A spectrum matching method for accurate frequency estimation of real sinusoids

Area of Science:

• Signal Processing
• Digital Communications
• Spectral Analysis

Background:

• Incoherent sampling significantly degrades frequency estimation accuracy, particularly with short signal lengths.
• Existing two-step methods struggle to fully mitigate these detrimental sampling effects.

Purpose of the Study:

• To propose and evaluate a novel spectrum matching based frequency estimator.
• To address the challenges posed by incoherent sampling in real sinusoid frequency estimation.
• To compare the proposed method against four existing two-step techniques.

Main Methods:

• A spectrum matching process involving modulation and spectral analysis is employed to eliminate spectral interference.
• A secondary error correction step utilizing Fourier transform refines the frequency estimate.
• Performance is evaluated through simulations comparing the proposed method with four other techniques.

Main Results:

• The proposed spectrum matching method closely approaches the Cramér-Rao lower bound.
• The technique demonstrates superior performance compared to four other methods, especially for short signal lengths.
• No error floor was observed in the simulations for the proposed method.

Conclusions:

• The developed spectrum matching frequency estimator effectively overcomes the limitations of incoherent sampling.
• This method offers a robust solution for accurate frequency estimation, even with limited data.
• The proposed approach provides a significant advancement over existing methods for real sinusoid analysis.