Research On DOA Estimation Algorithm Based On Non-circular Property Of Complex Numbers

At present,one of the focal points of DOA estimation is to improve the performance at low SNR.Most of the research on using the non-circular property to improve the performance of DOA estimation make use of the non-circular property of signals in the phase space.The phase of such signals is usually modulated by special methods,such as BPSK signal.However,some signals do not have non-circular property in the phase space,such as single-frequency CW signal.But the additive non-isotropic observation noise contained in the CW observed signal is non-circular in the azimuth space after being received and recombined by the vector hydrophone.In order to expand the research methods of the non-circular property,this paper studies how to improve the performance of DOA estimation by using the property that the pseudo-covariance of non-circular complex observation noise in azimuth space is not zero.At the same time,this paper also improves the conventional algorithms that make use of the non-circular property in the phase space of BPSK.This paper mainly contains the following two parts:(1)For the complex signal with non-circular property in the phase space,such as the BPSK signal,this paper puts forward a recombinational NC-MUSIC algorithm using pseudocovariance of complex signal received by the vector hydrophone array.In order to solve the problem of different channel noise power,this paper has carried on the recombination to signal model,in addition,on the basis of the recombinational model,using augmented received signal covariance matrix to construct the criterion of estimating the number of source.Simulation results show that the proposed algorithm can effectively reduce the Mean Absolute Error of the DOA estimation results at low SNR,improve the resolution and the ability to identify multiple sources,and significantly improve the probability of successfully estimating the number of sources.(2)For the CW signal,it does not have non-circular properties in the phase space.In order to study the non-circular property,this paper considers the non-circular property in the azimuth space of non-isotropic observation noise,an Adaptive Augmented Complex Extended Kalman Filter algorithm for single vector hydrophone has been proposed.At the same time,the Sage-Husa measurement noise adaptive algorithm is extended to the augmented form,combined with multi-model Kalman algorithm,and has achieved the joint adaptive of the augmented observation and state noise covariance matrix.Simulation and experimental data processing show that this algorithm has higher accuracy and better stability compared with the mean sound intensity algorithm based on single vector hydrophone for scenes with poor observation conditions,low SNR and insufficient prior knowledge of noise.