Research On Angle Estimation Algorithms For Noise MIMO Radars

Motivated by the development and the tremendous achievements of multiple-input multiple-output (MIMO) technique in the wireless communication field, a new radar system named MIMO radar is proposed recently. According to the array configuration, MIMO radar is mainly divided into two kinds. One is the distributed MIMO radar, whose transmit and receive antennas are widely separated, meeting the condition of space diversity. This makes the radar observe a target from different angles, thus resisting the target radar cross section (RCS) scintillation and improving the detection performance. The other one is the centralized MIMO radar, whose transmit and receive arrays are the same as the ones of the traditional phased-array radar, and the antennas are closed. Through transmitting orthogonal waveforms, the centralized MIMO radar can form a larger virtual array using waveform diversity, which brings many advantages such as increasing the maximal number of identification targets and improving the angle resolution. This paper focuses on the problem of direction of departure (DOD)-direction of arrival (DOA) estimation in noise MIMO radar. In addition, the properties of pulse compression output for noise waveform are analyzed. A fast method for designing random polyphase codes is studied, and a simple and easy to implement scheme is given for MIMO radar transmit beamforming. Meanwhile, the range and angle estimation for high-speed targets using the noise phased-array MIMO radar is addressed. The main works and contributions are summarized as follows.1. Related theories of noise MIMO radar1) The mathematical expression of mainlobe to sidelobe ratio for the noise waveform is deduced by researching the probability distribution and the second-order statistics of the sidelobes after the pulse compression. The quantitative analysis of the relationships between mainlobe to sidelobe ratio and pulse compression length, Doppler frequency and other parameters is given.2) Under the criterion of minimizing weighted integrated sidelobe level, a re-weighted cycle algorithm is proposed for designing random polyphase codes. The designed codes have low auto-correlation and cross-correlation sidelobe levels, and can be used as the transmit waveforms for MIMO radar.3) The sub-beam synthesis algorithm is proposed for transmit beamforming in MIMO radar. In the method, the transmit waveforms are weighted individually to form sub-beams, which add in the space and focus the transmit energy on the interested space, meanwhile minimizing the energy in the other spaces, thus improving the signal to noise ratio (SNR). This scheme needs no any optimization tool, easy to implement, and has low computational costs.4) The signal model of bistatic MIMO radar is established, and the principle of forming virtual array is discussed. The Cramer-Rao bound (CRB) of angle estimation for bistatic MIMO radar is derived.2. DOD-DOA estimation using joint matrix decomposition for MIMO radar1) A joint matrix diagonalization (JMD) based scheme is proposed for DOD-DOA estimation. First, the DOD-DOA estimation is transformed to the joint matrix diagonalization problem using singular value decomposition (SVD) and the theorem of rank-1matrix determined. Then, the single-sweep iterative algorithm is used to solve it, and the manifolds of transmit and receive arrays are obtained. Finally, the DOD-DOA can be estimated by spectrum analysis methods. The proposed method utilizes all the information of matched filter output, avoiding two dimensional (2D) spectrum peak searching, and possesses an accurate closed form solution at each iteration. The accuracy of angle estimation is nearly the same as2D MUSIC method, better than ESPRIT method and PM method. The DOD and DOA are automatically paired.2) A joint matrix upper triangular (JMUT) based method is prestented for DOD-DOA estimation. First, the DOD-DOA estimation is transformed to joint matrix upper triangular problem. Then, the extended QZ iteration algorithm is used to solve it, and the manifolds of transmit and receive arrays are estimated. At last, the DOD and DOA are regressed through spectrum analysis methods. This method makes use of all the information of matched filter output, avoiding2D spectrum peak searching, and possesses an accurate closed form solution at each iteration. The DOD and DOA are automatically matched. The accuracy of angle estimation is nearly the same as the result of JMD based method with lower complexity.3. DOD-DOA estimation using tensor higher order singular value decomposition (HOSVD) for MIMO radarFirst, the matched filter output is modeled by a three-way measurement tensor according to its multi-dimensional structure. The covariance tensor and cross covariance tensor are defined. Then the noise/signal subspace is estimated by HOSVD of the above three tensors. Finally, combined with2D MUSIC method and ESPRIT method, the DOD and DOA are estimated. Meanwhile, the first-order perturbation of the estimated signal subspace is evaluated, and its analytical expression is derived. The theoretical analysis and simulation results show that when the target number is less than the number of transmit antennas or receive antennas, the accuracy of estimated signal subspace of the proposed method outperforms the result given by the traditional matrix SVD based method, thus improving the angle estimation performance. Additionally, the unitary ESPRIT (UESPRIT) method is extended to the tensor case, and HOSVD UESPRIT algorithm is given to estimate DOD and DOA. This scheme can further improve the angle performance, especially for the scenario that the scattering coefficients of targets are correlated.4. Four dimensional (4D) angles and Doppler frequency estimation using tensor factor decomposition for MIMO radarA tensor factor decomposition based method is proposed for4D angles and Doppler frequency estimation in bistatic uniform rectangular array MIMO (UAR-MIMO) radar. The4D angles include elevation direction of departure (EDOD), azimuth direction of departure (ADOD), elevation direction of arrival (EDOA) and azimuth direction of arrival (ADOA). First, the signal model of URA-MIMO radar is established. Then the matched filter output is modeled by a five order tensor using the definition of matrix representation of a tensor, and the estimation of parameter matrices is tramsformed to the problem of factor decomposition of this tensor. The alternating least squares (ALS) method is used to solve it to estimate the4D angle matrices and Doppler shift matrix. Finally, the4D angles and Doppler shift are recovered through spectrum analysis methods according to the Vandermonde structure of the parameter matrices. In addition, to save complexity, the HOSVD is used to reduce the dimension of the original tensor when the number of antennas and pulses is large. The linear search is used to accelerate the convergence of the proposed algorithm. This method needs no estimation of noise/signal subspace and no spectrum peak searching. The4D angles and Doppler frequency are automatically paired with better angle estimation performance.5. Range-angle estimation for high-speed targets using noise phased-array MIMO radarTo solve the problem that it is ineffective for the traditional single-input single-output (SISO) radar to detect high-speed targets using the long time coherent integration method, a scheme based on noise phased-array MIMO radar is presented. The detection performance degradation caused by the target migration in SISO radar can be removed by coherently processing multi-channel echoes in parallel within a short duration. The range and angle of a high-speed target can be unambiguously estimated through the correlation pulse compression-beamforming algorithm. The noise phased-array MIMO radar can achieve compromise between the transmitting coherent processing gain of phased array radar and the high angular resolution of MIMO radar by adjusting the number of the transmit sub-arrays. Furthermore, using the random polyphase codes as the transmit waveforms gives low probability of identification and good anti-jamming performance, which makes the noise phased-array MIMO radar more easily meet the requirements in the modern battlefield.