Nonuniform Sampling Based Super Resolution Direction Of Arrival Estimation

Super-resolution direction-of-arrival (DOA) estimation based on an array antenna system is the focus of the research of spatial spectrum estimation. In real applications, however, the estimation performance of super-resolution algorithm depends on the size of the array's aperture, signal-to-noise ratio and the number of snapshots, etc, the improvement of the accuracy of DOA measurements is one of the challenges in radar array system all along. Following the classical sampling theory and signal processing theory in the time domain, the existing DOA algorithms is unified in Nyquist sampling theorem framework based on uniform sampling. Theory and simulation in MATLAB:an integrated approach is used in this dissertation. Nonuniform sampling in the time domain is introduced to spatial array signal processing, and several key techniques on nonuniform sampling based super-resolution direction finding methods are developed, and their corresponding theoretic analysis and deductions are given. The simulation results show that the proposed algorithms are valid, feasible and robust. The main contributions of this dissertation are as follows:1) The channel outputs are sampled on nonuniform sampling sets, and a novel approach is presented for improving the performance of DOA estimation by using nonuniform sampling (NUS) on sensors. A nonuniform spatial sampling time function is constructed. Due to the random variation in time interval of nonuniform spatial sampling, the signal-to-noise ratio (SNR) of the sensor outputs is increased, which makes MUSIC algorithm gain lower the probability of detection error and better estimation accuracy, especially at low SNR regions. A probability theory based weighted average method is presented to eliminate the abnormal estimation data of DOA, which combines the cluster analysis of each experiment results to decrease the estimation error especially as detection failure is occurred in single independent experiment.2) According to the quantitatively analysis of the impact of array spacing to angle ambiguity distribution, an angle estimation method based on specified interpolation sectors with element space larger than half wavelengths for uniform linear array (ULA) is proposed. Since the spatial potential distribution of impinging radiation is random, a pre-processing technique is involved to null out the signal outside the sector of interest. With the aim of resolving the problems posed above, some fourth-order cumulants of properly chosen sensor outputs from the uniform linear array (ULA) is computed to construct virtual array with one times larger expanded aperture, a DOA estimation algorithm based on spatial fourth-order cumulant is proposed, in which the propagator method without eigendecomposition is applied to estimate DOA. The proposed algorithm doesn't require the number of signals is smaller the number of sensors, and is computational efficiency as it doesn't need eigendecomposition and spectrum peak search.3) Aim to the estimation of DOAs in presence of strong signal in the practical application, a method based on null preprocessing algorithm for weak signals'direction estimation is studied, which first estimates the DOA of strong signals based on classical spatial spectrum estimation algorithms, then constructs a null weighted matrix based on singular value decomposition (SVD) to suppress the strong signals, at last estimates the weak signal's direction based on the preprocessing sensor outputs with MUSIC algorithm. The angular resolution for the DOA estimation of multi-sources with large power difference is improved compared with that of classical MUSIC algorithm. The proposed algorithm possesses the better estimation performance over JJM. In addition, NUS is beneficial to the performance improvement of the proposed algorithm.