Broadband Array Signal Doa Estimation And Tracking

DOA estimation and tracking for multiple wideband sources is an important research direction in array signal processing. With the development of military electronic technology and communication technology, wideband or ultra-wideband signals are used more widely. To meet the need of passive reconnaissance for such complex signals with large time-bandwidth product and large relative bandwidth, effective wideband array processing methods should be proposed. This dissertation is devoted to the study of two aspects of the wideband signal processing, the DOA estimation and tracking. The main researching work is as follow.(1) Effects of nonzero bandwidth on narrowband array algorithm. Along with the increase of bandwidth, the algorithms which are applicable to estimating the number of narrowband signals, such as AIC and MDL, will appear over-estimation of source number. When the number of snapshots is enough and the relative bandwidth of the signals are all less than 5%, MDL algorithm can accurately estimate the number of sources at moderate or high signal-to-noise ratio (SNR). Neither of two algorithms can accurately estimate the number of sources even if only one source's relative bandwidth of the signals is greater than 5%. Meanwhile, even if we can get the right number of the sources, the estimation error of classic DOA estimation algorithms will also increase as the bandwidth increases. Among all the arrivals, even if only one's bandwidth increase, the estimation errors of all sources will increase.(2) Four wideband DOA estimation algorithms are studied. Two of them are focusing methods which require initial DOA estimates: Rotational signal-subspace method (RSS), Two-sided correlation transformation method (TCT). Two of them are focusing methods which don't require initial DOA estimates: Array manifold expansion method (AMFE), Robust coherent subspace method (R-CSM). The problem to select the optimal focusing frequency is analyzed by simulation experiments. The simulation results show it will minimize the focusing error to select the center frequency of signals as focusing frequency when the signals have smooth or symmetrical power spectrum. The algorithms based on unitary focusing matrices have better estimation performance than others based on nonunitary focusing matrices. The simulation results demonstrate the two methods which don't require initial DOA estimates are effective. The AMFE method can estimate DOAs of multiple wideband sources, but the estimation accuracy is limited. The R-CSM method can improve its estimation accuracy by multiple iterations.(3) A DOA tracking method based on focusing matrices updating is studied. The simulation result show it can effectively track the change of multiple wideband sources'DOAs. We must determine appropriate size of observation window to make the method has both high response speed and good tracking stability. A new wideband DOA tracking method is proposed. The new method has good tracking ability when the appropriate forgetting factor is selected. The simulation result show the new method has better tracking ability than the first method.(4) An outfield experiment is carried out. The objective of the experiment is to demonstrate the validity of a wideband DOA estimation algorithm based on two array elements. The results show the algorithm has ability to estimate multiple wideband sources'DOAs and good estimation accuracy.