Study On Target Detection Algorithms In The Time Domain Using Wideband Radar

Compared with narrowband radars, wideband radars can provide many advantages, such as, better target detection performance in clutter, more flexible design for waveforms and systems, better spatial resolution, better electromagnetic compatibility and lower probability of interception. Wideband radars become the trend of radar development for its potential applications. However, nowadays the applications of wideband radar are mainly for target imaging, recognition and classification, how to use wideband radar to fulfill the conventional radar task (such as target detection and tracking etc) is of significance in both the theory and application.The work of this dissertation mainly focuses on issues of the wideband radar target detection in the time domain. The main content can be summarized as the following five aspects:(1). Based on a scattering center model, the physical property of wideband radar target return is discussed. We point out that how to deal with the target-scattering function changing caused by rotational motion and scatters'migration through range cells caused by translational motion is a challenging task for wideband radar target detection. The translational motion compensation can be obtained by envelope alignment and initial phase correction. This forms the basis for the following study.(2). The problem of wideband radar target detection in Gaussian noise by one pulse is researched. Firstly, the advantage and problem of detecting targets by wideband radar are discussed; several existing algorithms are reviewed with their advantages and disadvantages analyzed in detail; then we point out that the difficulty of wideband radar detection is the absence of the knowledge about the spatial distribution of the target scattering centers in real applications; To overcome this difficulty we design a novel binary integrator using multiple quantization thresholds, which we term as the modified binary integrator (MBI). Theoretical analyses and simulation results show that MBI is easily implemented and can achieve a more robust performance than the present algorithms(such as energy integration detector and scatter density dependent GLRT detector etc) for sparse scattering target when the prior knowledge of the targets'scattering centers distribution is absent.(3). The problem of wideband radar moving target detection in Gaussian noise based on feature of multiple adjacent pulses is researched. Based on the characteristics of the wideband target return, we point out that the tracks of target with a constant velocity in range-time data space are composed of several straight lines with the same slope. Based on the target and noise models we design a wideband radar Hough detector using Hough transform. The statistical properties of the detector are discussed completely and rigorously. Experimental results show that the detector can integrate the target returns from multiple pulses efficiently, and has a much better performance than the existing noncoherent detectors.(4). The problem of wideband radar target coherent integration detection in Gaussian noise by the returns of multiple adjacent pulses is researched. The conventional envelope alignment method suffers severe performance degradation under low signal-to-noise ratio. To overcome this problem we design a novel envelope alignment scheme based on cumulative distribution function mapping and constrained misalignment. After envelope alignment we propose two detection schemes. In the first one, we integrate the signal from every pulse by the initial phase steering vector estimated from the observed data. In the latter one, we integrate the signal from every range cells using prior training target data based on principal component analysis. Theoretical analyses show that the proposed detection schemes achieve constant false alarm rate (CFAR). Furthermore, in the detection experiments based on measured data, the proposed schemes can obtain better detection performance than the existing coherent detectors.(5). The problem of robust detection in wideband radar is investigated. The steering mismatches during the coherent processing interval and its impact on coherent integration are introduced firstly; then we propose a new detector based on convex programming, which we refer to as improved robust generalized likelihood ratio test (IRGLRT).A detailed analysis on IRGLRT shows that compare with the existing robust generalized likelihood ratio test (RGLRT) detector, it can greatly reduce the computation complexity without loss of detection performance. Finally, experimental results for measured data of three planes show that the proposed algorithm achieves a visible performance improvement with respect to the conventional GLRT, especially in the presence of severe steering vector mismatches.