Direction of arrival estimation based on high-order cumulant by sparse reconstruction of underwater acoustic signals

The high-order cumulant has the characteristics of Gaussian noise suppression and array elements expansion,introduced it into the Direction Of Arrival(DOA) estimation of the underwater acoustic signals,an algorithm of DOA estimation based on high-order cumulant by off-grid sparse Bayesian learning reconstruction is proposed.This method uses the natural blindness of the high-order cumulant to Gaussian noise,and calculates the fourth-order cumulant of the array signal to filter out the Gaussian noise,then the array elements doubles the original structure.The selection matrix is constructed to eliminate the redundant items in the fourth-order cumulant,the array elements can be expanded again,and the new observation model obtained has better statistical performance.Then,using the sparsity of the spatial domain,the off-grid sparse representation model under the fourth-order cumulant is derived,the maximum posterior probability of the source signal is calculated by Bayesian learning,and the target azimuth estimation is realized.Numerical simulation and sea measured data show that,this method can achieve a resolution probability of more than 95% when the azimuth interval between adjacent sound sources is 40,the root mean square error of the target azimuth estimation is within 1° when the signal-to-noise ratio is greater than-5 dB.This method can significantly suppress background noise interference,it can accurately and robustly estimate the azimuth of the underwater acoustic target even under the condition of densely distributed multiple sound sources.