Wideband DOA Estimation Under Clutter Using MIMO Radar With Sparse Array

Direction of arrival(DOA)estimation is an important research topic in array signal processing.In the presence of clutter,the strength of target echo is usually much lower than the clutter,which leads to the degradation of target DOA estimation performance.Multiple-input multiple-output(MIMO)radars are equipped with multiple transmit antennas to transmit different waveforms independently and process the signals received by multiple receive antennas jointly,which can improve the target parameter estimation performance.Employing wideband signals to carry more information can enhance antiinterference ability,and thereby is favorable for target detection,parameter estimation,and feature extraction,etc.In this paper,we investigate wideband DOA estimation under clutter using MIMO radar,and a wideband DOA estimation method based on compressive sensing after beamforming(WCSAB)is proposed,which is shown to have better performance than traditional wideband Capon technology(WCT).In practice,the number of available antennas may be limited due to the system complexity and cost constraints.When the number of antennas is limited,the DOA resolution of the traditional radar with full array structure may not be satisfying.To address this problem,the sparse array is introduced,and the WCSAB method is extended to the sparse array case.Considering that the performance of DOA estimation is not only related to the beamforming weights,but also to the sparse array structure,we formulate the joint optimization problem of beamforming weights and sparse array antenna positions based on minimizing Bayesian mean square error(BMSE).Considering that this problem is a combinatorial optimization problem,a simplifying iterative algorithm is presented to solve this joint optimization problem.The results show that the performance of the sparse array designed by the algorithm is close to the optimal sparse array,and is better than the nested array and co-prime array.Although the iterative algorithm is highly effective in solving the joint optimization problem of beamforming weights and sparse array antenna position design,it involves high computational complexity such as matrix inversion,etc.Deep neural network(DNN),by extracting useful features from the training set,can achieve high accuracy with reduced computational complexity.In this work,a DNN structure is proposed to match the considered signal model and to solve the joint optimization problem.The objective function and constraints of the original joint optimization problem are used as the loss function of the DNN during the training process.Through training the neural network,the optimum connection weights of DNN can be obtained,which can be viewed as the solution of the joint optimization problem.With the optimized weights and positions,the output of DNN gives the estimated DOA.The higher training and testing accuracy of DNN imply better DOA estimation performance.The performance of the system using DNN is analyzed through numerical examples,the accuracy of DNN is provided.The results show that when signal-to-noise ratio(SNR)and signal-to-clutter ratio(SCR)are high enough,the accuracy of the trained DNN based sparse array is close to that of the full array.