H. Krim et Jh. Cozzens, A DATA-BASED ENUMERATION TECHNIQUE FOR FULLY CORRELATED SIGNALS, IEEE transactions on signal processing, 42(7), 1994, pp. 1662-1668
We present a novel method for estimating the number of signals impingi
ng on a uniform linear array using observed sensor data. Unlike other
algorithms that apply Rissanen's minimum description length (MDL) prin
ciple to the observed data for source enumeration, this method applies
it to the prediction errors of a linear model that has been fitted to
an appropriate data matrix. It is a 1-D method that achieves improved
performance even for fully correlated signals over contemporary appro
aches, particularly with short data records and closely spaced signals
. Asymptotic consistency is shown and substantiating simulation exampl
es are included.