Fast blind channel shortening using a prediction-error filter aided by Autocorrelation Minimization

A hybrid algorithm for blind adaptive channel-shortening of ADSL communication channels is here proposed. The prediction-error filter is a well-known technique that can equalize minimum-phase channels for Multi-Carrier Modulation (MCM) modulated signals. Another well-known algorithm, Sum-Squared Autocorrelation Minimization (SAM), also suited to blind adaptive channel-shortening of MCM signals, is used to aid the prediction-error filter. SAM exhibits fast convergence, but has high computational cost and an unstable behaviour. The objectives of the hybrid algorithm are fast convergence and stable steady-state behaviour for modelled ADSL channels from one channel-shortening algorithm; we show the performance of the hybrid fulfils the objectives.