Implementation of swarm intelligence based optimized adaptive filtering technique for ECG data analysis system

Abstract

In biomedical signal processing, the removal of noise is one of the newlineimportant challenges faced to avoid medical information loss. The ECG newline(Electrocardiogram) signal is the most important signal used to diagnose the newlinewellness of heart s activity. Adaptive filters find wide application in several newlinebiomedical signal processing and communication units. The ECG pre-filters newlineare used to remove noises. This improves signal to noise ratio and enhances newlinethe estimation process in ECG signal. Several architectures are implemented newlineand presented in the literature for adaptive filters implementation. The major newlinecomponents are decimators, interpolators, delay elements, multipliers and newlineadders. Optimized designs are required for the processing elements with less newlinearea and power consumption. The existing adaptive algorithm cannot be newlineapplied with the multimode error surface. To minimize the cost function, this newlinework uses an approach by combining MRMN algorithm with ABC algorithm. newlineThe LMS algorithm fails to converge when impulsive noise is more in the newlinesignal. To enhance the convergence behaviour the LMS algorithm is newlineprocessed using the MRMN algorithm. The ABC algorithm has been solved newlineby combinatorial process and uni-modal/multimodal numerical optimization newlinewith MRMN algorithm. newlineIn this research, a novel VLSI architecture for adaptive filter design newlineusing Robust Mixed Norm (RMN) algorithm and Ant Bee Colony newlineoptimization is proposed. Swarm based methods were used to optimize the newlineconvergence behaviour of the adaptive filter. In this work, implementation of newlineadaptive filtering using swarm-based optimization techniques for biomedical newlinesystem on chip architecture is presented. The investigation for different newlinearchitectures compared with the proposed method shows its better newlineperformance. newline