Evaluation of dispersion constants for signed-error blind equalization algorithms

Abstract

Blind equalization isatechnique foradaptive equalization ofacommunication channel without theaidoftheusual training sequence. Stochastic gradient based blind equalization algorithms equalize thedispersive signals byexploitingthehigh-order statistics ofthetransmitted signal which areembedded insomepre-calculated constants. These constants, usually termed asdispersion constants, contain the information about thesize, shape andenergy ofthetransmitted signal. Dispersion constants also provide propel' gain oftheequalizer suchthat themeanfluctuation inadaptation process iszero(orminimum). Constant modulus algorithm (CMA)isawidely-stutdied andprobably themost famous blind equalization method in literature andpractice. Inthis work, wecontribute alittle by providing simple formulas forthedispersion constants used inCMAandsigned-error CMAforPAMandQAMsignals.