Eigenbased modeling of nonhomogeneous airborne radar environments

Space-time adaptive processing generally requires estimating a covariance matrix to compute the adaptive weights. Intrinsic to this process is an assumption that the environment is homogeneous, supplying a sufficient quantity of independent and identically distributed samples to train the adaptive filter. However, measured data analysis indicates the airborne radar environment can appear quite nonhomogeneous. This paper considers simple descriptions of nonhomogeneity, demonstrates measured data is nonhomogeneous via analysis of multichannel airborne radar measurements (MCARM) data, and discusses a means to excise outliers from covariance matrix estimation.