Analysis of systematic errors of frequency and damping estimation by interpolated DFT BY1 algorithm

This paper discusses systematic errors of frequency and damping estimation by Bertocco-Yoshida order 1 (BY1) interpolated DFT (IpDFT) algorithm. Systematic errors are caused by the spectral leakage and may be unacceptable high especially for the signal containing small number of oscillations. It is shown by comparison with Cramér-Rao Lower Bound (CRLB) that systematic errors may be significantly higher then estimation variance caused by additive Gaussian noise even for quite low (30-35 dB) signal to noise ratio. Systematic errors depend on signal phase, but quite surprisingly the maximum of those errors for frequency and damping estimation occurs for different signal phase.