Study Of Adaptive Detection For Radar Targets

Detecting and tracking targets are fundamental tasks for radar.Radar technology has been developed rapidly with the diversified targets and complex environment.The range resolution turns from low resolution to high resolution,the channel structure turns form low channel to high channel,and the spatial dimension of the signal to detect turns from one dimension to multiple dimensions.In this dissertation,we consider the problems of the lack of the training data,the mismatch between the actual and nominal steering vectors,the nonhomogeneous environment and the invalidation of the point target model and then discuss the adaptive detection in the limited-training case,the adaptive detection in the nonhomogeneous environment,the adaptive detection in the mismatched signal case,the adaptive detection of extended targets.The detailed description is as follows:1.In order to overcome the detection degradation for the conventional detectors in the limited-training environment,two detectors are proposed by modeling the disturbance as an autoregressive process with unknown parameters.The first detector proposed for the point-like target is derived resorting to a two-step design procedure: first derive the generalized likelihood ratio test under the assumption that the parameters of the autoregressive process are known.Then,the maximum likelihood estimates of the parameters,based on the training data,are substituted in place of the true parameters into the test.The detection performance of the new receiver shows that the proposed receiver can lead to a noticeable performance improvement over the conventional detectors.The second one proposed for the range-spread target is derived by resorting to the Rao design criterion.Meanwhile,the asymptotic expressions for the probabilities of false alarm and detection are derived in closed form,which show that the newly proposed detector is asymptotically constant false alarm rate with respect to the disturbance covariance matrix.Experimental results have demonstrated the effectiveness of the detector.2.The problem of detecting distributed target in Gaussian disturbance with unknown covariance matrix is dealt with.The partially-homogeneous environment is considered wherein the primary data and the training data share the same clutter covariance matrix structure while having different power levels.Two receivers based on the Rao test and the Wald test design criteria are derived at the design stage assuming that both the covariance matrix of the clutter and the steering vector have the persymmetric structure.The performance assessment conducted by resorting to simulated data and real data,show that the proposed detectors can improve the matched detection performance with a small set of secondary data.Additionally,one of the proposed detectors exhibits better rejection capabilities of mismatched signals than the previously proposed detectors,while the other is robust under mismatched conditions.3.The problem of adaptive detection of subspace signals embedded in thermal noise and clutter that depends on the transmitted signal is considered.At the design stage,we assume that the signal-dependent(SD)clutter shares the same subspace as the target signals.As customary,a set of secondary data,free of signal components,is also assumed available.Three adaptive detectors are proposed by resorting to the Rao test,Wald test and two-step likelihood test design criteria.Unlike the classical tests,which are derived by dividing the complex parameter into the real and imaginary parts,the proposed detectors treat the complex parameter as a single quantity to reduce the computational burden.Moreover,we derive the theoretical false alarm probabilities and detection probabilities which show that all the three proposed detectors exhibit the constant false alarm rate(CFAR)property.Simulation results demonstrate that the proposed detectors achieve a detection performance improvement over the conventional multidimensional detectors.4.In order to deal with the adaptive detection problem in the presence of mismatch between the actual steering vector and the nominal steering vector,mismatch selective detectors for the point-like targets in the partially homogeneous are proposed.Mismatch of the signal steering vector may exist due to many factors such as multipath propagation,array calibration uncertainties,beampointing errors.To enhance the selectivity,we add a fictitious interference which is orthogonal to the target steering vector in the quasi-whitening observation space under the null hypothesis.Then,we resort to the likelihood design criterion to design the one-step and two-step mismatch selective detectors.Numerical results show that the proposed detectors exhibit better mismatch rejection capabilities of mismatched signals than the traditional ones.5.Considering the mismatch between the actual steering vector and the nominal steering vector,we propose mismatch selective detectors for the range-spread target in the compound-Gaussian clutter environment.We assume that the target signal belongs to a subspace and add fictitious subspace signals that are orthogonal to the steering vector in the truly-whitening observation space under the null hypothesis.Then we assume that the covariance matrix is known and resort to the generalized likelihood ratio test and maximum a posteriori design criteria to derive the range-spread target detectors when the texture follows the gamma distribution and inverse gamma distribution.(1)The fixed point covariance matrix is calculated and substituted into the two test statistics with known covariance matrix and two adaptive mismatch selective detectors are obtained.(2)We assume that the covariance matrix follows the inverse Wishart distribution and derive maximum a posteriori estimate of the clutter covariance matrix.Then,the estimate of the covariance matrix is substituted into the generalized likelihood ratio test statistic with known covariance matrix and the adaptive Bayesian mismatch selective detectors are obtained.The selective detectors exhibit better mismatch discrimination capabilities of mismatched signals than the traditional detectors.Meanwhile,the adaptive Bayesian mismatch selective detectors outperforms the other two proposed mismatch selective detectors for matched signals.