Research On Signal Detection And Bearing Estimation Of Underwater Sound Source

Weak signal detection in the sea ambient noise background is an interesting research field of underwater signal processing, and has wide applications in the aspects of underwater target's detection and recognition, communication, remote control, telemetry, and ocean exploration, etc. Available signal detection performance of low signal to noise ratio means enlarged reception distance and improved localization accuracy to targets for the acoustic fuze of underwater weapons, or results in need of less signal source power for underwater communication system which will keep the communication more secret.Based on the rapid development of signal detection and processing theories in the last 20 years, the applications of some modern signal processing methods in underwater signal detection and processing are studied with emphasis in this dissertation, including harmonic frequency estimation based on modern cross-correlation spectrum, higher order statistics, chaotic oscillator theory, cyclostationary statistics, LFM signal detection based on Laguerre transform, and multi-sinusoids detection based on frequency warping, etc. On the other hand, the realization and improvement of some traditional signal detection and bearing estimation methods are also studied, such as searching method of time coherent averaging, polarity correlation, symmetrical correlation, adaptive filtering, some traditional time delay estimation methods, and adaptive FIR time delay estimation based on the least square principle, etc.The main contributions of this dissertation are as follows:1. The relation between the processing gain of time averaging and the departure of truncation period from integral times of signal period is studied. Based on this research, a searching method of time averaging is proposed, which may be applied to search line spectrum components in a certain frequency span. The SNR lower bound for effective detection of this method reaches - 36 dB under the simulation conditions in this dissertation.2. The principles of harmonic frequency estimation are studied by using Levinson method, Pisarenko method and MUSIC method, all of which being based on cross spectrum and cross higher order cumulant. After comparing the detection performances of the methods mentioned above, a new method to lower the order of the extended characteristic equation is proposed. With this method, the false peaks of spectral estimation arising from SVD (singular value decomposition) extended characteristic equation will be cleared to attain precise frequency estimation.3. A high precision FIR time delay estimation method based on the least square principle and adaptive filter theory is proposed, then the computer simulation analysis and anechoic water pool experiment are performed. Based on this method, bearing estimation of the stern part of a ship is studied. The results of simulation and experiment show that this time delay estimation method is effective.4. A new method to detect the LFM signal based on the Laguerre Transform is proposed. By properly setting the Laguerre coefficients, the Laguerre transform compensates for the frequency variation of a LFM signal, and reduces the signal to an almost perfectly periodic one. Since all the energy is concentrated into a single frequency bin, the LFM signal buried in the sea ambient noise can be detected effectively even under a very low SNR of about -30dB.5. A new method to detect an band-limited multi-sinusoids based on frequency warping is proposed, which coherently superimposes all the sinusoids onto a single line below the spectrum lines, thus the line will stand out of high level ambient noise. This method plays an important role in the detection of communication signal such as multi-frequency modulation signal and time-frequency code signal.