Comparison of continuous-and discrete-time modelling of polynomial-based interpolation filters
Vesa Lehtinen, Djordje Babić, Markku Renfors · 2004
Polynomial-based interpolation filters provide a flexible means for arbitrary and asynchronous sample rate conversion. They can be modelled with a continuous-time piecewise-polynomial impulse response. In sample rate conversion by rational factors, however, only discrete points of the impulse response affect the output signal. This leads to a discrete-time model of the interpolator. In this paper, properties of the continuous- and discrete-time models in optimisation of filters for sample rate conversion are discussed and compared so that the best model can be chosen for each filter design task. Interpolators optimised using the discrete-time model are optimal when a rational conversion factor is fixed. However, interpolators optimised using the continuous-time model can be more robust against changes of the conversion factor, which property is useful in applications that require flexibility.