Blind channel equalization based on iterative weighted least-mean squared algorithm
Dongxin Xu, Hsiao‐Chun Wu · 2005
The expectation-maximization (EM) criterion has been widely applied for sparse observed data, such as time-varying wireless channel equalization. Previous EM techniques for joint channel estimation and symbol detection had computational complexity exponentially proportional to channel model order. In this paper, we derive an efficient iterative weighted least mean squared (IWLMS) algorithm, based on EM, for blind equalization. Our new IWLMS algorithm greatly outperforms the popular blind equalization method, based on the constant-modulas criteria, according to Monte Carlo experiments.