Study Of Array Calibration And DOA Estimation In Nonuniform Noise

The conventional high-resolution direction-of-arrival (DOA) estimation algorithms are based on the ideal steering vector in the presence of white Gaussian noise, i.e., there are no array errors, and the background noise is white Gaussian noise with known statistical properties. However, in practical applications, array errors are inevitable due to the array antenna craft or the circumstance around the array and so on. On the other hand, the background noise usually is colored noise with unknown statistical properties. Consequently, DOA estimation in the presence of array errors and colored noise is a research focus in array signal processing. In this thesis, the methods for DOA estimation in the presence of array errors and nonuniform noise are discussed.1. The conventional methods for array gain and phase uncertainties calibration require that the sources are impinging on the array with known locations. While the global convergence of auto-calibration based on iterative approach can't be guaranteed. Additionally, the DOAs and phase uncertainties are not independently identifiable for unifrom linear array. In this thesis, a method for array gain and phase uncertainties auto-calibration with instrumental sensors is presented. Two well-calibrated sensors are added into the original array. With the principle of ESPRIT, the DOAs and errors can be estimated through eigen-decomposition simultaneously. Compared with conventional methods, the proposed method has no requirement of calibration sources with known locations or iterative and multidimensional search procedures, thus the computational complexity is less. Moreover, the method can overcome the ambiguity in DOA and phase uncertainties estimation. Computer simulations demonstrate the validity of the proposed method.2. Spatial smoothing algorithm is a common decorrelation method. However, its performance will be greatly degraded by the mutual coupling between array elements, so that the DOAs of coherent signals can not be estimated precisely. To deal with such a problem, a method for DOA estimation in the presence of mutual coupling as well as coherent signals is addressed. Based on the principle of subspace, the mutual coupling matrix can be estimated with independent signals among the impinging signals. After compensation for the mutual coupling, the DOAs of coherent signals among impinging signals can be estimated precisely with spatial smoothing algorithm. The simulation results also show the feasibility of the method.3. The performance of conventional high-resolution algorithms will deteriorate in colored noise. In this thesis, we present a new method for DOA estimation in the presence of unknown nonuniform noise whose covariance is an arbitrary diagonal matrix. The proposed method can estimate the noise covariance matrix effectively based on the Toeplitz structure of array covariance matrix of independent signals. The negative effect of the nonuniform noise on DOA estimation can be eliminated by subtracting the noise covariance matrix from the array covariance matrix, and the conventional MUSIC algorithm can be applied directly. The proposed method is computationally simple and achieves a better performance in DOA estimation.