Exploiting the Trusted Phase Noise Model in CV-QKD

System

The threat actor (Eve) manipulates the classical signal used for phase-reference alignment in CV-QKD systems, exploiting the trusted phase noise model to increase correlations with the secret key while masking her influence.

The manipulation is invisible to the two parties because the only check they run returns one lumped number. Parameter estimation gives them a single total excess noise and no way to see which part of it moved, so the threat actor can take exactly as much extra noise as she frees by shrinking the part they trust, and the total they measure stays where they expect it [Ren2019].

What they are left with is a key the threat actor partly or wholly knows, while they believe it is secret and that they have earned the key rate they are using [Ren2019].

The weakness is about the reference light as it arrives at the receiver rather than about any one piece of equipment, so several different routes reach the same end state: a lower-loss path for the reference pulses, an amplifier, an attenuator held down while the receiver calibrates, a rotation of the pulses the receiver never measures, or interception and re-preparation of the reference [Shao2022,Shao2022a,Zhao2019].

Literature 10
[Pan2026] Yan Pan, Mingze Wu, Heng Wang, Junhui Li, Yun Shao, Jie Yang, Yaodi Pi, Ting Ye, Ao Sun, Lin Jiang, Lianshan Yan, Wei Pan, Yang Li, Wei Huang, Song Yu, Yichen Zhang, Bingjie Xu. "Long-distance discrete-modulated continuous-variable quantum key distribution over 126.56 km of fiber with a local local oscillator" (2026) 10.1038/s42005-026-02718-0.
[Qi2018] B. Qi, C. C. W. Lim. "Noise analysis of simultaneous quantum key distribution and classical communication scheme using a true local oscillator" (2018) 10.1103/PhysRevApplied.9.054008. arXiv:1708.08742. Read in full for the A-0103 review. Origin of the trusted-phase-noise assumption that A-0103 exploits, and the only source in that review's set that reports a measurement on hardware, over 25 km of single-mode fibre. The copy read is arXiv:1708.08742v2, whose footnote 1 on page 4 already sketches the attack; whether that footnote survived into the journal version was not verified, so it supports no field value anywhere in SQOUT..
[Ren2019] S. Ren, R. Kumar, A. Wonfor, X. Tang, R. Penty, and I. White. "Reference pulse attack on continuous variable quantum key distribution with local local oscillator under trusted phase noise" In: J. Opt. Soc. Am. B 36, B7–B15. (2019) 10.1364/JOSAB.36.0000B7.
[Ren2019a] S. Ren, R. Kumar, A. Wonfor, X. Tang, R. Penty, and I. White. "Noise and Security Analysis of Trusted Phase Noise Continuous Variable Quantum Key Distribution using a Local Local Oscillator" In: 2019 IEEE 20th International Workshop on Signal Processing Advances in Wireless Communications (SPAWC), Cannes, France, 2019, pp. 1-5. (2019) 10.1109/SPAWC.2019.8815501.
[Shao2022] Y. Shao, Y. Li, H. Wang, Y. Pan, Y. Pi, Y. Zhang, W. Huang, and B. Xu. "Phasereference-intensity attack on continuous-variable quantum key distribution with a local local oscillator" In: Phys. Rev. A 105, 032601. (2022) 10.1103/PhysRevA.105.032601.
[Shao2022a] Y. Shao, Y. Pan, H. Wang, Y. Pi, Y. Li, L. Ma, Y. Zhang, W. Huang, B. Xu. "Polarization Attack on Continuous-Variable Quantum Key Distribution with a Local Local Oscillator" (2022) 10.3390/e24070992. Read in full for the A-0103 review. MDPI and Unpaywall both refused the PDF, so the article was read first through PubMed Central (PMC9320017) and then from a local copy of its text, rendered from the Europe PMC JATS XML, which the operator supplied on 2026-10-19. Supplies the polarisation route to the trusted phase-noise term, which needs no amplifier, no attenuator and no privileged fibre, only a polarisation controller acting on the phase-reference pulses a practical receiver never polarisation-measures. Cite this paper by section and never by page: the copy this review verified against carries the article's text rather than the publisher's layout, so its pagination does not match Entropy 24(7), 992. All five quotes drawn from it in this batch were checked against that text and every one matched exactly..
[Shao2026] Yun Shao, Yan Pan, Heng Wang, Mingli Zhu, Yaodi Pi, Ting Ye, YaZhuo Jiang, Li Ma, Yang Li, Wei Huang, Bingjie Xu. "Passive Local-oscillator Monitoring Assisted Parameter Estimation for Continuous-variable Quantum Key Distribution" (2026) 10.1088/1742-6596/3293/1/012010.
[Shen2026] T. Shen, J. Ma, X. Wang. "Deep learning-based defense against phase-injection attacks in pilot-multiplexed continuous-variable quantum key distribution" (2026) 10.1364/OE.590610. Read in full for the A-0103 review. Two roles. It adds a fifth lever on the same phase-reference-alignment surface, injecting structured non-Gaussian perturbations into a published pilot tone, with the same signature of biasing the estimate without noticeable changes in the estimated channel parameters. And it settles A-0103's maturity question. Sec. 2.3 adds the perturbation numerically to recorded phase values, but Sec. 3.1 performs it optically, at a threat-actor terminal of a polarisation controller, a polarising beam splitter with a power meter and an FPGA-driven phase modulator, on one pilot pulse in ten over 20 km of fibre, and Sec. 3.2 measures the resulting biased estimates. A-0103 stays theoretical on scope, not on evidence: this route biases the phase estimate, not the trusted phase-noise term A-0103 exploits, and the four trusted-term routes are simulation only. Its own maturity belongs to a pilot-tone row SQOUT does not yet hold..
[Yamano2023] S. Yamano, T. Matsuura, Y. Kuramochi, T. Sasaki, M. Koashi. "General treatment of Gaussian trusted noise in continuous variable quantum key distribution". (2023) arXiv:2305.17684. Read for the A-0103 review (Secs. I and II), from the Zotero library. Cited on A-0103 for a reading that runs against first impressions: making trusted-noise accounting general and proof-independent widens the exposure rather than closing it, because the framework's soundness rests on the trusted quantity being physically beyond the threat actor's reach, which holds for detector efficiency and electronic noise but not for a term parameterised by light that crossed the channel. It generalises how to spend a trusted term, not how to verify that one deserves to be trusted..
[Zhao2019] W. Zhao, R. Shi, D. Huang. "Practical Security Analysis of Reference Pulses for Continuous-Variable Quantum Key Distribution" (2019) 10.1038/s41598-019-54249-0. Read in full for the A-0103 review, resolved through Unpaywall after the Zotero library came up empty. Supplies the measure-and-resend route to the trusted phase-noise term, which Ren2019 examined and rejected as self-defeating; the two results are reconciled rather than ranked, because Zhao2019 removes the cost Ren2019 objected to by spending effort on a Bayesian phase estimator instead of on the pulse intensity. Also the only source in that review's set that gives a run-time test specific to this route, a differential comparison of the phase-compensation error on signal and on reference pulses..
QID: A-0103
UUID: 3835a207-b126-4a3d-98bc-5bb41213692b
Tier: T0
Type: Hybrid
Domain(s):
  • Quantum Communication
Tactic: Execution
Created: 2024-08-19
Updated: 2026-11-13
Maturity: Theoretical
Technique properties

