Exploiting Imperfect Channel Transmittance in CV-QKD

System

The threat actor (Eve) can exploit an imperfect channel transmittance model in continuous-variable quantum key distribution (CV-QKD) systems to overestimate the secret key length by biasing parameter estimation through intercept-and-resend attacks, leveraging issues like atmospheric turbulence and background noise.

Update ([Usenko2012,Castilloceleita2025]): Two conditions bound this attack, and naming them matters as much as naming the attack. First, the transmittance must fluctuate within the block of data over which the transmitter (Alice) and receiver (Bob) estimate their parameters. Second, and more important, the parties must have left the fading out of their key-rate calculation. Parties who compute their security bound from the transmission statistics they actually observed are covered, and no cited source shows the threat actor gaining anything against them. So the exposure is not created by the turbulence; it is created by a security analysis that treats a fading link as a steady one.

Literature 8
[Castilloceleita2025] Miguel Castillo-Celeita, Matteo Schiavon. "The role of averages in CV-QKD over fast fading channels" (2025) 10.48550/arXiv.2511.12721. arXiv:2511.12721.
[Chai2020] Geng Chai, Peng Huang, Zhengwen Cao, Guihua Zeng. "Suppressing excess noise for atmospheric continuous-variable quantum key distribution via adaptive optics approach" In: New Journal of Physics 22, 103009 (2020). (2020) 10.1088/1367-2630/abb47c.
[Guo2017] Y. Guo, C. Xie, Q. Liao, W. Zhao, G. Zeng, and D. Huang. "Entanglement-distillation attack on continuous-variable quantum key distribution in a turbulent atmospheric channel" In: Phys. Rev. A 96, 022320. (2017) 10.1103/PhysRevA.96.022320.
[Huang2019b] W. Huang, Y. Mao, C. Xie, and D. Huang. "Quantum hacking of free-space continuous-variable quantum key distribution by using a machine-learning technique" In: Phys. Rev. A 100, 012316. (2019) 10.1103/PhysRevA.100.012316.
[Huang2021] Duan Huang, Susu Liu, Ling Zhang. "Secure Continuous-Variable Quantum Key Distribution with Machine Learning" In: Photonics 8, 511 (2021). (2021) 10.3390/photonics8110511.
[Kish2024] S. P. Kish, C. Thapa, M. Sayat, H. Suzuki, J. Pieprzyk, S. Camtepe. "Mitigation of Channel Tampering Attacks in Continuous-Variable Quantum Key Distribution" (2024) 10.1103/PhysRevResearch.6.023301. arXiv:2401.15898.
[Ruppert2019] L. Ruppert, C. Peuntinger, B. Heim, K. Guenthner, V. C. Usenko, D. Elser, G. Leuchs, R. Filip, C. Marquardt. "Fading channel estimation for free-space continuous-variable secure quantum communication" In: New Journal of Physics 21, 123036 (2019). (2019) 10.1088/1367-2630/ab5dd3. arXiv:2011.04386.
[Usenko2012] V. C. Usenko, B. Heim, C. Peuntinger, C. Wittmann, C. Marquardt, G. Leuchs, R. Filip. "Entanglement of Gaussian states and the applicability to quantum key distribution over fading channels" In: New Journal of Physics 14, 093048 (2012). (2012) 10.1088/1367-2630/14/9/093048. arXiv:1208.4307.
QID: A-0100
UUID: 79fa3f0c-1e26-47cd-86ae-acf35faac25b
Tier: T0
Type: Quantum
Domain(s):
  • Quantum Communication
Tactic: Execution
Created: 2024-08-19
Updated: 2026-11-11
Maturity: Theoretical
Technique properties

Attack type: Door opener/Lowering security, Key extraction
Attack category: Intercept-and-resend attack
Target party: Receiver
Target function: Quantum layer, Key management
Quantum channel: Fibre optic, Free space
Security layers: L1 — Physical Layer (Quantum & Optical) L5 — Post-Processing & Cryptographic
Maturity: Theoretical
Operational properties

Technique → Countermeasures 2
Description Countermeasure

Proper parameter estimation charges all the noise the parties observe to the threat actor. Over a fading channel this means computing the security bound from the observed covariance matrix, averaged over the channel transmittance, with the noise that the fading adds included. Gaussian optimality then still bounds the threat actor [Castilloceleita2025], and the averaged estimate stays secure because it overestimates the threat actor's information [Kish2024]. No source computes a gain for the threat actor against parties who estimate this way.

CM-0007 Proper Implementation of Quantum Communication Protocol and QBER Estimation

Keeping the transmittance estimate close to the truth is a real defence against this technique, because the attack lives on the gap between the modelled channel and the actual one. Regular and randomised recalibration narrows that gap.

It helps, and it does not close the exposure. On a fast-fading link no recalibration cadence tracks the instantaneous transmittance at all: [Castilloceleita2025] defines that regime by contrast with "the slow fading channel, in which the transmittance is stable long enough to allow an accurate estimation". Where the link fades faster than it can be measured, the honest repair is to stop assuming a single transmittance, which is what CM-0159 does and why that control is linked alongside this one.

CM-0095 Proper Quantum Channel Calibration
Technique → Security Layers 2
Description Category Impact Type

Physical Layer marked prerequisite rather than secondary, and the distinction is deliberate. The attack needs two things from this layer and damages neither. It needs the channel to fluctuate, because "the transmission coefficient of a free-space channel is unstable and fluctuating randomly in time due to the turbulent atmospheric environment" [Guo2017], and it needs in-beam optical access so the threat actor can tap the signal and resend. Nothing optical is defective and nothing optical is altered: the receiver detects correctly throughout, which is why [Guo2017] places its own subject beyond the scope of the imperfections of practical devices. So the layer supplies conditions for the attack rather than suffering an effect from it.

L1 — Physical Layer (Quantum & Optical) Prerequisite for Attack

Post-Processing and Cryptographic is the primary layer. The object the attack turns on is the arithmetic that converts measured quadratures into a key-rate decision and into an amount of privacy amplification: [Guo2017] measures the overestimate as a gap between two computed key rates, and ETSI GS QKD 005 clause 4.4.1(e) places the obligation to quantify model deviations in this layer by requiring that they be taken into account during privacy amplification. The repair is made here too, by carrying the fading term or by binning the block before estimating. The linked base vulnerability puts its primary layer in the same place.

L5 — Post-Processing & Cryptographic Primary Target
Related Techniques 1
Relation type Description Technique
Refines

This technique is a variant of the intercept-and-resend attack, and says so in its own terms: [Guo2017] describes its strategy as an intercept-resend attack performed by applying an entanglement-distillation operation to the transmitted non-Gaussian mixed states. The chain is the familiar one, tap the signal, measure, decide, regenerate and resend with an electro-optic modulator. What makes it a distinct variant is where the cover comes from. A-0045 silent route buys its cover from the channel loss budget by sending vacuum when unambiguous discrimination fails, and A-0097 and A-0098 buy theirs by injecting light that moves a quantity the receiver measures. This variant pays nothing: the cover is the parties own mis-modelled fading, so the threat actor alters no device. That is also why it needs the non-Gaussianity window, which is a condition none of the other variants has.

A-0045 Intercept-and-Resend Attack
See Also
A-0098 Saturation Attack
A-0099 Manipulation of Vacuum-Noise Estimation
A-0101 Power Consumption Analysis Attack on State Preparation in QKD Systems
A-0102 CV Detector-Control Attack
A-0103 Exploiting the Trusted Phase Noise Model in CV-QKD