A Study on Radar Signal Processing for Accurate DOA Estimation of Multiple Signal Sources

DoctorThis dissertation discusses a study on radar signal processing for accurate direction-of-arrival (DOA) estimation of multiple signal sources. The organization of this study is as follows. First, we proposed a novel scheme to improve the accuracy of the DOA estimation of multiple targets (i.e., signal sources) by combining the independent component analysis (ICA) and the phase comparison monopulse (PCM) technique. The proposed scheme is performed in three steps: 1) preprocessing 2) signal separation of each target’s signal components using complex-valued FastICA algorithm with measurements of a few successive radar return signals 3) angle estimation by applying the traditional PCM technique to individual separated interference-free signals. Because of this completely new approach, both mutual interference among echo signals from the multiple targets and the noise components leading to an angular error can effectively be mitigated, reuslting in an accurate angle estimation of each target. Consequently, much accurate DOA estimation in two-dimensional space can be made by the proposed scheme. Next, we proposed a new scheme using ICA as well as fractional Fourier transform (FrFT) to determine accurate DOAs of multiple wideband linear frequency modulated (WLFM) signals without both the cross-term interferences and severe noise components. The proposed scheme is performed in three steps: 1) preprocessing 2) signal separation that two homogeneous WLFM signals associated with each signal source are extracted by performing ICA algorithm only once. 3) DOA estimation using phase difference between the two homogeneous WLFM signals at the energy-concentrated domain, which can be transformed by FrFT. In this case, we can determine the rotation angle corresponding to the energy-concentrated domain of each WLFM signal by comparing the maximum peak values of individual FrFT results with different fractional Fourier domains. Thanks to the proposed scheme, the cross-term interference problem between the WLFM signals can be avoided. In addition, the noise components induced in the received signals are effectively mitigated. As a result, an enhanced DOA estimation accuracy can be achieved, as compared to the above mentioned existing methods. Finally, we propose a new efficient sparse representation algorithm based on the orthogonal matching pursuit (OMP) using a step-by-step search approach in order to rapidly achieve accurate DOA estimation of multiple targets using a single measurement vector (i.e., array signal vector). Regardless of the high mutual coherence of redundant dictionary (i.e., array manifold matrix), the optimal sparse solutions for DOAs of multiple targets can be yielded with a relatively low computational cost, as compared to the iterative local searching OMP method that is up-to-date. Furthremore, we showed that the proposed scheme is more suitable for DOA estimation of multiple targets using the single measurement vector than the existing subspace-based methods in a noisy environment