Study On The Direction Of Arrival (DOA) Estimation Of Wideband Coherent Signals

Direction-Of-Arrival(DOA)estimation is one of key techniques in mobile communication and digital audio technologies.It has very large application fields such as radar, sonar,mo-bile communication, Multiple-Input Multiple-Output (MIMO)system,sensor networks, sound source tracking,and so on,MUltiple Signal Classification(MUSIC) and Estimationof Signal Parameters via Rotational Invariance Techniques(ESPRIT) are two of the representative DOA estimationmethods which are often used and so-called"superresolution"DOA estimation algorithms.The DOA estimation algorithm for wideband signals is a hot research subject at home and abroad. Because wideband signal processing technology can take advantage of more extensive information of targes, compared with narrowband technology,it has the superiority and its breakthrough research in underwater targeting, battlefield target identification and radar tracking of low elevation has an important significance.DOA estimation methods for narrowband signals, narrowband coherent signals and wideband signals are studied deeply in this paper.It presents some different array manifolds and makes comparison of the patten characteristics of ULA,UCA and IUCA. It mainly makes lots of simulation and experimental performance analysis for MUSIC algorithm which is based on signal subspace.The experimental results show that the more the number of array element, the greater the number of snapshots, the higher signal to noise ratio, the better the estimation performance of MUSIC algorithm; for the three array models,when other conditions are same, the estimation performance based on ULA is best, IUCA is followed by, and UCA is less.For narrowband coherent signals, Some compensation methods for reducing signal correlation have been studied like the Spatial Smoothing(SS), and Forward-Backward SS(FB-SS). as they are often employed for correlation reduction.But their ability of decorrelation is obained by sacrificing the effective number of arrays, there is a certain degree of loss of the array aperture. and at the same time, they tend to decrease the effective number of array elements that may degrade DOA estimation accuracy.This paper presents two efficient DOA estimation algorithms for narrowband coherent signals—A modified method based on the MMUSIC and a reconstruction algorithm based on the Toeplitz matrix.The former constructs a new matrix by the direct data snapshots and its conjugate inversion and gets the singular value decomposition of covariance matrix of the original data and the new matrix ,then the DOA of coherent signals are estimated by the conventional MUSIC without aperture loss.The simulation results show that the improved algorithm has better estimation performance at the low Signal to Noise Ratio(SNR) than spatial smoothing algorithm;For the latter,an upper triangle Toeplitz matrix is designed by rearranging the correlation functions of snapshot vectors. The covariance matrix of orignal data is substituted by the average of the Toeplitz matrix and its conjugate inversion. And the DOA of coherent signals are estimated by the conventional MUSIC algorithm based on the new matrix. The algorithm doesn't need array smoothing without aperture loss. The simulation result proves that the maximum number of DOA estimation for coherent signals can be M-1,and its estimation performance exceeds spatial smoothing algorithms.For the DOA estimation algorithm of wideband signals,it mainly presents the ISM algorithm and CSM algorithm which are based on signal subspace. On the base of Wideband signal model, including time domain model and frequency domain model, two modeling ways for wideband signals are introduced. And then makes lots of the analysis about the non-coherent signal subspace method (ISM) and Coherent Signal Subspace Method (CSM).Firstly, according to the idea of the ISM method, it decomposes the wide-band signal into many narrow-band signals and decreases the relativity of every narrow-band signal by using the improved MMUSIC method. It successfully applies the decorrelation method for narrowband to the wideband.Simulation showes the method is effective for the DOA esimation of wideband coherent signals. For the CSM method,most of the current focus ways need to pre-esitmate the directions of arrival.Secondly,A CSM method is proposed whicn is based on the resampling concept and reconstruction of the virtual array. It makes use of space resampling technique to adjust the interelement spacing. Through the interpolation processing and construction of virtual array, it makes the array manifolds of different frequencies tend to the same, and complete focusing. The simulation results show the algorithm at the small number of snapshots still keeps a good performance.and its performance is not so good at low signal to noise ratio, but with the SNR becoming higher, the performance of the algorithm will increase rapidly. And the algorithm is easy to realize, it's no need to pre-esimate the direction, and can be able to complete DOA estimation for.the coherent and incoherent broadband sources...