Summary

This blog by Rick Lyons presents three time-domain algorithms for estimating the frequency of a noise-free real or complex sinusoid using raw samples rather than FFT bins. It also contrasts these simple estimators with FFT-based approaches and highlights a key principle about the limits of so-called "exact" mathematically derived DSP formulas in practical implementations.

Key Takeaways

• Implement three low-complexity, time-domain frequency estimation algorithms that avoid performing an FFT.
• Compare the accuracy and computational cost of time-domain estimators against FFT-based frequency estimation.
• Evaluate how finite precision, sampling assumptions, and model mismatch affect the practical performance of "exact" DSP formulas.
• Apply the presented methods to both real and complex sinusoids and recognize their domains of applicability.

Who Should Read This

Intermediate DSP engineers and algorithm designers working on frequency estimation for communications, radar, or audio applications who want low-complexity, time-domain alternatives to FFT-based methods.

Still RelevantIntermediate

Topics