PASSIVE RADAR SIGNAL PROCESSING FOR TRACKING OF TARGETS IN CLUTTER
One of the great challenges with passive radar systems, apart from interference of the monitoring signal caused by the direct signal of the illuminator used, is the ground clutter of the transmitted signal. This clutter can mask weak targets and therefore prevent detection and tracking. In passive radar systems, the presence of targets is detected on the basis of the delay between the arrival of the direct signal from an illuminator and the signal reflected on the target by the illuminator. For determination purposes, the monitoring signal is correlated with the direct signal and the Doppler-shifted reference signal so that the targets can be detected and localized in the range-Doppler domain.
In general, the monitoring signal not only contains target reflections but also the direct signal of the illuminator and its clutter, which interferes with the detection of targets. Ground clutter has no Doppler modulations – hence, the clutter may be created by slight oscillations of the illuminator or the slight movement of ground structures. Due to the correlation sidelobes in the range-Doppler domain, these weak modulations of the ground clutter cause great interference in the Doppler range. These can mask weak targets and prevent detection in large parts of the range-Doppler domain with the result that tracking is more difficult.
This problem is clearly illustrated in Figure 2, which shows a range-Doppler map recorded with FHR's passive radar system ATLANTIS. The clutter around zero Doppler is present over the entire distance range. Sidelobes of the ground clutter, which jut well into the Doppler range, are particularly evident at a distance of 4 to 5 km. A cooperative target, the GPS track of which is shown in pink, intersects the zero Doppler line. Target tracking, based on the detections of the passive radar system, is shown in yellow. The target starts with a positive Doppler shift (approx 50 Hz) and changes its direction of movement in such a way that the Doppler shift changes to negative frequencies (approx -100 Hz). The tracking of a non-cooperative target approaching with positive Doppler shift can also be seen. The tracking will be interrupted as soon as the cooperative target crosses the zero Doppler line as detection of the target will not be possible due to the ground clutter and its sidelobes. Tracking will only recommence when the target exits the area around the zero Doppler line.
Various approaches are used to suppress clutter; all are based on the principle of projecting the monitoring signal into a subspace that is orthogonal to the direct signal. In particular, the Extensive Cancellation Algorithm (ECA) can remove interference caused by individual ground clutters from the monitoring signal through projection of the monitoring signal into an orthogonal subspace of the ground clutter.
Due to the relatively large bandwidth, DVB-T-based passive radars display frequency-dependant effects, e.g. frequency-dependant fading due to multi-path propagation. To take account of these effects, a technique was developed in which the projection is carried out in the frequency domains so that ground clutter can be suppressed in a frequency-dependant manner. This allows the suppression of all ground clutter in a single step. The Doppler modulation of the ground clutter must also be taken into account. Consequently, the ECA-CD algorithm (ECA by Carrier and Doppler Shift) was developed at Fraunhofer FHR.
The effectiveness of this algorithm is shown in Fig. 3. The suppression of all ground clutter not only leads to its elimination near the zero Doppler line – the strong sidelobe signals have also almost completely disappeared. This significantly improves detection and tracking with small Doppler shifts. Tracking now continues while the cooperative target is passing through the zero Doppler line.