Blind equalization with controlled delay robust to order over estimation

Blind identification algorithms from second order statistics, while achieving good performance from limited observation size, are shown to be dependent on an (improbable) accurate estimation of the channel order. An algorithm proposed by Gazzah et al. (see IEEE Transactions on Signal Processing, October 2000) corrects this drawback while providing with only those equalizers with minimum and maximum delays. The latter are known to not have the best equalization performance. Inspired by this approach, we propose a new algorithm that blindly estimates the channel response and computes equalizers with any desired delay, while at the same time, can handle over estimated values of the channel order