Architetture radar adattive per il rilevamento di obiettivi in ​​scenari con carenza di campioni

The main goal of a radar system consists in detecting and locating objects (e.g., air crafts or ships) present in an operating scenario by emitting electromagnetic energy. The energy backscattered from the surrounding environment primarily indicates the presence of an object, which, by comparing the received echo with the transmitted waveform, can be located along with other related information (e.g., size, speed or type). However, under real circumstances, radar behavior is typically affected by interferences that turn the detection stage into a random process, characterized by missed detections and false alarms. Then, to optimize performances, several techniques have been developed by the radar community over the years. Among them, an important role is assumed by the family of adaptive techniques: one of the most powerful is the space-time adaptive processing, that is based on the simultaneous processing of signals received by multiple antennas (space domain) and/or multiple pulses (time domain) of a coherent processing interval. Here, special attention is paid on the suppression (or, at least, the mitigation) of clutter and/or jamming effects. In particular, the matched filter’s output is used in a suitable detection scheme, optimized according to the spectral/statistical properties of the interference. In fact, according to a priori information, it is possible to define a decision rule that guarantees certain performances in terms of detection and, equally important, false alarms. The contributions of the present work can be framed in the context of adaptive tech niques focusing on two main kind of interference environment, i.e., homogeneous and heterogeneous. In the first one, the disturbance given by the environment (free of sig nal components) shares the same spectral properties of the noise in the cell under test; otherwise, it is called heterogeneous environment. i ABSTRACT As for the homogeneous environment, in this work we address adaptive radar detection and localization of point-like targets in Gaussian clutter with unknown covariance matrix, assuming a low volume of training samples. To this end, we first exploit the symmetrically structured power spectral density of the clutter to transfer data from the complex to the real domain. Then, the spillover of target energy is incorporated into the design criteria to come up with two architectures capable of guaranteeing improved detection performances and range estimation. For what concerns heterogeneous environments, instead, we devise four adaptive radar architectures. The first relies on a cyclic optimization exploiting the maximum likeli hood approach in the original data domain, whereas the second detector is a function of transformed data that are normalized with respect to their energy and is obtained by estimating the unknown parameters through an expectation-maximization-based alter nate procedure. The remaining two architectures are obtained by suitably combining the estimation procedures and the detector structures previously devised. The performance assessments of the first two detectors (homogeneous environment), conducted on both simulated data and real recorded datasets, demonstrate the effective ness of the newly proposed detectors compared with the state-of-the-art counterparts that ignore either the clutter spectral symmetry or the energy spillover. Performance analy sis of the last four detectors (heterogeneous environment), conducted on both simulated and measured data, highlights that the architecture working in the transformed domain guarantees the constant false alarm rate property with respect to the interference power variations and a limited detection loss with respect to the other detectors, whose detection thresholds nevertheless are very sensitive to the interference power. The reminder of the work is arranged as follows: in Chapter 1 we introduce the topic of adaptive detection by focusing on the specific mathematical models exploited. Chapters 2 and 3 contain the novelty of the present work, i.e., the derivation and the analyis of a set of detectors with improved detection and localization performance, suitable either for homogeneous or heterogeneous environments. Finally, the last chapter contains the conclusions and introduces possible future implications of this work.