Fractional lower-order statistics for efficient adaptive temporal and spatial methods in non-Gaussian environment

Abstract

A bst rwctThis paper nddresscs thc problem of blind equalization for digital communicntiotis using constant tnodulus signals in the prescnce of heavy-tnilcd additive channel iioisc. We compare the performance of R tcrnporal filter antenn;r systciii with the ones obtained with an array of sensors usctl at ihc receiver to copy thc information sequeiicc. First, wc deinonstrale the negative cffecls of channcl tioisc io tlic origind CMA cost function in terms of reliability and crmvcrgence. Thcri, we introduce n new CMA criterion Tor bolh tempornl aiid spatial systetiis bascd on the fractional lower-order statistics (PLOS) of thc rcccived dum.We perform an analytical study of the properties of the iicw cost function nnd we illustrntc its colivergence belinvior through cornputcr simulations.