Peningkatan Kualitas Sinyal Elektrokardiogram dengan Filter Adaptif Recursive Least Squares

Abstract

This paper presents a comparative analysis of two adaptive filtering algorithms, namely Least Mean Square (LMS) and Recursive Least Square (RLS), for denoising electrocardiogram (EKG) signals contaminated with Gaussian noise. A five-second EKG signal was extracted from a PhysioNet database and artificially corrupted by additive white Gaussian noise. The noisy signal was then processed using both LMS and RLS filters, where parameter tuning was applied to each algorithm to optimize performance. The evaluation metrics used in this study include Mean Squared Error (MSE) and Signal-to-Noise Ratio (SNR). Experimental results demonstrate that the LMS filter achieved its best performance at a learning rate (μ) of 0.01, resulting in an MSE of 0.1218 and an SNR of 4.18 dB. In contrast, the RLS filter provided superior results with a forgetting factor (λ) of 0.99, achieving an MSE of 0.1179 and an SNR of 4.32 dB. These findings indicate that while RLS outperforms LMS in terms of signal recovery accuracy, LMS remains advantageous in computational simplicity and is more suitable for real-time applications. The comparative analysis suggests that the choice between LMS and RLS should consider the trade-off between denoising performance and computational efficiency, especially in portable or embedded biomedical devices.

Keywords: Electrocardiogram, Noise Reduction, Recursive Least Squares, Signal-to-Noise Ratio, Filter Performance