New blind equalization schemes based on constant cross algorithm

Abstract

In this work, we present two new blind equalization schemes suitable for transmitting cross cross-QAM signals. Based on the framework of constant cross algorithm (CXA), a modified CXA (MCXA) is obtained by adding a constellation-matched error (CME) term to the cost function of the CXA. For the MCXA, the CXA term provides global convergence and the CME term helps to decrease the residual error and accelerate convergence. Furthermore, we propose a decision-directed version of the MCXA, called DD-MCXA, which can improves the performance of the MCXA. Preserving the good initial convergence characteristics of the CXA, the DD-MCXA automatically decreases the effect of the CXA errors when the equalizer is near convergence. Both simulation and analysis demonstrate the good performance of the proposed schemes.