A Simplified Minimum Error Entropy Criterion and Related Adaptive Equalizer Algorithms

초록

The minimum error entropy (MEE) has been successfully applied to equalization, signal processing, classification, state estimation and machine learning under non-Gaussian noise environments. However, the implementation of MEE faces heavy computation caused by double summation operations inherited in the original MEE. To this end, we utilize the fact that statistical expectations or sample means can be replaced with its instant values and propose a new MEE criterion that has no double summations. We also introduce related algorithms for the weight update in the tapped delay line (TDL) filter structure. Experimental results with adaptive equalization for multi-path fading channels with impulsive noise are presented to verify the effectiveness in calculation and performance of the proposed MEE. © 2023, Korean Institute of Communications and Information Sciences. All rights reserved.

키워드