Research On DOA Estimation Of Wideband LFM Signals Based On Fractional Fourier Transform

Direction of arrival(DOA) estimation is a significant research point in array signal processing, which is playing an increasingly important role in radar, sonar, seismic survey systems and biomedical engineering. Especially DOA estimation of wideband linear frequency modulation signal has extensive practical value which can carry large amounts of information and has been widely used in radar and underwater acoustic communication systems. Some approaches proved to be effective have been proposed on the research of DOA estimation of coherent LFM signals during these years of development. However, achieving a robust algorithm with higher estimation accuracy and little computation complexity is still a challenging and promising problem due to the complexity of physical environment and various interferences.Based on the research and simulation of algorithms of DOA estimation of LFM signal on the basis of fractional Fourier transform, this dissertation intensively investigated an effectively improved algorithm of DOA estimation of coherent wideband LFM signals to overcome the shortcomings of the existing algorithms with array redundancy, aperture loss, and inadequate estimation accuracy. In addition, we made an effective and successful attempt to apply genetic algorithms to the field of DOA estimation of coherent wideband LFM signals in the dissertation. The major achievements of the dissertation are as follows:Firstly, proposed an effectively improved algorithm of DOA estimation of coherent wideband LFM signals. Based on the dechirping property of LFM signal in fractional Fourier domain, the time-invariant direction matrix and a new array data model in that dechirping FRF domain was constructed, which has achieved more prominent performance of estimating the DOA of adjacent coherent wideband LFM signals, owing to being combined with filtering and fully excavating the time-frequency information included in observed signals compared with the conventional method. In addition, the proposed approach increased the ability of discernment and decorrelation on the same SNR with no array redundancy and aperture loss and also increased the number of DOA that can be resolved, combined with reusing the conjugate data being rearranged and a modified MUSIC with low rank approximation closely and deeply.Secondly, proposed an effective method greatly reducing the computational complexity in which genetic algorith was applied to the field of DOA estimation of coherent wideband LFM signals successfully. The spectrum peak search of LFM signal’s FRFT received by the reference array was accomplished accurately and efficiently with genetic algorith which obtains the ability to quickly search and strong robustness in the algorithm of DOA estimation of coherent wideband LFM signals based on forward/backward spatial smoothing in FRF domain, in the algorithm of DOA estimation of coherent wideband LFM signals based on Toeplitz matrix and in the improved method proposed. The approch of genetic algorithms have greatly reduced the computational complexity of the algorithm of DOA estimation and have been adapted to the application in real-time occasion, easing the contradiction between the computational algorithm and estimation accuracy efficiently.