Font Size: a A A

Research On Array Design And Sparse Robust Signal Processing Algorithms For MIMO Radar

Posted on:2017-02-14Degree:DoctorType:Dissertation
Country:ChinaCandidate:J YangFull Text:PDF
GTID:1108330488472912Subject:Signal and Information Processing
Abstract/Summary:PDF Full Text Request
Multiple-input multiple-output (MIMO) radar is an emerging radar system that is developed on the basis of MIMO wireless communication, and it has drawn considerable attention from researches during the past decade for a number of its remarkable performance superiority in target detection, tracking, identification, parameter estimation and etc.. Unlike a standard phased-array radar, the MIMO radar paradigm can simultaneously transmit multiple orthonormal probing waveforms via its transmitting antenna elements to illuminate the target, and have the ability to jointly process signals received at multiple receive antennas. The waveform diversity afforded by MIMO radar enables significant superiority over its phased-array counterpart, including improved spatial resolution, increased degrees-of-freedom (DOF), enhanced interference rejection capability, and flexibility for transmit beampattem design. MIMO radar can be classified into two basic regimes, namely distributed and collocated, according to the specific antenna configuration selected. In distributed MIMO radar, the antennas are widely separated. This enables each antenna pair to view the target from different angles, and captures the spatial diversity of the target’s radar cross section. In collocated MIMO radar, all the antennas are closely spaced, so that the viewing angle does not vary between different antenna pairs, but the waveform diversity offered by such a regime can still enable the MIMO radar system to achieve improved performance. In this dissertation, we investigate signal processing algorithms using collocated MIMO radar, with special attention on array geometry optimization, direction-of-arrival (DOA) estimation algorithm design, robust adaptive beamforming algorithm design and sparse imaging algorithm design. The main contributions can be summarized as the following four aspects:The maximum number of targets that can be uniquely identified by the traditional uniform MIMO radar is limited by the number of virtual sensors. To alleviate this issue, we design a novel antenna array geometry in MIMO radar which is based on the concept of nested array in the first part, and a corresponding spatial sparsity-based DOA estimation method is also developed. By utilizing this new MIMO array geometry, which is also called nested MIMO radar, we are capable of resolving more independent targets than the number of virtual sensors. The reason for such an increasement in the available DOF is that, the transmitting and receiving antenna elements of the aforementioned nested MIMO radar are properly arranged in order to achieve the desired nested sample virtual array, which can be synthesized by transmitting orthogonal waveforms and extracting the waveforms in each receiving element with a set of matched filters. Furthermore, by vectorizing the covariance matrix of the signals received by virtual array, we can obtain a covariance vector, which is reminiscent of the received signal of single snapshot from a uniform linear array with an extended aperture, known as the difference co-array of the virtual array. Considering the special structure of nested array, the DOF of the corresponding difference co-array can be definitely larger than that of physical array, and as a result, DOA estimation can be done for underdetermined case. Additionally, since the traditional subspace techniques fail to yield reliable DOA estimates under the single measurement vector environment, we develop an alternatively efficient high-resolution DOA estimation approach, which is based on a sparse Bayesian learning (SBL) criterion. The proposed algorithm well reserves the superiorities of its counterparts in superresolution and adaptation to nonideal signal environment, and resorts to a computationally tractable polynomial rooting procedure to eliminate the discretized sampling grid error one by one based on the reconstructed spatial power spectrum simultaneously.In the second part, we extend the coprime array idea to the case of MIMO radar and address the antenna array design problem of coprime MIMO radar, and somewhat similar arguments discussed in the first part can be applied for this novel structure as well. Specifically, by using the proposed configuration optimization method to design the transmitting array and the receiving array, respectively, one can construct a MIMO virtual array with a certain desirable characteristic, i.e., a coprime virtual array, so as to utilize the increased DOF of the coprime co-array in the light of the correlation domain, and detect more sources than the number of virtual sensors. Different from nested array, some holes exist in the difference co-array of the coprime array, and the application of the existing spatial smoothing-based subspace method essentially reduces the usable DOF to approximately one-half of the total number of contiguous co-array elements. To overcome this problem, similar to what we have done in the first part, we extend the mathematical theory of SBL to DOA estimation using coprime MIMO radar, by virtue of the fact that the sparse reconstruction algorithms owns much enhanced adaptation to arbitrary array geometry. It should be noted that, to accelerate the speed to converge to the global minimum, we derive a fast hyperparameter update strategy during the learning procedure based on the majorization-minimization (MM) method. In addition, we circumvent the underlying modeling error incorporated in the spatial scope discretization procedure of the traditional sparsity-driven algorithms in a different manner as it is done for nested MIMO radar, that is, refine the direction parameter set along with the sparse signal such that the parametric dictionary will approach the true sparsifying dictionary.In the third part, we mainly investigate robust adaptive beamforming for MIMO radar. We generalize the traditional phased-array radar’s robust adaptive beamformer design rule to the MIMO radar case, in order to fully exploit the enhanced DOF offered by the virtual array. The essence of our proposed robust adaptive beamforming method is to reconstruct the interference-plus-noise covariance matrix efficiently and estimate the true desired signal steering vector precisely. More specifically, the reconstruction process is performed by projecting the sample covariance matrix into interference subspace or eliminating the desired signal component from the sample covariance matrix. The estimation process is performed by solving a new optimization problem, the objective function of which is minimizing the beamformer sensitivity. Notice that, the solution of the newly constructed optimization problem can be obtained in closed form by using the Lagrange multiplier methodology, which has a comparable computational complexity with that of the standard capon beamformer. After conducting the above operation, we show that the proposed robust adaptive beamforming algorithm can maintain satisfactory adaptation in cases with imprecise signal model. Additionally, the proposed robust adaptive beamforming algorithms are applicable to both nested MIMO radar and uniform MIMO radar scenarios, thus can suppress more interferences than the number of virtual sensors in the aforementioned nonideal signal environment.In the fourth part, we mainly study compressive sensing (CS) MIMO radar imaging based on the fact that the scatterers of the target are often distributed sparsely, and better imaging performance can be obtained by exploiting the arising CS technology. Firstly, considering the linear frequency modulation waveform and frequency hopping waveform, the echo model is respectively established for colocated MIMO radar, and the effect of waveform selection on CS-based MIMO radar imaging using sparse model is also analyzed from a ambiguity function viewpoint. Then, to calibrate the mismatch between the tacit assumption by CS-based algorithm that scatterers exactly lie on pre-discretized grids and the nature that scatterers are arbitrarily distributed continuously in spatial space, we introduce a modified sparse learning via iterative minimization (SLIM) algorithm, which linearizes the off-grid errors using the second order Taylor approximation, to the field of CS-based MIMO radar imaging. Unlike traditional CS-based approach, the proposed algorithm not only gets rid of grid dependence, but also enhances the imaging quality.
Keywords/Search Tags:Multiple-input multiple-output radar, direction-of-arrival estimation, nested array, sparse Bayesian learning, coprime array, robust adaptive beamforming, ambiguity function
PDF Full Text Request
Related items