Phase Noise

The power spectral density of phase noise creates a distinctive broadening around the carrier frequency. Unlike ideal theoretical signals, real oscillators exhibit a 1/f³ and 1/f² decay profile that varies between devices due to manufacturing tolerances in crystal resonators and phase-locked loop components. This skirt shape—its slope, knee frequency, and spurious tone content—forms a highly discriminative feature for Specific Emitter Identification (SEI).

Phase noise is characterized across different frequency offsets from the carrier, each revealing different physical mechanisms:

• Close-in phase noise (< 1 kHz offset): Dominated by flicker noise in the oscillator's active devices and resonator Q-factor variations
• Mid-range (1 kHz – 100 kHz): Reflects phase-locked loop bandwidth and loop filter component tolerances
• Far-out phase noise (> 100 kHz): Determined by thermal noise floor and buffer amplifier characteristics

Each region provides independent, complementary fingerprint features.

The PLL settling behavior during frequency synthesis creates a transient phase noise signature. When a transmitter changes channels or initiates transmission, the loop's lock acquisition produces a characteristic phase perturbation pattern. This transient response is governed by the loop filter component values—resistors and capacitors with 5-10% manufacturing tolerances—creating a device-specific dynamic fingerprint distinct from steady-state operation.

Crystal oscillators exhibit a frequency-temperature stability curve (typically AT-cut or SC-cut characteristics) that modulates phase noise behavior across operating temperatures. The thermal coefficient and inflection point vary between devices due to crystal blank orientation tolerances during manufacturing. Advanced fingerprinting systems use temperature-compensated baseline models to normalize these environmental effects and extract the invariant device signature.

Discrete spurious frequency components appear in the phase noise spectrum due to power supply ripple coupling, digital clock leakage, and mechanical vibrations. These spurs—their frequencies, amplitudes, and harmonic relationships—are highly device-specific because they depend on:

• PCB layout and decoupling capacitor placement
• Switching regulator frequency tolerances
• Mechanical mounting and crystal package stress

Spur patterns often provide stronger discrimination than continuous noise floor measurements.

A related hardware impairment where a portion of the unmodulated carrier signal leaks through the mixer, creating a DC offset in the baseband constellation. While phase noise is a dynamic, short-term fluctuation, LO leakage manifests as a static origin offset—both originate from the same local oscillator subsystem and together form a composite fingerprint of the transmitter's frequency synthesis chain.

The static difference between the intended and actual carrier frequency of a transmitter, caused by local oscillator inaccuracies. CFO represents the long-term, stable frequency error, while phase noise captures the rapid, stochastic variations around that offset. Both are derived from the same oscillator and serve as complementary identifying features in RF fingerprinting systems.

The spillover of signal energy into adjacent frequency channels caused by non-linear amplification. Phase noise directly contributes to spectral regrowth by broadening the carrier's spectral skirt. When a signal with existing phase noise passes through a non-linear power amplifier, the spectral regrowth pattern becomes a unique, device-specific out-of-band fingerprint that can be analyzed for emitter identification.

A higher-order spectral analysis technique that transforms a signal to reveal quadratic phase coupling, providing a noise-resistant feature space. Bispectrum analysis is particularly effective for extracting phase noise characteristics because it suppresses Gaussian noise while preserving the non-Gaussian phase relationships introduced by oscillator imperfections, making it a powerful tool for emitter identification.

A comprehensive metric quantifying the deviation of measured constellation points from their ideal reference positions. Phase noise contributes directly to EVM by causing angular rotation of constellation points. While EVM aggregates multiple impairments into a single score, advanced fingerprinting systems decompose EVM to isolate the phase noise component as a distinct, device-specific identifier.

An algorithmic mechanism that continuously updates a device's fingerprint baseline to account for slow, natural variation of hardware impairments. Phase noise characteristics drift over time due to temperature changes and component aging. Drift compensation algorithms track these gradual shifts in the phase noise profile to maintain authentication accuracy without requiring frequent re-enrollment of trusted devices.