A multiple input-output theorem leading to a new equalizer structure.

Abstract

The study of the multiple-input/output inverse theorem (MINT) to realize exact inverse filtering has led to the proposal of new solution types. Digital processing functions of MINT other than the exact inverse filtering are explored. One such realization of MINT is the selective reception, filtering and distribution of signals in a multichannel system. This is demonstrated with the design of equalizers for a speakerphone system. Practical implementation problems associated with common zeros or close zeros in the channels are discussed and different solutions are proposed to address these issues. One of them is the use of the Minimum Norm Solution with system delay. Finally a novel equalizer structure, based on MINT, is proposed for realizing inverse filtering of a single analog transmission channel. Simulations show that its performance is comparable to that of the conventional equalizer based on the Least Square Error criterion.