The traditional NLS (Nonlinear Least Square) based on beampattern synthesis method to MIMO radar is studied in the dissertation. We first decompose the covariance matrix of the transmit waveform. Then the cost function of NLS is constructed to optimize the decomposition factor, so as to obtain a desire spatial transmit beampattern. Compared with the beampattern synthesis for MIMO radar in existence, the proposed method can adjust the width of transition region and mainlobe region ripple by changing the penalty function and has less mean square errors. In additional, a subspace based DOA (Direction-Of-Arrival) estimation method for MIMO (Multiple-Input Multiple-Output) radar is studied in this paper. By transforming the data covariance matrix, we can construct the projection matrix of signal subspace and noise subspace. The computation complexity of the algorithm is reduced as it does not need to perform eigendecomposition. By using the over estimated theory, this algorithm avoid estimating the number of signal sources and can keep a good estimate performance in the case of lower SNR (Signal-to-Noise Ratios) and multi-target. Computer simulation results confirm the effectiveness of the algorithms. |