A novel recursive Fourier transform for nonuniform sampled signals: application to heart rate variability spectrum estimation.

摘要: We present a novel method to iteratively calculate discrete Fourier transforms for time signals with sample intervals that may be widely nonuniform. The proposed recursive transform (RFT) does not require interpolation of the samples uniform intervals, and each iterative update N frequencies has computational order N. Because inherent non-uniformity in between successive heart beats, an application particularly well suited this is power spectral density (PSD) estimation rate variability. compare RFT based spectrum Lomb–Scargle Transform (LST) estimation. PSD on LST also samples, but greater than Nlog(N). conducted assessment study involving analysis quasi-stationary various levels randomly missing beats. Our results indicate leads comparable performance significantly less overhead complexity applications requiring estimations.