Data-adaptive training selection for radar applications

In this paper we devise and assess two procedures for selecting secondary data in a very heterogeneous radar scenario including the presence of outliers. The new algorithms are based on the theory of the generalized likelihood function, either one-step or two-step and achieve an effective rejection of the outliers. We also present performance curves showing that, for some operational scenarios with more than one outliers, the new procedures can outperform some previously proposed data selection algorithms.