Improvement of empirical mode decomposition under low sampling rate

Empirical mode decomposition (EMD) is a new data analysis method which depends on the exact location of the extrema in the signal. Therefore, the available EMD algorithm only works under high sampling rate in order to find the extrema exactly. Fourier interpolation has been used to solve the problem. This paper discusses the error expression of Fourier interpolation and derives its error upper bound for general band-limited signals, which implies Fourier interpolation yields errors especially near the boundary when the signal is non-integer-period sampled. Motivated by this, a hybrid extrema estimation algorithm based on Fourier interpolation is proposed. Simulation results show that the hybrid algorithm is efficient in improving the performance of the EMD under low sampling rate.