Parameter Estimation Of Linear Frequency Modulated Signals Based On MP Algorithm

LFM signals are widely used in multitudinous signal processing area, such as radar, audio, sonar and so on.Therefore parameter estimation algorithms of LFM signals are deeply researched. This dissertation is mainly concerned with the estimation methods of LFM signals in gaussian and impulse noise case .Currently, time-frequency analysis is widely used in array signal parameter estimation field, via establishing the spatial time frequency distribution (STFD) matrix and then combining the traditional methods of array signal processing. However, such parameter estimation method performance will be affected by the selection of the correct time-frequency points and the cross-terms, and there would be a better estimation only under the condition of higher signal-to-noise ratio (SNR). Sparse decomposition of signals is emerging as a new signal decomposition and denotation method, which has been applied to array signal processing, It is mainly studied the application of MP algorithm in wideband array signal processing area in this dissertation.The following is the summarization of the main work:1,The character of LFM signals in time-frequency domain are investigated, and also some of the typical time-frequency analysis method have been studied,such as the time-frequency coherent signal subspace method (TF-CSM) in gaussian noise; the FLOM-based spatial-time-frequency domain MUSIC (FLOM-TF-MUSIC) and the FLOM-based RAT method (FLOM-RAT) in impulse noise.finally, the adaptive signal -- MP decomposition algorithm is discussed in detail.2,A new method based on MP decomposition is presented to estimate the 2-D direction of arrival of LFM signals in wideband array signal processing. The 2-D DOA of sources can be found by MP decomposition of the array output, over an adaptive over-complete matrix dictionary, the atoms of which are established according to the LFM signals' frequency parameter and the array geometry. Numerical simulations of the resolution show that the proposed method's performance superior to the conventional time-frequency coherent signal subspace method (TF-CSM ), especially in the case of lower SNR and more signals. 3,The high-resolution array direction finding algorithms are deduced in the presence of impulsive noises. The aim of this dissertation is to make full use of the MP algrithm, and to combine the Fractional Lower Order Statistics(FLOM) for parameter estimation of nonstationary signals.The FLOM-MP and infinity-norm normalization pre-processing based MP (IN-MP) algorithms are developed ,whose performance superior to the general FLOM-TF-MUSIC and FLOM-RAT algorithm.