Quantum Channel Dazzling

System

Parent:

Quantum Channel Dazzling floods the receiver of a free-space or satellite quantum key distribution link with bright extraneous light, raising detector noise and the quantum bit error rate until key generation halts. It is the optical-domain equivalent of radio-frequency jamming, and it takes no key material: the link simply stops producing one. The threat actor (Eve) reaches the receiver (Bob) by either of two routes. In the direct route the laser is aimed into the receiver's own field of view, which is only a few to a few hundred microradians wide and leaves the attacker almost nowhere to stand. In the reflective route, which is the one the founding study analyses and the one all three sources treat as the realistic threat, the laser is aimed instead at the transmitting satellite (Alice) from a terminal that can sit a thousand kilometres away from the ground station; the satellite's own structure scatters a fraction of that light downward, and it arrives at the tracking receiver looking like ordinary background [Gozzard2021,Krelina2023,Simmons2023]. Scope here is the free-space and satellite ground-laser case; for the general fibre-and-free bright-light denial technique see Optical Jamming / Bright-Light Channel Denial [A-0119].

Literature 4
[Gozzard2021] D. R. Gozzard, S. Walsh, and T. Weinhold. "Vulnerability of Satellite Quantum Key Distribution to Disruption from Ground-Based Lasers" In: Sensors 21, 23. (2021) 10.3390/s21237904.
[Krelina2023] M. Krelina. "Quantum Communication Countermeasures" (2023) 10.48550/arXiv.2310.08728. arXiv:2310.08728.
[Simmons2023] Cameron Simmons, Yoann Noblet, Alfonso Tello Castillo, Ross Donaldson. "An investigation of jamming in free-space quantum key distribution" In: Proc. SPIE 12335, Quantum Technology - Driving Commercialisation of an Enabling Science III, 123350C (11 January 2023). (2023) 10.1117/12.2644866.
[Tan2024] Vincent Tan. "Framework and Classification of Indicator of Compromise for physics-based attacks" (2024) 10.48550/arXiv.2401.08127. arXiv:2401.08127.
QID: A-0039
UUID: c78aa481-0a33-4db6-9c46-9a5e8b19d151
Tier: T0
Type: Quantum
Domain(s):
  • Quantum Communication
Tactic: Impact
Created: 2024-01-19
Updated: 2026-11-12
Maturity: Theoretical
Technique properties

Attack type: Denial of service
Attack category: Saturation attack
Target party: Receiver, Transmitter
Target function: Quantum layer
Quantum channel: Free space
Security layers: L1 — Physical Layer (Quantum & Optical) L7 — Operational Security
Maturity: Theoretical
Operational properties

Technique → Countermeasures 4
Description Countermeasure

The device-design umbrella, and what it carries here that its two linked children do not. The concrete measures for this technique sit on those children, telescope and receiver optics on CM-0028 and satellite architecture on CM-0029. What belongs at the umbrella is the design and evaluation process around them, which Krelina2023 sets out as a certification recommendation: for each quantum channel receiver, experimentally estimate both the off-axis receiving imprint, that is the in-scattering profile, and the scattering imprint, that is the reflection profile, and build in a system for detecting and localising the source of out-of-field attacks [Krelina2023]. Neither measurement is made today, and without them an operator cannot say where their own receiver sits against the coupling figures Gozzard2021 reports [Gozzard2021].

CM-0030 Quantum Device Optimal Design and TEMPEST

Detector time gating, which opens the detector only during the short windows when legitimate photons are expected. Gozzard2021 lists receiver gating first among its countermeasures and notes that improved satellite-ground synchronisation narrows the gate [Gozzard2021]. Against the attack that paper actually models, the gate does much less than that ordering suggests: the terminal is continuous wave and its excess photons are uniformly distributed in time, so a gate rejects them only in proportion to its duty cycle. Krelina2023 states the limit directly, that the mitigation is less effective against a continuous-wave laser, whereas a dazzling pulsed laser would need to align with the narrow time gate [Krelina2023], and Simmons2023 gives the general form, that time gating relies on the jammer having no access to the timing of the channel and is therefore conditional [Simmons2023].

CM-0026 Time Gating

Moving the session to a path the attacker is not illuminating. Krelina2023 lists rerouting among its counter-countermeasures for this attack, describing it as switching to backup Quantum Key Distribution links or rerouting the quantum connection, and states its limit in the same breath, that it can be particularly challenging in the early stages of quantum networks, where redundancy is minimal, and rerouting consumes substantial resources [Krelina2023]. Simmons2023 supplies a distinct form of the same idea for the satellite case, and it is not the same control as switching paths after detection: launching several satellites into similar orbits means that targeting and jamming a large number of them simultaneously would require separately actuated and targeted telescopes and lasers, with attacker cost scaling roughly with the number of satellites to be jammed, while launch cost can be made less than linear through multiple-launch systems [Simmons2023].

