Blind channel approximation: effective channel order determination

Information theoretic criteria have been widely suggested for the determination of the order of an unknown channel. Unfortunately, when used for the estimation of the order of measured microwave radio channels they are very sensitive to the SNR and the number of data sample variations. In order to develop an efficient criterion for the estimation of the order of the significant part of the channel, called the effective channel order, we use an approach based on numerical analysis arguments. The derived criterion provides a "maximally stable" decomposition of the range space of an "over-modeled" data covariance matrix into signal and noise subspaces. It is shown to be robust to the SNR and number of data sample variations. It provides useful estimates, leading to sufficiently good blind approximation/equalization of measured real-world microwave radio channels.