, the feasibility of this method is further verified through sea trial, and the recognition rate of sea trial data is above 93%. The simulation and sea trial results show that the proposed method is suitable for underwater acoustic channels.