The Research On Blind Signal Separation Method For Low Angle Radar

At present, using the low-flying strategic weapons such as aircrafts and cruise missiles which flight in the low-altitude below 1000 meters’ height to raid the ground targets has become one of the important means of modern warfare. Due to the difficulty for the conventional radar to find and keep track of the objects, the research on the stability and precision of the radar’s detecting and tracking aiming at the low-flying targets is a key subject to the information-based war and it has a great practical significance in the field of military and air defense.The multipath effect of electromagnetic wave propagation is formed when the radar is tracking the low-flying targets. It was because that the surface of the ground and sea or the surrounding terrain may cause the mirror reflection and diffuse scattering. Multipath effect makes the radar receive the superposition of direct wave signal and signals from other propagation path. This not only makes the echo amplitude and phase of signals received by the radar change, but also leads to the big errors in the target detection parameters. It may even prevent radar from stable detecting and tracking for the targets. Therefore, in order to realize the precise measurement and tracking of the low altitude target, the multipath signal of radar’s measurement signals must be removed.Then we can extract the direct wave signal and improve the precision of the radar to target detection parameters.This thesis is mainly to study how to use the blind source separation algorithm to extract the direct wave signal effectively, restrain the error of the multipath effect, and further improve the measurement precision of the radar. First of all, this thesis expounds the task background, significance and research status at home and abroad about the detecting and tracking of the low-angle radar. Then the thesis expounds the general situation of the blind source separation and the principle of multipath effect. The geometric model of the multipath effect is analyzed and how the multipath effect makes impact on the echo amplitude, phase and echo power of signals received by the radar is discussed. After that some commonly used algorithms for the blind source separation are studied. The blind source separation algorithms include AMUSE algorithm, JADE algorithm, CuBICA algorithm and FastICA algorithm. Through the simulation analysis and comparison about the similarity coefficient between the separation signals and the source signals of various algorithms, the root mean square error of the separation signals and the curves of the output signal to noise ratio, it is verified that the separation performance of the CuBICA algorithm is superior to other algorithms and the CuBICA algorithm has higher separation accuracy and better robustness. Then,by using the CuBICA algorithm, third-order and fourth-order cumulants are diagonalized simultaneously in order to realize the independent component analysis. It is verified that the CuBICA algorithm has a strong ability to adapt to the different signals. It is able to handle the super-gaussian and sub-gaussian signals at the same time, and also has a good performance on the linear mixtures of symmetric and skew distributed source signals. In the end, the CuBICA algorithm of blind source separation algorithm is used to process the multipath data measured at an experimental base. By comparing the root mean square error curve about azimuth, elevation angle and distance before and after the blind source separation at different sampling points, and the ratio of their root mean square error before and after the blind source separation, finally it is realized that the measured radar data is processed, the radar direct wave signal is extracted effectively and the error of the multipath effect is restrained. It is also verified that this method has the effectiveness and practicability. This method can also be applied to radar data processing of the high-altitude target detection.