Research On Coherently Distributed Chirp Source Localization

Estimating the azimuth and elevation of source,source number and spatial positons of sourses by using the spatial characteristics of source on the sensors of array,are important tasks for the fields of mobile communication,radar detection,sonar system,etc.Chirp signal is an important class of non-stationary signals whose frequency varys with time.This class of signals has excellent performance of time resolution,larger time-bandwidth product,and compared to narrowband signals,it possesses a stronger ability to resist interference Doppler shift.Besides,the two important paraments of this class of signals possess location informations.Therefore,chirp signal is very suitable for applying to the fields of target detection and positioning.The research regarding the direction of arrival(DOA)estimation problem for chirp signal has important significance.Most of chirp signal's DOA estimation methods are based on the assumption of idealized point source in the far field.However,targets in practical applications have more complex spatial distribution generally,such as,lacking of direct wave between source and revieved array or random fluctuation during the observation period.Hence,a new data model is required and the corresponding high accuracy spatial parameter estimators are needed as well.To solve the problems above,this thesis carries out the following main works:1.Distributed chirp source modelingFor multiple chirp sources case,discrete and continuous distributed chirp source models are respectively constructed in the time domain for two different application environments that is the number of scattered sources is relatively small and relatively large.Then,using the time-frequency characteristics of chirp signal in the fractional Fourier domain,the models in the time domain are simplified in which eliminating the influence of time parameter on the location vector.It offers the possibility for achieving that the classical narrowband DOA estimators are extended to the counterparts in the case of distributed chirp source.2.The spatial parameters estimation of discrete coherently distributed chirp sourceWhen the number of scattered coherently sources is relatively small,the local scattering environment can be simutlated by discrete coherently distributed chirp source model.The steering vector in the proposed model can be simplified and calibrated by using Taylor expansion.In the Energy-concentrated domain and the Dechirping domain,a proper spatial parameter estimation estimator for the distributed coherently chirp source is proposed.Besides,the feasibility of the proposed methods is demonstrated by simulations.3.The spatial parameters estimation of continuous coherently distributed chirp sourceWhen the number of scattered coherently sources is relatively large,based on the continuous coherently distributed chirp source model,the subspace decomposition algorithm,the ESPRIT algorithm and the maximum likelihood estimation algorithm are respectively improved that they can be exploited in the continuous coherently distributed model in the fractional Fourier domain.Besides,the corresponding generalized formulations of these estimators are proposed for estimating spatial parameters estimation of distributed chirp source.For reducing the computational complexity,a low-complexity algorithm based on 1D MUSIC spatial parameter search is proposed.The simulation results are applied to evaluate and compare the performance of proposed algorithms.4.The spatial parameters estimation of chirp source at correlated signal scenariosIn order to improve the applicability of the spatial parameter estimation of chirp sources in high correlated signal environment and the resolution capability,a novle sparse recovery algorithm based on the time-frequency dictionary is proposed.The corresponding time-frequency dictionary and sparse model in the spatial domain is constructed according to the time-frequency characteristics of chirp signal in the fractional Fourier domain.This algorithm combines with the filtering technology,which effectively improves the anti-interference capacity.The optimum solutions of DOAs in the proposed model can be estimated by iterative L1-norm minimation.Finally,the numerical experiments demonstrate the effectiveness of the proposed algorithm.