Asymptotic efficiency of a blind maximum likelihood sequence detector

We consider the performance of blind maximum likelihood sequence detection (MLSD) when the recursive least-squares (RLS) algorithm is used to update channel estimates. We employ asymptotic efficiency analysis to characterize the performance of the detector as the signal-to-noise ratio (SNR) approaches infinity. Asymptotic efficiency analysis allows us to quantify the loss in performance due to the presence of inter-symbol interference (ISI) and the lack of channel knowledge. We show that, under certain conditions, the asymptotic efficiency of the detector depends only on a single most-likely noise realization. Our results indicate that the performance of the RLS-based detector is strongly dependent on both the magnitude of the ISI and the number of data samples available.