Studies On Bearing Estimation And Its Performance Analysis For Spatially Distributed Sources

In applications of array processing such as radar, sonar, and wireless communications, the array observed signal is best modeled as a distributed, rather than a discrete source. The principal mechanism for making the source appear to be distributed in space, is unresolvable multipath scattering. So far, many methods have been proposed to estimate the nominal bearing of spatially distributed source. However, most of them are merely appropriate to the case of small angular spreading. When angular spreading is large, it is difficult to trade off between robustness and computational lost. Also, much related research requires the narrowband assumption, which is not always realistic in practical applications. It is necessary to analyze the effects of finite bandwidth to bearing estimation of spatially distributed source. Additionally, more complex noise models should be dealt with in practical situations as well as that of white Gaussian noise. This thesis mainly extends the research of bearing estimation for spatially distributed source from the above three aspects. The innovations of this thesis are as the follows:1) It is analyzed that the lower-order Jacobi-Anger expansion model for an incoherently distributed source has better accuracy performance than some other approximate source models such as space frequency model, two-point source model, and low-rank approximate model. Using the parametric decoupling property of the JA-based model, a novel angular parameter estimator is proposed, which is very robust in case of large angular spreading.2) The concept of space frequency distribution is introduced into reevaluation of incoherently distributed source model. Based on the model, two low-complexity robust bearing estimators are developed. It is proved theoretically that their estimates are asymptotically unbiased and consistent. 3) Using perturbation analysis, the space-correlation loss and the bearing-estimated error for finite-bandwidth incoherently distributed sources are evaluated analytically, respectively. It is disclosed that bandwidth-induced bearing bias increases monotonously with the signal's relative bandwidth. When the relative bandwidth is not small enough to be negligible, the spatially-only processing cannot lead to accurate bearing estimate, that is, the joint space and time processing is necessary for high-precision bearing estimation.4) A uniform CRB expression of bearing estimation for spatially distributed sources in unknown noise environments is derived. More exactly, the CRB performance of bearing estimation for spatially distributed sources in several typical spatially-correlated noises is evaluated theoretically and numerically, respectively. It is disclosed that the estimated errors mainly depend on space ambiguity between signal and noise components.5) Based on the noise model with a Centro-Hermitian covariance matrix, a differential denoising bearing estimator is developed for coherently distributed sources in unknown noise fields. Its accuracy and resolution performance is analyzed theoretically and numerically. In order to improve its estimated performance in case of low signal-to-noise ratio, a modified differential denoising estimator is formulated. Simulation results show the effectiveness of the proposed estimators under the low SNR case. The impacts of angular spread and number of sensors are also investigated.