Enhanced design of digital filter and its implementation on fpga

Abstract

quotDigital filters are commonly used in biomedical signal processing to remove or suppress noise and interference, reject or pass certain frequency bands, shape the signal, enhance specific frequency ranges, and more. The complexity and challenges involved in processing bioelectric signals have attracted the attention of researchers to contribute in this domain. Electrocardiogram (ECG) signals, acquired by placing electrodes on the human body, are often contaminated with different noise signals in various frequency ranges. Removing or suppressing these noises while retaining useful information related to cardiac activities is a challenging task for digital filter design. Finite Impulse Response (FIR) digital filter-based designs are proposed in this research work due to their attractive properties, such as linear phase delay, stability, and less complexity involved for hardware implementation. newlineIn this work, adaptive-based filter and multiband filter designs are proposed to efficiently eliminate Powerline Interference (PLI) with its sub-harmonics, Baseline Wandering (BW), Electromyogram (EMG), and high-frequency noises from the ECG signal. The first design of FIR digital filter is proposed based on adaptive noise cancellation, considering the non-stationary nature of ECG signals. Various Least Means Square (LMS) algorithms are realized to update the coefficients of the adaptive filter for processing ECG signals. The performance of the adaptive filter is evaluated on the physionet MIT-BIH arrhythmia and MIT noise stress database. In addition to the adaptive filter design, a multiband filter-based FIR filter design is also proposed in this research work. The main advantage of this design is that it eliminates the need for noise measurement, which is required in the adaptive filtering method. The coefficients of the FIR multiband filter are obtained using least square-based optimization algorithms. The performance of the designed multiband filter is evaluated on the Physionet ECG ID database, which contains