Zooming Into Precision: The Zoom TFD for High-Resolution Analysis of Non-Stationary Signals

This paper introduces Zoom TFD, a time-frequency decomposition (TFD) method designed for the high-resolution analysis of non-stationary signals, achieving optimal energy concentration, high time-frequency resolution, and inherent cross-term suppression. Traditional TFDs, such as the Wigner-Ville Distribution (WVD) and Choi-Williams Distribution (CWD), often struggle with resolution trade-offs and cross-term interference, while methods like Zhao-Atlas-Marks Distribution (ZAMD) attempt to mitigate these effects at the cost of higher computational complexity. Zoom TFD takes a different approach, integrating adaptive Fourier Transform windowing with a minimization-based spectral selection mechanism. This formulation enables the dynamic refinement of the time-frequency representation by selectively enhancing dominant spectral components while suppressing noise and unwanted harmonics, ensuring a more precise and focused analysis. The effectiveness of the Zoom TFD is assessed against ten leading state-of-the-art TFDs using the Boashash—Sucic Normalized Instantaneous Resolution and Heisenberg uncertainty performance measures. This evaluation is conducted across a variety of experimental and simulated signals, such as frequency-modulated chirps and multi-component signals. The results indicate that the Zoom Time-Frequency Distribution (TFD) consistently outperforms existing methods. It achieves the highest energy concentration and the lowest uncertainty, demonstrating its robustness for precise and interference-free time-frequency analysis. This work establishes the Zoom TFD as a powerful tool for signal analysis, setting a new benchmark for high-resolution time-frequency representation applications, ranging from biomedical engineering to communications.