Parameter Estimation Of MIMO Radar Based On Compressed Sensing

MIMO radar target parameter estimation is of paramount importance for the development and application of the new radar system. In this thesis, the problems of MIMO radar target parameter estimation such as high sampling rate, large data and so on are investigated based on the huge advantage of the target sparse property and the compressive sensing(CS) technology. The parameter estimation methods for MIMO radar are emphasized based on CS, including CS-based DOA estimation, joint estimation of DOA and Doppler, as well as the CS-based design of the measurement matrix.In chapter 1, the background and significance of the research are analyzed. The development and the state-of-the-art research of the MIMO radar system and the parameter estimation methods are reviewed. Meanwhile, CS theory and its application are summarized. The CS-based MIMO radar parameter estimation methods are also explained in detail.The second chapter is the basic theory. The CS theory is introduced and the mathematical model and key factors for CS are discussed. Based on the MIMO radar parameter estimation theory based on CS, the MIMO radar parameter estimation principle based on CS is analyzed, together with the RIP condition, sparsity analysis of the echo, dictionary creation, data acquisition of CS data and the CRB of parameter estimation. Inspired by the numerical simulations, the method of using CS for MIMO radar parameter estimation is validated, which is the fundamental basis of further research.MIMO radar DOA estimation based on CS is investigated in chapter 3. By intruding the traditional methods of MIMO radar DOA estimation, the MIMO radar DOA estimation principle is clearly shown. Firstly, single target DOA estimation method for MIMO radar based on CS is proposed. Both one-dimension and two-dimension models are constructed and how to use these models in DOA estimation is analyzed. Simulation results demonstrate the advantages of the proposed methods. Secondly, multi-target DOA estimation method for MIMO radar based on CS is proposed. Similarly, one-dimension and two-dimension models are constructed and analyzed. The resolution limit for DOA estimation is also addressed. The superiority of the method and the importance of resolution limit are shown via numerical results.In chapter 4, the joint estimation of target DOA and Doppler for MIMO radar based on CS is focused on. The model of DOA-Doppler joint estimation for MIMO radar echo is obtained first, followed by the introduction of Doppler ambiguity function and the basic method of DOA-Doppler joint estimation. Then we investigate how to use CS to jointly estimate DOA and Doppler, including the based principle, the grid discretization method of DOA-Doppler plane, and the unambiguous velocity interval. The CS-based method of DOA-Doppler joint estimation is proposed and analyzed. The advantage of the method is illustrated by analyzing the SIR performance. The Signal-Infrerennce Ratio(SIR) of stationary target and slow-moving target using random measurement matrix and advanced random measurement matrix is calculated. The performance of the method is compared with that of the matched filter method. Experiments of whether the target lies in the DOA-Doppler grid or not shows that the proposed method outperforms the matched filter method. The importance of grid discretization is obtained by the experiment that target lies outside the grid.In chapter 5, the measurement matrix design of CS-based MIMO radar is addressed. Some related problems like random measurement matrix, reconstructing condition, reconstructing condition of mutual coherence, reconstructing condition of cumulative mutual coherence, and so on are introduced, which form the basis of the optimization of the measurement matrix. The optimization problem based on the mutual coherence is proposed using the model of the measurement matrix design problem. Theoretic analyses prove the feasibility of the method. Another measurement matrix optimization method is proposed based on the cumulative mutual coherence. Simulation results show that comparisons between the two methods and the Gaussian random measurement matrix, which provide the validation of the two methods.The thesis is concluded in chapter 6. Main innovations of the thesis are shown and future research tracks are provided.