Evolutions of mathematical statistics and corrections to CFAR and STAP theories
Y.D. Shirman, V.M. Orlenko · 2008
Since Fisher’s maximum likelihood (ML) method is correct only for asymptotically great number of samples, the exile of Bayessian approach from mathematical statistics (MS) hampered development of CFAR and STAP theories. Many heuristic corrections to these theories appeared therefore. But, it seems better to begin creating the generalized Bayessian theory, providing the processing algorithms for fast varying conditions. The new Pareto — Gaussian a priory model of total interference (TI) intensity is therefore reasoned. Investigation of its use in CFAR and STAP theories is begun. The contours are outlined of future combination of such theories with fast progress in “knowledge aided signal processing”.