Inferensys

Glossary

Power Supply Rejection Ratio (PSRR)

A measure of a circuit's ability to suppress ripple and noise on its power supply rail from appearing at its output, where poor PSRR allows supply variations to modulate the signal and create a unique, environmentally-coupled fingerprint.
Supply chain manager using AI negotiator on laptop, supplier data visible, casual office afternoon setup.
ANALOG CIRCUIT METRIC

What is Power Supply Rejection Ratio (PSRR)?

Power Supply Rejection Ratio (PSRR) quantifies a circuit's ability to prevent power rail noise from corrupting its output signal, a critical parameter in RF fingerprinting where supply-induced modulation creates unique device signatures.

Power Supply Rejection Ratio (PSRR) is the ratio of the change in supply voltage to the resulting change in output voltage, expressed in decibels (dB). It measures how effectively an amplifier, data converter, or regulator isolates its output from ripple, noise, and transient fluctuations present on its DC power rail. A higher PSRR indicates superior isolation.

In RF fingerprinting, poor PSRR becomes an exploitable feature. Supply noise—from switching regulators or digital circuitry—couples into the analog signal path, amplitude-modulating or phase-modulating the transmitted waveform. This environmentally-coupled, device-specific interaction between the power delivery network and the signal chain creates a unique, hardware-dependent signature distinct from static non-linearity.

POWER SUPPLY MODULATION

Key Characteristics of PSRR in Fingerprinting

Power Supply Rejection Ratio (PSRR) quantifies a circuit's ability to prevent ripple and noise on its power rail from corrupting the output signal. In RF fingerprinting, poor PSRR is an exploitable feature, as it allows power supply variations to modulate the transmitted waveform, creating a unique, environmentally-coupled signature.

01

Definition and Mechanism

PSRR is the ratio of the change in supply voltage to the resulting change in the output signal, expressed in decibels (dB). A low PSRR means a significant portion of power supply noise couples directly into the signal path. This coupling occurs through finite output impedance of current sources and incomplete isolation in amplifier stages. The specific frequency response of PSRR is a device-unique trait, varying due to on-chip decoupling capacitance and parasitic inductances.

02

PSRR as a Fingerprinting Feature

A device's PSRR is not a perfect, flat line; it degrades at higher frequencies. This frequency-dependent attenuation curve is shaped by the physical layout and component values of the voltage regulation and bias networks. An attacker or authenticator can inject a known, low-amplitude pilot tone onto the power rail and measure its modulated remnant at the RF output. The amplitude and phase of this remnant, relative to the injected tone, form a unique hardware signature.

03

Environmental Coupling and Drift

Unlike static non-linearity, PSRR-based fingerprints are environmentally reactive. A device's power supply noise is a function of its activity (e.g., CPU load, transmit power) and external factors (e.g., battery voltage droop, temperature). This creates a dynamic signature that reflects the device's instantaneous operating state. While this complicates authentication, it also provides a rich, multi-dimensional feature space that is extremely difficult to clone or replay.

04

Key Metrics and Measurement

Characterizing PSRR for fingerprinting requires more than a single dB value. Critical metrics include:

  • PSRR vs. Frequency: The signature is the entire curve, not a single point.
  • Supply-Induced Jitter: Noise on the supply modulates the zero-crossings of clock buffers, creating a unique phase noise profile.
  • Intermodulation with Signal: The mixing of supply noise with the intentional signal creates deterministic sidebands that are a direct product of the PSRR non-linearity.
05

Distinction from Other Impairments

PSRR is distinct from static non-linearity (INL/DNL) because it is a dynamic, stimulus-dependent impairment. While INL is a fixed deviation from an ideal transfer curve, PSRR dictates how a time-varying disturbance on the power rail maps to the output. It is also distinct from thermal noise, as PSRR-related artifacts are deterministic given a known supply fluctuation. This makes it a complementary feature for multi-modal fingerprinting systems.

06

Exploitation in DACs and ADCs

In a Digital-to-Analog Converter (DAC), poor PSRR on the reference voltage or output buffer supply directly amplitude-modulates the reconstructed waveform. In an Analog-to-Digital Converter (ADC), supply noise couples into the comparator and sample-and-hold circuits, causing gain errors and aperture jitter that are supply-dependent. Time-interleaved ADCs are particularly susceptible, as supply noise modulates the already-present interleaving mismatch spurs, creating a complex, coupled signature.

POWER SUPPLY REJECTION RATIO

Frequently Asked Questions

Explore the critical role of Power Supply Rejection Ratio in RF fingerprinting, where a circuit's inability to perfectly reject power rail noise creates a unique, environmentally-coupled signature that can be exploited for device identification.

Power Supply Rejection Ratio (PSRR) is a measure of a circuit's ability to suppress ripple and noise present on its power supply rail from appearing at its output. It is defined as the ratio of the change in supply voltage to the resulting change in output voltage, typically expressed in decibels (dB). A higher PSRR indicates better isolation. In an operational amplifier or data converter, the internal bias circuits and amplifier stages rely on a clean, stable supply. When the supply rail carries ripple—from switching regulators, digital clock noise, or mains hum—finite PSRR means a fraction of this noise couples into the signal path. This coupling occurs through mechanisms like channel-length modulation in transistors, where drain-source voltage fluctuations directly modulate the bias current, or through parasitic capacitive paths between supply and signal nodes. The result is that the output signal becomes a function not only of the input but also of the power supply's instantaneous voltage, creating an environmentally-coupled modulation that is unique to each physical instance of a chip due to microscopic manufacturing variances in its power distribution network and decoupling elements.

Prasad Kumkar

About the author

Prasad Kumkar

CEO & MD, Inference Systems

Prasad Kumkar is the CEO & MD of Inference Systems and writes about AI systems architecture, LLM infrastructure, model serving, evaluation, and production deployment. Over 5+ years, he has worked across computer vision models, L5 autonomous vehicle systems, and LLM research, with a focus on taking complex AI ideas into real-world engineering systems.

His work and writing cover AI systems, large language models, AI agents, multimodal systems, autonomous systems, inference optimization, RAG, evaluation, and production AI engineering.