Device-Agnostic Microwave Noise Metrology for Nonlinear Cryogenic Quantum Devices
Source: arXiv:2605.28808 · Published 2026-05-27 · By Andrea Celotto, Alessandro Alocco, Bernardo Galvano, Luca Fasolo, Emanuele Palumbo, Luca Callegaro et al.
TL;DR
This paper addresses the metrological challenge of accurately characterizing noise properties of nonlinear cryogenic microwave quantum devices—such as Josephson Traveling Wave Parametric Amplifiers (JTWPA)—which are key in solid-state quantum technologies for near-quantum-limited microwave signal processing. Existing noise calibrations often place the noise source in series with the Device Under Test (DUT), causing calibration models to depend on assumptions about the DUT's nonlinear internal dynamics, which limits accuracy and generality.
The authors propose a device-agnostic, in situ noise metrology protocol that separates readout-chain calibration from the DUT by substituting the DUT with a controllable noise source using cryogenic microwave switches. This approach jointly combines Planck spectroscopy with a Variable Temperature Stage (VTS) that sweeps thermal noise temperature, and a Short-Open-Load-Reciprocal (SOLR) scattering parameter calibration to refer noise and scattering measurements to the same cryogenic reference planes. The result is a robust, model-independent noise calibration at DUT ports.
They demonstrate the protocol on a JTWPA operated in a strongly nonlinear multimode regime. The method successfully extracts gain and input-referred added noise as functions of pump power and frequency, revealing pump-dependent noise increases consistent with complex nonlinear noise channels. The work presents a methodologically solid and portable approach to accurate microwave noise metrology of quantum devices, overcoming biases inherent in serial configurations and enabling reliable noise characterization without simplified DUT noise models.
Key findings
- Substitution configuration calibration decouples readout-chain calibration from DUT nonlinear dynamics, improving robustness over serial configuration.
- Planck spectroscopy with a Variable Temperature Stage (VTS) enables continuous, precise thermal noise referencing from approx. 120 mK to 2 K within the cryogenic environment.
- Full Short-Open-Load-Reciprocal (SOLR) S-parameter calibration combined with noise calibration aligns noise and scattering parameters at the same cryogenic ports, aiding accurate de-embedding.
- Application to a JTWPA shows pump-power-dependent increase in added noise, exceeding quantum limit at higher powers, illustrating multimode nonlinear effects.
- Model-based serial configurations risk underestimating added noise due to unmodeled noise channels, demonstrated with synthetic data showing measurable bias.
- Using cryogenic switches allows rapid switching between DUT and noise source, facilitating in situ calibration under identical conditions.
- The VTS implementation with a −20 dB attenuator on a thermally isolated stage provides a physically traceable noise source enabling noise power calibration at cryogenic temperatures.
Threat model
The adversary is effectively the nonlinearity and multimode dynamics internal to the DUT that induce energy-dependent noise transmission and scattering channels unknown a priori and not captured by simplified linear calibration models. These cause bias in calibration if the DUT is in the calibration path (serial configuration). The substitution configuration method assumes the DUT cannot be injected with noise during calibration and that the readout chain behaves linearly and stable when the DUT is removed, enabling independent calibration of system gain and noise temperature.
Methodology — deep read
The paper begins with the threat model of a typical active microwave quantum device (DUT) that operates under complex nonlinear, multimode processes and is immersed in a cryogenic environment with distributed microwave components and lossy stages. The adversarial challenge is measurement bias introduced by nonlinear DUT noise processes if the calibration includes the DUT in series and assumes linear noise transmission models.
Data is collected from a custom cryo-electronic setup including a Josephson Traveling Wave Parametric Amplifier (JTWPA) as the DUT, cryogenic microwave switches, a Variable Temperature Stage (VTS) serving as a controllable thermal noise source, and a Nonlinear Vector Network Analyzer (NVNA) that enables simultaneous scattering parameter and noise power spectral density (PSD) measurements.
The core innovation is in the architecture: the DUT and VTS are placed on the cold stage of a dilution refrigerator between cryogenic electromechanical RF switches, enabling quick substitution. The VTS assembly consists of a matched −20 dB attenuator mounted on a thermally isolated copper stage with heater and thermometer, allowing Planck spectroscopy-based continuous noise tuning from ~120 mK to 2 K.
The calibration procedure is a multi-step workflow: (1) place DUT, VTS, and impedance standards between switches, (2) perform Short-Open-Load-Reciprocal (SOLR) S-parameter calibration to define reference planes, (3) measure DUT S-parameters at operating point, (4) measure DUT output noise PSD without injected signals, (5) measure VTS S-parameters, (6) sweep VTS temperature to fit Planck thermal noise model extracting the system gain Gsys and noise temperature Tsys of the readout chain, (7) switch back to DUT and measure output noise PSD, (8) combine calibrated outputs with DUT scattering data to extract device-specific noise observables such as input-referred added noise.
