Wideband DOA Estimation For Chrip Signals Using STFD Matrices

Direction-of-arrival estimation algorithms have been extensively studied forits significance in array signal processing field. It has been more mature intendedto adapt the modern application environments. The research objects turn to bewideband signals more than ever, and the background condition is developed fromthe overdetermined condition in which the sensor number is larger than the sourcenumber to the underdetermined condition in which we aim to estimate moresources than the number of array sensors. The study objects have been classified asnarrowband signals, wideband stationary signals and wideband nonstationarysignals, during which the wideband signals has been more widely studied becauseit bears more information than narrowband signals.In recent years, the spatial time-frequency distribution (STFD) method hasbecome a hot topic of DOA estimation methods which can be applied to widebandsignals DOA estimation. It considers three dimensions information of spatial,temporal and frequency domains of received signals, which has better performancecompared to many other algorithms under the same conditions. Meanwhile, in thecase of a fixed array structure and the number of sensors, how to accuratelyestimate more number of sources even larger than sensor numbers, becomes aconsiderable problem in many cases, as which we call it underdetermined condition.In this paper we address the problem of DOA estimation of wideband chirp signalsusing STFD method in the condition of underdetermined. All the study in thisthesis are as follows:Firstly, the background of array signal processing and the significance of thisresearch are introduced. The research status and development prospects of DOAestimation algorithms are reviewed and summarized. We studied the main methodused in this thesis of STFD method and several DOA estimation methods inunderdetermined conditions as well.Secondly, the wideband signal model and common array models are thendiscussed in chapter2. The main work in this part is to study two importantwideband DOA estimation methods, which are the coherent signal-subspacemethod (CSM) and the spatial time-frequency distribution method (STFD). Thecore idea of CSM is to create reasonable focusing matrices and select focusingfrequency. As well as the DOA estimation problem in the condition ofunderdetermined, we introduced an underdetermined DOA estimation methodnamely Khatri-Rao (KR) subspace approach, which has been developed in recentyears and has a relatively good performance. Thirdly, we present an STFD-CSM approach which combines the conventionalCSM algorithm with spatial time-frequency distribution method in overdeterminedcondition to estimate the DOAs of wideband chirp signals. This approach takes fulladvantage of the signal’s time-frequency characteristics, and the estimationaccuracy of the STFD-CSM method is significantly higher than the conventionalCSM and many other algorithms, thereby verifying the validity of the proposedapproach.In the last part of the thesis, we discussed a DOA estimation algorithm forwideband chirp signals in underdetermined conditions. Considering of thenonstationary characteristics and its approximate stationary characteristics duringrelatively short time interval, we propose a STFD-KR algorithm which combinesthe STFD method and KR subspace algorithm for underdetermined DOAestimation. By the derivation and simulation, it demonstrates that this STFD-KRalgorithm can be used in underdetermined DOA estimation cases. Meanwhile, ithas the higher accuracy of estimation than the existing methods.