Related Experiment Videos
A negentropy minimization approach to adaptive equalization for digital communication systems.
1INC, University of California, San Diego, CA 92093-0523, USA. sychoi@ucsd.edu
IEEE Transactions on Neural Networks
|October 6, 2004
Summary
This study introduces a new adaptive equalization method, NEGMIN, which uses approximate negentropy to improve performance over traditional MMSE methods. The NEGMIN equalizer achieves better bit error rate (BER) and requires fewer training symbols.
Area of Science:
- Signal Processing
- Information Theory
- Machine Learning
Background:
- Adaptive equalization is crucial for mitigating signal distortions in communication systems.
- Traditional methods like Minimum Mean Squared Error (MMSE) equalization primarily rely on second-order statistics.
- Limitations of MMSE include suboptimal performance in the presence of higher-order statistical information and nonlinear distortions.
Purpose of the Study:
- To introduce a novel adaptive equalization method, Negentropy Minimization (NEGMIN), for finite-length equalizers.
- To enhance equalizer performance by incorporating higher-order statistics through approximate negentropy minimization.
- To compare the performance of the proposed NEGMIN equalizer against existing MMSE and Adaptive Minimum Bit Error Rate (AMBER) equalizers.
Main Methods:
- Developed a new performance criterion based on minimizing approximate negentropy of the estimation error.
- Utilized non-polynomial expansions of the estimation error to capture higher-order statistical information.
- Investigated two solutions for the NEGMIN equalizer based on normalization parameter ratios, including an MMSE-equivalent solution and a distinct alternative solution.
- Optimized the NEGMIN equalizer by adjusting normalization parameters to maximize output power (variance).
Main Results:
- The NEGMIN equalizer demonstrates superior Bit Error Rate (BER) performance compared to traditional MMSE equalizers.
- The alternative solution of the NEGMIN equalizer exhibits performance characteristics similar to the AMBER equalizer.
- The proposed NEGMIN equalizer requires significantly fewer training symbols than the AMBER equalizer for comparable performance.
- NEGMIN shows increased robustness against nonlinear distortions compared to the MMSE equalizer.
Conclusions:
- Minimizing approximate negentropy offers improved convergence, performance, and accuracy in adaptive equalization.
- The NEGMIN equalizer provides a robust and efficient alternative to existing equalization techniques.
- The proposed method effectively leverages higher-order statistics for enhanced signal processing in communication systems.