Research On DOA Estimation Method Under The Background Of Impulse Noise Based On Coprime Array

The estimation accuracy of DOA estimation method based on traditional uniform arrays is limited by the array aperture,and a large number of array elements are required to improve the estimation performance,while the coprime arrays with sparse characteristics can obtain a larger array aperture and higher estimation accuracy with the same number of array elements and can achieve orientation estimation under the condition that the number of source is greater than the number of physical array elements,which has received extensive attention from scholars at home and abroad.However,most algorithms assume that the environmental noise obeys the Gaussian distribution.In the actual marine environment,the noise sometimes has strong impulse characteristics in the time domain.At this time,the high order statistics of the data received by the array do not exist.The performance of DOA estimation methods mentioned above are severely degraded.Therefore,it is of great significance to study the DOA estimation method under the background of impulse noise based on coprime array.This paper first studies the DOA estimation method based on the coprime array,analyzes two generalized coprime array structures and the reason why the coprime array can obtains lots of degrees of freedom,which is far greater than the number of physical sensors.Compared the performance of MUSIC methods based on two kinds of structures,the advantages of the two structures are analyzed.For the virtual array received signal model,this paper studies the DOA estimation methods in the context of coherent sources: spatial smoothing method,Toeplitz matrix reconstruction method and vector singular value decomposition method.These methods all achieve accurate estimation of the target azimuth at the expense of the virtual array aperture loss.To reduce virtual array aperture loss,DOA estimation methods based on sparse signal reconstruction are studied: SMV-OMP and SMV-BP.Because the virtual array receiving signal model is a single snapshot model,the estimation accuracy of SMV-OMP and SMV-BP in the case of multiple targets is not high.In order to improve the estimation accuracy in the case of multiple targets,the virtual array model under the single snapshot is converted to a virtual array model captured by multiple snapshots,and And use the 1l-SVD to achieve DOA estimation.The simulation results show that the 1l-SVD based on spatial smoothing still has good estimation performance in the multi-source background,and the estimation performance of this method is better than that of the SS-MUSIC method in the case of low signal-to-noise ratio and small snapshots.Under certain conditions,this method can resolve the azimuths of two sources with an angular interval of 3?.Aiming at the performance degradation of DOA estimation method based on Gaussian white noise model under the background of impulse noise,combined with the principle of increasing the degree of freedom of the coprime array,the impulse noise is processed from the perspectives of the array receiving data matrix preprocessing and the fractional low order statistics.The DOA estimation methods under the Gaussian white noise model are extended to the background of impulse noise.The simulation results show that the methods in this paper can still achieve the target orientation estimation under strong impulse conditions.Finally,considering that the underwater acoustic signals are mostly broadband signals,in order to improve the practicability of the methods in the marine environment,the DOA estimation methods under the narrowband signal model mentioned above are extended to the broadband signal model,and experimental data processing verifies the effectiveness of these methods.