Research On Direction Finding Approach For MIMO Radar Based On Intelligent Computing

The technology of radar should meet the higher requirements in the complex environment today.In order to meet the requirements of anti-stealth and anti-interference,multiple-input-multiple-output(MIMO)radar has appeared in recent years.The new radar system takes advantage of the technology of MIMO in the field of communication,which contributes to the higher accuracy of detection compared with the phased-array radar.Thus,MIMO radar has an advantage for estimation of parameters.In order to locate the targets accurately,direction finding is an important part of obtaining the location information.Many direction-finding approaches of MIMO radar have been studied for additive Gaussian noise,but their performance can deteriorate in the non-Gaussian noise.Therefore,the direction-finding problem has to be solved in the impulse noise.However,many subspace direction-finding algorithms need to obtain high signal-to-noise ratio and adequate samples to get better performance of direction finding.And these algorithms need extra processing techniques when coping with the coherent sources.In addition,the spatial spectrum function of MIMO radar is more complex in the impulse noise,which is high-dimensional,nonlinear and multimodal.Thus,the peak searching of spatial spectrum will also be a challenge.Intelligent computing is easier to avoid plunging into local optima in a derivation-free and black-box mechanism compared with many traditional optimization algorithms,so it will be an effective method to perform direction finding for MIMO radar.Intelligent computing has the capacity to guarantee the convergence and reduce the computational complexity greatly compared with the grid searching,which will meet the requirement of the real-time direction finding and provide a new idea for direction finding of MIMO radar.Direction finding approaches for MIMO radar based on intelligent computing are studied in this thesis.Several objective functions are designed to solve a variety of problems of direction finding,which includes the problems for impulse noise,inadequate samples,low signal-to-noise ratio,coherent sources and computational complexity.According to the characteristics of each objective function,several new intelligent computing algorithms are proposed after studying a variety of intelligent computing algorithms.The main content of this thesis can be summarized as follows:1.In order to improve the performance of suppressing the impulse noise,the fractional low-order covariance matrix is proposed.Then,the maximum likelihood algorithm based on the fractional low-order covariance is proposed for direction finding of bistatic MIMO radar,because the maximum likelihood algorithm is not sensitive to the snapshots and signal-to-noise ratio.To solve the objective function,a quantum-inspired cat swarm algorithm is devised to acquire the global optimal solution.This approach is able to locate the independent and coherent sources with a small number of snapshots in the impulse noise and has the capacity to obtain high accuracy and success rate of estimation.2.In order to avoid selecting the parameters of the traditional low-order moments and improve the performance of suppressing the impulse noise,the infinite norm normalization is applied to suppress the impulse noise.Then,the weighted signal subspace fitting algorithm based on the infinite norm normalization is proposed to obtain better performance of direction finding for bistatic MIMO radar.To solve the objective function,a quantum-inspired grey wolf algorithm is devised to acquire the global optimal solution.This approach is able to locate the independent and coherent sources in the impulse noise,and obtain higher accuracy and success rate of estimation with a small number of snapshots.3.In order to avoid solving signal and noise subspace by means of eigendecomposition or singular value decomposition in the impulse noise,the propagator method based on the infinite norm normalization is proposed to reduce the computational complexity for direction finding of bistatic MIMO radar.Then,a quantum-inspired cuckoo search algorithm is designed to solve the two-dimensional multimodal searching problem.This approach can accurately locate the sources in the impulse noise,and reduce the computational complexity greatly compared with the grid searching,which will meet the requirement of the real-time direction finding.4.In order to analyze and assess the performance of direction-finding approaches for bistatic MIMO radar,the Cramér-Rao bound of parameters estimation for bistatic MIMO radar is derived in the impulse noise,which generalizes the Gaussian Cramér-Rao bound.