Two-parameter power-variable-training STAP

Post pulse compression and partially adaptive multi‐waveform STAP

“…Notwithstanding the variety of ways in which training data can be modified/down‐selected (e.g. [7–10, 14, 15, 17, 19, 20, 23, 25, 26]), the standard SCM estimate is obtained as using the set of primary snapshots in L prime with cardinality n ( L prime ). The exclusion of range indices comprising the CUT and surrounding guard cells from the training data is generally used to avoid including possible moving targets in/near the CUT.…”

“…To further supplement the many robust SCM estimators that have been developed (e.g. [4, 6–26]), the SIMO µ‐STAP formulation in [29] proposed the use of additional training data obtained from the K secondary filters in (2). For example, a ‘no primary’ µ‐STAP form of SCM based only on secondary training data can be realised as which does not exclude the CUT or guard cells, as doing so is pointless because of the range‐smearing effect of the secondary filters.…”

“…[4, 6–29]) with varying trade‐offs, assumptions, and degrees of success. A prominent trend among these is the down‐selection/modification of the training data itself as a means to achieve improved homogeneity [7–10, 14, 15, 17, 19, 20, 23, 25, 26]. In a bit of a departure from these methods, the recently developed μ‐STAP formulation [27–29] involves the generation of additional training data via the application of multiple pulse compression filters that possess relatively low cross‐correlation with the actually emitted waveform.…”