A New Design Paradigm for MIMO Dispersive Channel DF Equalizer

Abstract

We consider the issue of the equalizer design in a modern block-adaptive application scenario where the receiver processing power allows us to implement the filters, designed according to an indirect approach, by using a software logical structure. We propose a new design paradigm that, unlike the classical one, counts that the filter structure is not defined in hardware and can be, therefore, modified in correspondence of each new channel estimation. An important step of the proposed paradigm concerns the choice of the most suited implementation structure for the equalizer filters; such a solution obviously depends on the constraints imposed to the implementation quality parameters (e.g., implementation delay, computational complexity, memory requirements); moreover, the most suited solution also depends on the channel impulse response as well as on signal and noise spectra and, therefore, it is useful to be able to modify it in a time-varying wireless scenario. In particular, when at least one external filter pole is sufficiently close to the unit circle, the forward-backward implementation of the ideal noncausal filter (often discarded in the equalizer design) can provide significant reduction in computational complexity over all alternative filter structures.