Abstract: In this paper an adaptive noise cancelation (ANC) model is presented to remove baseline wander (BW) noise from mathematically modeled ECG signals. The ANC model is designed to have a trade-off between the correlation properties of noise and reference signals. Matlab is used to simulate ECG signals artificially, to represent different sinus rhythms and leads of ECG waveform. Furthermore contamination of an important artifact (baseline wander) is simulated for normal ECG lead II, and then identified using LMS algorithm and its preconditioned versions: NLMS and TDLMS algorithms, to get denoised ECG signals. Experimental results are presented for a comparison of these adaptive algorithms, which shows preference of TDLMS algorithm and Fx LMS algorithm over the rest.
Keywords: Mathematical model, almost periodic, adaptive algorithm, adaptive noise canceler.