A modified DFT-based Frequency Estimator for sinusoidal signals

In this paper, a fine resolution algorithm based on Discrete Fourier Transform (DFT) for frequency estimation of sinusoidal signals is proposed by combining the Candan algorithm with the 2N-point DFT algorithm. The proposed algorithm first employs the Candan algorithm to get a coarse estimate of the frequency, which is used to refine the incoming signal. The refined incoming signal is further input to the 2N-point DFT algorithm and a fine estimate of the possible residue frequency-offset can be well extracted. By taking the advantages of both algorithms, the proposed algorithm can perform better than either of the algorithms. The reason can be attributed to the additional frequency-offset step. As the proposed algorithm must run two component algorithms sequentially, it simply takes the sum of two individual algorithms in complexity. Simulation results show that the Root Mean Square Error (RMSE) of the proposed frequency estimator performs very close to the Cramer-Rao lower bound (CRLB).