The Research On Key Technique Of Direction Of Arrival For Passive Detection System

Direction of arrival (DOA) estimation by array is an important research branch in the field of modern signal processing. It has a great variety of applications in many fields, such as communications, radar and sonar, etc. In order to make passive detection system have the ability of measuring multiple radiant signals arrived simultaneously and improve the angle resolution of DOA estimation, application of super-resolution DOA estimation algorithm by array to passive detection system is researched deeply in the dissertation. According to the complexity of electromagnetism condition faced to passive detection, the dissertation extend the research of key techniques for passive detection system, including fast DOA estimation, complex radar signal DOA estimation, source number estimation and array channel calibration. Each subject is studied in detail and corresponding algorithms are proposed. The contributions and fruits of the dissertation are listed as follow.In the aspect of DOA estimation for multiple narrow signals, the dissertation focuses on fast subspace calculation for the data received, because the implementation of subspace methods in passive detection application with real-time operation usually experiences a bottleneck in the calculation of the signal or noise subspace. Propagator Method (PM) for fast computation of signal or noise subspace has introduced. Superior to conventional DOA estimation, PM can obtain the noise subspace and signal subspace fastly in stead of Eigen-Value Decomposition (EVD). The performances of the Propagator Method in terms of the mean squared error on DOA estimation are investigated in detail. Based on Propagator Method, several 2D fast algorithms of DOA parameter estimation have designed on L shape of array structure. Furthermore, a joint estimation of source number and DOA is proposed in order to conquer the inherence of limitation in PM for application. Computer simulation confirms the proposed methods valid impressively.Many methods of DOA estimation are essentially limited to processing narrow-band data. This is indeed a realistic assumption in many applications (e.g. active radar and communication). However, in the cases of passive detection system, the received signals may be broadband or narrowband distributed in broadband. In the environment of multiple narrow-band signals distributed in broadband, the subspace structure of data received by array can not be generally used for DOA estimation without the frequency information of each signal received. So the joint frequency-DOA methods of multiple narrowband signals are researched. Characteristic of time-space equivalence is properly used to joint frequency-DOA estimation. After analysis of joint frequency-DOA estimation techniques by uniform linear array (ULA), Joint frequency-DOA estimation by arbitrary array structure is researched deeply. Based on State-space joint estimation algorithm, a novel joint estimation algorithm, named SOBI (Second-Order Blind Identification) joint frequency-DOA estimation is shown by changing the structure of non-central auto-covariance matrix of data. Compared to the State-space algorithm, the new algorithm can realize the frequency-DOAs estimation in the condition that source number is more than the number of sensors in array. At last, a fast algorithm of joint frequency-DOA is explored for real-line application.For DOA estimation of wideband signals, developing spatial time-frequency technology is much suited for non-stationary signals. Spatial time-frequency distribution (STFD) is illuminated and its structure is analyzed, simulations show that STFD act effective for DOA estimation than the common methods. For dwelling with the large computation load of this kind of methods, two fast algorithm are derived. As the further progress of STFD, spatial ambiguity distribution of the signals received by array are exploit for the scenarios of LFM signals instead of STFD. Ambiguity transformation converts the absolute time and frequency of LFM signal into the domain of relative time lag and frequency difference, which bring more attractive steering vector structure than that in time-frequency domain. Several algorithms for DOA estimation in ambiguity domain for coherent and incoherent LFM signal are proposed, which can be applied to arrays of any aperture size and arbitrary chirp rate signals. The frequency of the sources can be higher than the non-aliasing frequency of array designed for signals and the number of sources can be greater than the number of sensors. The simulations demonstrate the improvements in the DOA estimation of the wideband LFM signal.At last but not the least, the factor analysis technique is introduced for sources number estimation and array channel calibration, which are two key problems to influence the performance of DOA estimation. Factor analysis model is shown to approximate the model of covariance matrix of data received. Then the statistic signal processing method for factor analysis is applied to estimate the sources number in colour noise condition. The simulation proves that the performance of the method for sources number estimation is better than that of GDE (Gerschgorin Disk Estimator) in certain false alarm rate. The two novel methods of array channel calibration, Least Square algorithm and Column Ration algorithm, are proposed by factor analysis technique. Based on an auxiliary testing source on microwave front-end, the two algorithm can extracting the amplitude-phase gain and noise parameters from the estimated covariance matrix in real-time.