Morlet Wavelet

The Morlet wavelet achieves the theoretical lower bound of the Heisenberg-Gabor uncertainty principle, providing the optimal trade-off between time and frequency resolution. Unlike the Short-Time Fourier Transform, which uses a fixed window, the Morlet wavelet's Gaussian envelope ensures that the product of temporal and spectral variances is minimized. This property makes it uniquely suited for analyzing transient phenomena where precise localization in both domains is critical, such as isolating the exact onset of a RF fingerprinting preamble.

A defining mathematical property is the direct mapping between wavelet scale and Fourier period. For the standard Morlet (with central frequency ω₀ ≥ 5), the scale-to-frequency conversion is nearly exact:

• Scale-to-Frequency: f = ω₀ / (2π * scale)
• Fourier Period: λ = (4π * scale) / (ω₀ + √(2 + ω₀²)) This allows engineers to interpret a scalogram directly as a spectrogram with adaptive windowing, bridging the gap between classical spectral analysis and multi-resolution wavelet theory.

The Morlet is a complex analytic wavelet, meaning its Fourier transform is zero for negative frequencies. This construction yields two critical outputs:

• Magnitude: Provides the time-varying spectral energy density.
• Phase: Captures the instantaneous phase progression of the signal. In RF fingerprinting, this phase information is vital for detecting subtle hardware-induced phase noise and I/Q imbalance artifacts that purely real-valued transforms would miss.

The wavelet is defined as ψ(t) = π^(-1/4) * e^(iω₀t) * e^(-t²/2). The Gaussian window e^(-t²/2) provides smooth, non-oscillatory decay to zero, eliminating the spectral leakage and sidelobe artifacts caused by the abrupt truncation of rectangular windows. This smoothness is essential for analyzing cyclostationary features in communication signals, where sharp windowing can introduce false harmonics that obscure the subtle periodicities of the modulation scheme.

To be a valid wavelet, the Morlet must satisfy the admissibility condition, requiring zero mean. This is enforced by adding a correction term: ψ(t) = π^(-1/4) * (e^(iω₀t) - e^(-ω₀²/2)) * e^(-t²/2). For ω₀ > 5, the correction is negligible. This condition guarantees that the Continuous Wavelet Transform is invertible, allowing perfect reconstruction of the original signal from its wavelet coefficients—a crucial property for denoising and signal synthesis in emitter identification pipelines.

Unlike the fixed-resolution STFT, the Morlet wavelet's effective window length scales inversely with frequency. At high frequencies (small scales), the wavelet compresses to capture fast transient signal edges with high time resolution. At low frequencies (large scales), it stretches to provide high frequency resolution for analyzing steady-state carrier drift. This adaptive zooming capability makes it the standard basis for extracting both the sharp turn-on transients and the slow steady-state waveform imperfections used in device fingerprinting.

The formal transform that uses the Morlet wavelet as its most common mother wavelet. The CWT maps a one-dimensional time series into a two-dimensional time-frequency representation by continuously varying the wavelet's scale (inversely related to frequency) and translation (time shift). Unlike the discrete wavelet transform, the CWT provides an overcomplete, highly redundant representation that is ideal for visual analysis and feature extraction from non-stationary signals.

A special case of the Short-Time Fourier Transform that uses a Gaussian window function. The Morlet wavelet is mathematically equivalent to a complex sinusoid modulated by a Gaussian envelope, making it the wavelet-domain counterpart to the Gabor transform. Both achieve the Heisenberg-Gabor uncertainty principle's lower bound, providing the optimal trade-off between time and frequency resolution. The key distinction is the Gabor transform's fixed window versus the Morlet's frequency-adaptive window.

The visual representation of the absolute value of CWT coefficients plotted as a function of time and scale. When using the Morlet wavelet, the scalogram provides a multi-resolution heatmap where scale is inversely related to frequency. Key characteristics include:

• High scale values correspond to low frequencies and better frequency resolution
• Low scale values capture high-frequency transients with superior time localization
• Directly used to identify transient events in RF fingerprinting applications

The analytic signal is a complex-valued representation created by adding a real signal to its Hilbert transform (a 90-degree phase shift) as the imaginary component. The Morlet wavelet is inherently analytic because its Fourier transform has zero energy at negative frequencies. This property enables unambiguous extraction of instantaneous amplitude, instantaneous phase, and instantaneous frequency—critical features for identifying the subtle hardware impairments exploited in RF fingerprinting.

A time-frequency reassignment technique that sharpens a CWT scalogram by reallocating coefficients along the frequency axis based on instantaneous frequency estimates. When applied to a Morlet-based CWT, SST concentrates diffuse energy onto the true time-frequency ridges, dramatically improving the readability of component trajectories. This is particularly valuable in RF fingerprinting for isolating the subtle, persistent waveform distortions caused by DAC non-linearities and I/Q imbalance.

The time derivative of the unwrapped instantaneous phase of an analytic signal, representing the dominant frequency at a specific moment. The Morlet wavelet's complex nature enables direct computation of instantaneous frequency from the CWT coefficients. In RF fingerprinting, deviations from the ideal instantaneous frequency trajectory reveal transmitter phase noise and clock jitter—hardware-specific imperfections that serve as unclonable device identifiers.