Adaptive spatial filtering with data-dependent training for improved radar signal detection and localisation in structured interference

Adaptive spatial filtering is applied in modern surveillance radar systems to improve the detection and localisation of desired signals embedded in severe and dynamic interference from clutter as well as jamming. In practical situations of interest, the interference can only be cancelled to sufficiently low levels by significantly compromising the sidelobe response of the adapted pattern. Attempts to effectively stabilise the sidelobes, by means of diagonal loading or rank-reduction for example, gives rise to suboptimal patterns which potentially lead to intolerable degradations in interference suppression. On the other hand, the application of an adaptive pattern with unregularised sidelobes to a test cell containing a signal that is not incident from the system steer direction may cause an unwanted detection or false alarm. This paper proposes a data-dependent training strategy for adaptive spatial filters which provide enhanced interference rejection with a reduced likelihood of false alarms while simultaneously avoiding target self-nulling when the beam is pointed in the signal direction.