T/2 soft decision feedback blind equalizer

Vladimir R. Krstić · 2004

T/2 soft DFE is the based on idea of transforming a classical DFE to a cascade of uncoupled linear devices. The crucial component of this cascade is a white filter, which supplies the linear T/2 FSE with uncorrelated samples. The adaptive whiting of input sequence is controlled by a stochastic gradient algorithm based on joint entropy maximization (JEM-W) criteria which introduces soft decision properties at the very beginning of blind equalization. Soft decisions probably smooth the error surface and allow the algorithm to escape from local minima.

Read the paper · More papers on PaperTik