Research On Wideband Signal Off-Grid DOA Estimation Methods

The array signal processing is wildly used in military and civilian purposes,such as sonar,electronic countermeasures and seismic exploration.This paper mainly focuses on the Direction-of-Arrival estimation problems,which is a critical part of array signal processing.The traditional DOA estimation algorithms are mostly based on the signal subspace methods which mainly consist of subspace fitting and subspace decomposition.These methods highly rely on the signal-to-noise ratio(SNR)and snapshot number.To overcome the shortcomings of the traditional DOA estimation algorithms and push the limitation of Nyquist sampling theorem,the sparse signal reconstruction system is established along with the introduction of Compressive Sensing(CS)theory.Simultaneously,the DOA estimation can be solved by Bayesian Compressive Sensing(BCS)when the Bayesian theory is combined with CS methods.The thesis starts from the Bayesian Compressive Sensing theory.The wideband signal estimation methods and off-grid DOA estimation methods are mainly studied.Combining these two sorts of methods,the off-grid methods which aim to wideband signal are proposed.Based on the contents above,the works and contributions of this thesis are illustrated as followings.The thesis first illustrates the model of array signal processing and introduces the traditional DOA estimation algorithms.Based on this model,the compressive sensing theory and spatial discretization theory are introduced.Some of the mature methods of sparse signal reconstruction are classified and illustrated.With the explanation of Bayesian theory,the thesis illustrates the application of BCS which is used in DOA estimation.Then the classic BCS based DOA estimation algorithm is realized and discussed.To overcome the dependency on snapshots number,a DOA estimation method based on RTO-MH is raised in the thesis.The proposed method has better accuracy with fewer snapshots and shorter processing time.The off-grid signal model is studied for solving the grid mismatch problem in practical DOA estimation.Based on this model,the Off-Grid Sparse Bayesian Inference(OGSBI)method is realized.And its performance is evaluated.Simultaneously,the RTO-MH based method is improved so that it can reconstruct the off-grid signal.This new method is called OGRTO(Off-Grid RTO-MH).For the comparison of OGRTO method and OGSBI method,the simulation results indicate the proposed method performs better with fewer snapshots.The traditional DOA estimation algorithms usually aim to narrowband signals.However,there exist lots of wideband signals in the practical situations.The thesis first illustrates the definitions of narrowband signal and wideband signal.Two common methods which are used to estimate the wideband signal are introduced.The limitations of these two methods are analyzed.To deal with the limitations,the joint sparsity of wideband signals is exploited.The off-grid DOA estimation method for wideband signals is studied based the combination of joint sparsity and OGRTO method.