Design and analysis of improved high-speed adaptive filter architectures for ECG signal denoising

Abstract In this paper, improved high-speed adaptive filter-based denoising architectures are proposed and implemented using the Xilinx Virtex-6 series FPGA platform. The performance of the proposed DF-RDLMS, TDF-RDLMS and TF-RDLMS implementations is analyzed and compared with that of the existing adaptive filter architectures as well as state-of-the-art wavelet-based denoising architectures. The proposed adaptive filter implementations are found to perform better than existing adaptive filter architectures and wavelet-based architectures. As compared to the existing adaptive filter implementations, the proposed architectures facilitate design flexibility in choice of the step-size parameter and also allow processing of input signals in fixed-point format as well as floating-point format. Simulation results exhibit the effectiveness of the proposed architectures in efficiently denoising ECG signals in noisy environments and prove their high suitability for low-cost high-performance denoising applications in medical field. Moreover, it is observed that the proposed high-speed adaptive filter architectures require significantly less hardware as compared to state-of-the-art wavelet-based architectures.