A Study on Performance Improvement of Decision Error Correction of Decision-Directed Equalizers Using Matched Filters

In this paper, we propose a method for detecting and correcting residual symbol decision errors after equalization, rather than focusing on the optimal solutions for equalizers or channel estimators. The proposed structure retains the conventional equalizer and symbol decision process while reconstructing the received signal using the estimated channel impulse response and symbol decision sequence. The core concept involves statistical analysis for the difference between the reconstructed and actual received signals to detect and correct symbol decision errors. Traditional approaches determine the likelihood of symbol decision errors by applying threshold values to the in-phase and quadrature channels, enabling additional symbol error correction. However, in this study, instead of a simple threshold-based approach, we performed matched filtering by computing the complex inner product between the estimated transmission channel response and difference signal in the complex plane. This method enables a more accurate correlation assessment. Thus, we improved the correction capability for erroneously decided symbols and achieved a lower bit error rate (BER) at the same signal-to-noise ratio (SNR) compared with conventional methods.