An adaptive whitening matched filter/equaliser for the ISI channel

This paper addresses the problem of adaptive linear equalisation for the intersymbol interference channel. The paper derives the minimum variance linear estimator for the input data sequence, namely the Kalman filter. An adaptive law is derived for the Kalman filter when the channel and/or transmit filters are unknown. This adaptive law is derived by recursively minimising a novel cost function closely related to Kullback-Leibler distance. Our simulations suggest global convergence to an open-eye solution for any minimum phase channel, however we do not currently have a proof of this proposition. The performance however is significantly better than the fractionally spaced constant modulus algorithm.