Non-physiological Artifact Elimination in Brain Waves using Adaptive Variable step size Algorithms

Authors

  • M.V.V.S. Prasad , T.Ranga Babu

Abstract

This paper proposes novel adaptive signal processing techniques for brain waves in patient care monitoring applications. Based on Leaky Least Mean Square algorithm using sign regressor and sign of step value and/or sign of signal, these novel algorithms are developed. In this work, we developed various adaptive signal conditioning techniques for BW suitable for remote patient care monitoring. Our proposed hybrid techniques mainly designed for simplicity with respect to complexity. In order to achieve better performance of the artifact elimination process a combination of normalized least mean square algorithm, leaky and variable step size algorithm is utilized. This hybrid version is variable step size leaky least mean square (VSNL2MS) algorithm. The experimental results confirm that this algorithm is better in terms of convergence and filtering ability than the counter parts. Further, to minimize computational complexity, the proposed VSNL2MS is combined with sign-based algorithms. Among the versions of signum based algorithms the sign regressor VSNL2MS based noise canceller is well suited for patient care monitoring systems.

Published

2020-02-28

Issue

Section

Articles