On the optimality of Bussgang and super exponential blind deconvolution methods

Abstract

In this contribution, a generalization of the super exponential blind deconvolution method is discussed. The generalization consists in the definition of a "spikyness" criterion involving nonlinearities rather than only powers. This allows to rephrase Bussgang deconvolution in the framework of super exponential deconvolution using a spikyness criterion which takes into account the pdf of the input series to be recovered. Improved performance is expected when generalized super exponential deconvolution is tuned to suitable optimality criteria.