Application Of Toeplitz Structure In Signal Detection

Target detection of received echoes is the first priority in radar working.The traditional detection algorithms have higher requirements on the quantity of training samples.However,in the actual radar detection environment,there are many factors that cause the clutter data to be nonhomogeneous.The training samples may not be independent and identically distributed with the cell under test,and it is difficult to collect enough target-free echoes to achieve quantity requirement.In order to solve the decline of detection performance caused by small sample conditions,many research have been exploiting the structural information of clutter.The special structures of the covariance matrix caused by the the system or interference properties can be used to improve the performance of the detector.Incorporating the priori knowledge,there are many adaptive decision schemes that can optimize the detection performance,but there exist a large quantity of causes that bring about uncertainty to make the preset assumptions no longer valid.This paper mainly studies the problem of target detection when the training samples are insufficient and the structure of the received data deviating from the pre-assumed hypothesis.Firstly,this paper introduces several common Toeplitz matrix estimation algorithms for complex Gaussian clutter in the background of signal processing.The error of the estimation method is analyzed by utilized simulation data,and the time cost of the two iterative algorithms is compared.The appropriate estimation method is selected to be utilized for the subsequent classification and detection process.Secondly,adaptive detection of targets based on centrotoeplitz structure is studied.Based on the total adaptability of the adaptive matched filter,the centrotoeplitz structure is applied to the adaptive target detection based on the original clutter spectrum symmetry structure and the Hermitian-Toeplitz structure.Utilizing the simulation data and the measured data to verify the quantity demands reduction of training samples and the performance improvement of proposed method.Then,the application of the covariance structure classification technique in adaptive detection is considered.A covariance structure classification technique based on model order selection is proposed.Based on the hermitian,symmetric,centrohermitian and centrosymmet-ric structure,two structures,Hermitian-Toeplitz and centro Toeplitz,are added.The models are more accurately divided.The performance of the proposed classification technique is verified by simulation data and measured data.Finally,a signal detection system based on the pre-classification and re-detection of Toeplitz structure is designed.Incorporating the Hermitian-Toeplitz and centro Toeplitz structure into the covariance models,a pre-classification and re-detection architecture based on model order selection theory is designed,which includes a classification stage of covariance structure selection and a detection stage of detectors corresponding to different covariance structure.The detection thresholds for the detector bank and the false alarm probability are discussed.utilized the simulation data and the measured data,the improvement of the detection effect of the proposed detection system is proved.