Unintentional Radiated Emissions

Unintentional radiated emissions originate from manufacturing process variations and component tolerances that make every circuit physically unique. These microscopic differences in transistor dimensions, interconnect lengths, and dielectric properties cause each device to leak a subtly distinct electromagnetic signature. Unlike intentional transmissions, these emissions are a byproduct of internal switching activity, clock harmonics, and power supply ripple coupling onto unintended radiating structures such as PCB traces and cable shields.

The emitted signature spans a broad frequency range, often extending from kilohertz into the gigahertz region. Key spectral components include:

• Narrowband tones from clock oscillators and their harmonics
• Broadband noise from digital switching transients and power supply ripple
• Modulated leakage where internal data signals inadvertently amplitude or phase-modulate the emitted carrier This composite spectrum forms a highly dimensional feature space for device discrimination.

The emission profile exhibits quasi-stationary behavior—stable enough for reliable identification over short to medium timeframes, yet subject to slow drift from:

• Temperature variation affecting oscillator frequencies and amplifier gains
• Aging effects in electrolytic capacitors and semiconductor junctions
• Supply voltage fluctuations altering switching edge rates Robust fingerprinting systems must implement drift compensation algorithms to maintain authentication accuracy across environmental conditions.

Unlike the intentional modulated waveform used for communication, unintentional emissions are:

• Independent of protocol: The signature persists regardless of the data being transmitted or the modulation scheme in use
• Hardware-intrinsic: Cannot be altered through software configuration or firmware updates
• Involuntary: The device cannot suppress or mask its emissions without physical hardware redesign This makes the signature a compelling physical unclonable function for anti-counterfeiting and supply chain authentication.

The capture and analysis of unintentional emissions differs significantly based on measurement distance:

• Near-field (reactive region): Dominated by capacitive and inductive coupling, providing high spatial resolution for isolating emissions from specific PCB regions or IC pins
• Far-field (radiative region): Propagating electromagnetic waves suitable for stand-off identification at operational distances Practical authentication systems often operate in the transition region, balancing signal-to-noise ratio with operational stand-off requirements.

Capturing usable fingerprints requires specialized instrumentation and signal processing:

• High-dynamic-range receivers with low noise figures to detect weak leakage above the ambient noise floor
• Shielded anechoic chambers for controlled enrollment, eliminating external interference
• Higher-order statistics including bispectrum and cyclostationary analysis to extract features invariant to Gaussian noise
• Time-domain gating to isolate transient turn-on signatures from steady-state emissions The extracted feature vectors are then fed into machine learning classifiers for device identification.