CFAR subspace detectors with multiple observations in system-dependent clutter background

In this paper, the target detection problem is studied in system-dependent clutter (SDC) background where the clutter and signal of interest (SOI) propagate in the same channels. The received data model applied here, named as multiple observations data model, consists of multiple observations (primary data) and a set of secondary data (free of SOI). Three detectors with the constant false alarm rate feature are designed by using the generalized likelihood ratio, Rao and Wald tests. Interesting findings are: (1) Theoretical derivation proves that these three detectors are equivalent. (2) The primary data size D and the secondary data size K have nearly the same contribution to computational complexity and the probability of detection (PD) is more sensitive to D. (3) For desired probability of false alarm and PD, proper D and K can be selected based on the approximately estimated noncentrality parameters involved in probability density functions. A selection procedure is provided. (4) Based on those estimated noncentrality parameters, it is demonstrated that the relationship between PD and clutter-to-noise ratio is determined by the system response matrix with respect to clutter and SOI. Simulation results validate that the designed new detectors outperform other classic detectors in SDC background.