Detection Of Non Cooperative Underwater Acoustic Pulse Signal

Non-cooperative pulse-signal detection is a tough problem of passive detection.In the real ocean scene,non-stationary noise and strong line spectrum interferences(which exits in the sensor installation platform)are the main false alarm sources for the detection of the underwater acoustic pulse signal.Meanwhile,the weak pulse-signals can be masked by the strong line spectrum interference,which results in high false alarm probability and miss probability for beam power based detection algorithm and periodogram based detection algorithm.To solve the above problems,in this thesis,the time-frequency and time-spatial characteristics are integrated into the designs of algorithms,respectively.Then novel pulse-signal detection algorithms are proposed by combining the noise tracking and signal perception.The main contributions of this thesis can be list as follows:(1)The pulse-signal detection in frequency domain combining the noise spectrum tracking and signal spectrum perception.A noise spectrum tracking algorithm based on minimum smoothing statistics is designed to solve the problem of high false alarm caused by the nonstationary noise.A signal spectrum perceptron is designed to make up the weaknesses of periodgram for the signals with narrow pulse width and fast frequency changing.Based on the power spectrum of the data,the power spectrum component of the pulse signal can be extracted by the signal spectrum perceptron using MMSE(Minimum Mean Square Error)algorithm,and the interception of the pulse signal can be detected based on this component.In order to verify the effectiveness of the proposed algorithm,with different input signal to noise ratio(SNR)and signal to interference ratio(SIR),the gain of signal to noise ratio(SNR)and the gain of signal to interferer ratio(SIR)are analyzed,respectively,and the algorithm is compared with the periodogram based detection algorithm.The results show that the proposed algorithm outperforms the above periodogram based detection algorithm.(2)The pulse-signal detection in spatial domain based on noise beam power tracking and signal beam power perception.Firstly,the signal beam power perceptron is derived with the arbitrary array structure and generalized beamforming model.Furthermore,the noise beam power tracking algorithm is introduced,and the beam power observations and the noise beam power tracking results are used as the input of the signal beam power perceptron to extract the pulse signal components by MMSE method,this components can be utilized to perform the space detection of the pulse signal.In simulations,the proposed algorithm is evaluated by different input SNRs and input SIRs.And also the above simulation results show the effectiveness of the proposed algorithm.