Covariance prediction error filter (CPEF) tracking of time‐varying biomedical rhythms

Publisher Website
Google Scholar
Abstract
The tracking of frequency changes in rhythmic data is approached in this paper using an algorithm named Covariance Prediction‐Error Filter (CPEF). It is related to auto‐regressive spectral analysis in thatitobtainscoefficients for an ARM which fits the data being analysed. Matrixrelationshipsaresetupvia forwards and backwards scanning of the data and the required coefficients are found via solution of these matrix equations. Results using simulated sinusoids and a range of biomedical rhythms are presented. These indicate the superiority of CPEF over Fourier Transform, Yule‐Walker and Maximum Entropy methods for analysing short stretches of rhythmic signals. Improvements in frequency spectral shape, peak accuracy andcomponentdiscrimina‐tion have been obtained and hence CPEF can be suitable for instantaneous rate measurement of noisy signals.

This publication has 12 references indexed in Scilit: