Hybrid CFAR techniques for HF radar

Processing of received HF radar signals to detect target echoes generally involves formation of azimuthal beams, range information and Doppler information. Typical data rates in HF radar are high, so conventional processing of these three dimensions using fast Fourier transforms is often performed to enable real-time system performance. This restriction is however being relaxed with modern computer systems. Lees (1987) describes a conventional signal processing scheme for over-the-horizon (OTH) radar. Similar conventional processing could also be applied to HF surface wave radar. Conventional azimuth-range-Doppler (ARD) processing is often closely matched to the target echoes, however this depends on the system architecture, radar waveform, propagation medium and target dynamics. Target detection is usually performed in ARD space. False detections are introduced by radar system noise and environmental/man-made noise and clutter. Lees describes a data whitening process for reducing false alarms rates caused by the background noise and clutter. This noise and clutter is non-stationary and heterogeneous with unknown probability distribution. Hence non-parametric detection techniques are considered where a priori information about the sample distributions is not known. We investigate a number of CFAR techniques for use in HF radar.