The method separates the calibration of the linear readout chain from the nonlinear DUT behavior, avoiding reliance on model assumptions about internal noise scattering channels which plague serial calibration setups. The theoretical framework is detailed, outlining how serial calibration biases noise and gain estimates when unmodeled energy-dependent noise channels are present. Synthetic data simulations quantify this effect.
As an end-to-end empirical example, the JTWPA is pumped into a nonlinear multimode regime, measuring gain and noise as functions of pump power and frequency. The results reveal physically consistent increases in added noise, validating the device-agnostic calibration approach's sensitivity and accuracy.
The entire protocol relies on meticulous microwave switch insertion losses, coaxial cable thermalization, and a cryogenically compatible S-parameter calibration set, all integrated into a standardized platform for portability and reproducibility. Code and raw data availability are not explicitly discussed but could be requested from the authors. The VTS thermal model and uncertainty budget are ongoing work towards metrological traceability and primary standards.
Technical innovations
- Combining Planck spectroscopy via Variable Temperature Stage with SOLR microwave calibration to unify noise and scattering references in situ at cryogenic device ports.
- Employing a noise source substitution configuration through cryogenic switches to decouple readout-chain noise calibration from nonlinear DUT behavior.
- Analytical characterization of bias mechanisms in serial noise calibration approaches for nonlinear devices with multi-mode scattering.
- Implementation of a matched −20 dB attenuator on a thermally isolated copper stage as an accurately controllable cryogenic noise source.
Baselines vs proposed
- Serial calibration model: input-referred noise underestimation demonstrated quantitatively in synthetic test data (Appendix A) versus substitution calibration that robustly recovers gain and noise without bias.
- JTWPA device: measured input-referred added noise increases from near quantum limit at low pump power to significantly above at high pump power (Fig. 3), illustrating nonlinear excess noise activation.
Figures from the paper
Figures are reproduced from the source paper for academic discussion. Original copyright: the paper authors. See arXiv:2605.28808.
Fig 1: Flow diagram (not at scale) of the scattering of fluctuations impacting
Fig 2: Two placements of the absolute noise reference, labeled REF. (a)
Fig 3 (page 2).
Fig 4 (page 2).
Fig 5 (page 2).
Fig 6 (page 2).
Fig 7 (page 2).
Fig 8 (page 2).
Limitations
- Current implementation is not a primary noise standard; accuracy depends on VTS thermal model fidelity and calibration uncertainty budget under ongoing refinement.
- Thermal gradients inside the VTS and their quantitative impact on noise calibration are not fully characterized and remain future work.
- No adversarial or active attack evaluation of calibration robustness presented; focus is strictly on measurement methodology.
- Demonstrated on a particular JTWPA device; generalization to other nonlinear devices plausible but unproven experimentally.
- Potential errors from imperfect impedance matching and insertion losses require careful characterization outside this initial report.
Open questions / follow-ons
- What is the full quantitative uncertainty budget incorporating VTS thermal gradients and switch losses to assess traceability of measured noise powers?
- How does the protocol extend to multiport, multi-frequency, or more complex device architectures beyond two-port amplifiers?
- Can active adversarial scenarios, such as maliciously altered DUT behavior or intermittent nonlinearities, bias the substitution protocol?
- Is it possible to integrate this calibration approach into automated dilution refrigerator testbeds for high-throughput benchmarking?
Why it matters for bot defense
While this work focuses on noise metrology of nonlinear cryogenic microwave devices for quantum technologies rather than security per se, the core principle of establishing device-agnostic, model-independent calibration protocols is highly relevant to bot-defense and CAPTCHA engineering. Precise, absolute noise characterization with clearly defined reference planes and minimized assumptions enables robust discrimination between device-generated signals and noise, supporting the calibration of trusted entropy sources or physical unclonable functions used in bot detection.
Moreover, the methodology to decouple calibration from complex nonlinear device behavior via substitution and in situ referencing with switches could inspire analogous strategies in CAPTCHA sensing hardware or side-channel noise measurement. Ensuring measurement linearity and avoiding model-dependent errors is critical when calibrating hardware attestation features or physical proofs in adversarial environments. Thus, bot-defense engineers could apply similar substitution calibration and multi-parameter scattering characterization to verify trusted hardware signal integrity against sophisticated malware or replication attempts.
Cite
@article{arxiv2605_28808,
title={ Device-Agnostic Microwave Noise Metrology for Nonlinear Cryogenic Quantum Devices },
author={ Andrea Celotto and Alessandro Alocco and Bernardo Galvano and Luca Fasolo and Emanuele Palumbo and Luca Callegaro and Luca Oberto and Patrizia Livreri and Emanuele Enrico },
journal={arXiv preprint arXiv:2605.28808},
year={ 2026 },
url={https://arxiv.org/abs/2605.28808}
}