System identification via hybrid FIR-IIR adaptive filtering

This paper present an adaptive scheme applied to system identification, which involves both FIR and IIR adaptive filters. In the current work the scheme operates in a two-stage process for the identification of an unknown system having long impulse responses (IIR). The first part performs identification via an FIR adaptive filter, which gives a rough approximation to the unknown system. The second part shows the application of the Balanced Model Truncation (BMT) technique which maps the coefficients of the FIR filter obtained from the first step, into a set of coefficients that initialise the IIR filter to undergo adaptation. The further adaptation of the BMT initialised IIR filter performs fine-tuning on the system identification