Attack type: Door opener/Lowering security, Key extraction
Attack category: Phase-reference-alignment attack
Target party: Receiver
Target function: Quantum layer
Quantum channel: Fibre optic
Security layers: L1 — Physical Layer (Quantum & Optical) L3 — Firmware & Software Drivers L5 — Post-Processing & Cryptographic
Maturity: Theoretical
Operational properties

Technique → Countermeasures 1
Description Countermeasure

The upgraded trusted phase-noise model prices the margin in advance: the key rate is computed as though the phase-reference path were lossless, so removing loss from it buys the threat actor nothing she has not already been charged for. [Ren2019] proposes it for this attack by name, estimating the tolerable excess noise with a lossless reference channel, and reports that the upgraded trusted model still beats the untrusted one by 43.5 percent in transmission distance and roughly 9.5 times in key rate at 0.14 dB/km. [Ren2019a] gives the figures at the pessimistic setting, about 9 times the key rate and 26.3 percent more distance.

CM-0096 Upgraded Trusted Phase Noise Model
Technique → Security Layers 3
Description Category Impact Type

Primary at the physical quantum and optical layer, because every route acts optically on light in the fibre: peeling the reference pulses into a lower-loss path, amplifying them, attenuating them during calibration, or rotating their polarisation [Ren2019] [Shao2022] [Shao2022a]. The quantity that moves, the phase-estimation-error noise, is itself an optical measurement property of the receiver's detector.

L1 — Physical Layer (Quantum & Optical) Primary Target

Secondary at the post-processing and cryptographic layer, which is where the optical manipulation turns into a security failure. The parties run parameter estimation and privacy amplification honestly on a falsified input, and [Ren2019] gives the reason it cannot be caught there: the procedure returns only the overall excess noise as one value, so a shift between its components is undetectable while the total is unchanged. The result is a key length set from an estimate that credits the threat actor with less information than she has.

L5 — Post-Processing & Cryptographic Secondary Effect

Secondary at firmware and software drivers, because the polarisation route attacks a run-time policy rather than a component: the polarisation-compensation loop's decision to measure part of the reference train and apply one correction to the whole cycle [Shao2022a]. A defender auditing only the optical layer and the key-rate computation will not find that lever, and the repair is a change to this layer.

L3 — Firmware & Software Drivers Secondary Effect
Related Techniques 1
Relation type Description Technique
Combines with

This technique creates the room in which an attack on the signal runs undetected, and the sources count the key that attack takes as the technique's result; in [Ren2019] that attack is an intercept-and-resend. [Ren2019] makes the coupling explicit and quantitative, since the threat actor tunes the splitting factor of her intercept-and-resend until the additional excess noise equals the tolerance she freed by shrinking the trusted term. In that example A-0045 supplies the information gain and this technique the concealment, and the published key fractions are computed for the two together.

A-0045 Intercept-and-Resend Attack
See Also
A-0098 Saturation Attack
A-0099 Manipulation of Vacuum-Noise Estimation
A-0100 Exploiting Imperfect Channel Transmittance in CV-QKD
A-0101 Power Consumption Analysis Attack on State Preparation in QKD Systems
A-0102 CV Detector-Control Attack