CM-0085 Quantum Network Traffic Rerouting
From parent: Denial

Rerouting is the availability control for denial of a quantum network: when a node or link is denied, the session moves to a backup path rather than the attack being prevented. [Krelina2023] lists it as a counter-countermeasure, describing it as a natural mitigation strategy involving switching to backup Quantum Key Distribution (QKD) links or rerouting the quantum connection, while warning that it can be particularly challenging in the early stages of quantum networks, where redundancy is minimal, and rerouting consumes substantial resources. [Song2026] gives the satellite instantiation and the first quantitative result for it, pre-associating each entanglement session with multiple candidate satellites having overlapping visibility windows so that failover requires no real-time path calculation, and reassigning the session to a backup satellite on an anomalous error reading. Because each attack disrupts only one session, the threat actor (Eve) must compromise all quantum-enabled satellites simultaneously to stop key distribution, and the simulated improvement in cumulative key generation ranges from 35.8% at 250 eavesdropping attempts to 90.2% at 1500. A mega-constellation therefore supplies exactly the redundancy [Krelina2023] identifies as missing in early networks, so the control's value scales with network density.

The ITU-T QKDN resilience series adds three points.

It is two controls, not one. Protection allocates backup QKD modules, links or key relay routes before a failure occurs, and recovery computes a new route after it: the key relay route for protection is pre-set, while the key relay route for recovery can be automatically calculated [ITUY3815]. The distinction is operational rather than terminological, because a pre-set protection switch is fast and a computed recovery is not, and recovery carries its own deadline, since a QKDN controller is recommended to recalculate the key relay route for recovery within the toleration time of the cryptographic application [ITUY3821]. An attacker aiming to exceed that toleration time has a target the prose above does not name.

Redundancy on the same fibre is not redundancy. A protection QKD link is recommended to be deployed in separate optical fibre from the working QKD link [ITUY3821]. Where it is not, one act on that fibre takes both, and this family contains a child, A-0118, whose whole mechanism is one such act.

The standard names what defeats the control outright. Coordinated protection across the quantum and key-management layers is the last resort when the quantum layer has no protection component available, and if the failure of QKD modules and/or links affects all routes involved in key relay, including protection relay routes, the coordinated protection cannot be applied [ITUY3830]. That is the normative form of the limit [Krelina2023] reaches from the attacker's side, and it names the defeating behaviour precisely: not attacking harder, but attacking widely enough to take the backup path with the working one.

CM-0085 Quantum Network Traffic Rerouting
Technique → Security Layers 6
Description Category Impact Type

The physical quantum and optical layer is where the whole attack lives. The threat actor puts photons into the receiver's optical path, either directly through its field of view or by way of the transmitting satellite, and they end up in the single-photon detector: Gozzard2021 computes that about one photon in 10710^{7} incident on the receiver optic from high off-axis angles couples into the detector [Gozzard2021], and Krelina2023 identifies that detector as the most susceptible target [Krelina2023]. Nothing above L1 has to be defective for the attack to succeed, and nothing above L1 is reached, since the chain ends at the detector, before any classical control, post-processing or key-delivery function. Tan2024 fixes the level at which the layer stops working, about 10−1510^{-15} W of received power at the detector [Tan2024].

L1 — Physical Layer (Quantum & Optical) Primary Target

Operational security is a prerequisite layer for this attack rather than a target of it. Krelina2023 makes target acquisition the pivotal constraint, noting that satellite trajectories can be obtained from space surveillance while determining ground-station locations may rely on intelligence work, and lists keeping the locations of fibre routes, optical ground stations and satellite trajectories classified as a counter-countermeasure in its own right [Krelina2023]. Gozzard2021 makes the same point about the transmitting end, observing that Micius is relatively easy to locate because it transmits tracking beacons, which future systems could phase out to counter the threat [Gozzard2021].

L7 — Operational Security Prerequisite for Attack
From parent: Denial

Operational Security is a prerequisite layer for this family rather than a target of it, and five children map it: A-0039, A-0118, A-0120, A-0123 and A-0127. Two controls live there and both are named by the parent's own source as counter-countermeasures in their own right: keeping route and asset locations unknown, since [Krelina2023] states that knowledge of the locations of fibre optic cables, ground optical stations, or the trajectories of quantum satellites is crucial for a successful attack and that keeping such information classified and unknown is therefore imperative, and physical protection of the route and the sites at either end. Where both hold, the four quantum-layer routes that map this layer do not start, A-0039, A-0118, A-0120 and A-0123, and the fifth, A-0127, loses the relay its adversary has to hold. The same source makes the point quantitatively by naming target acquisition as the foremost challenge the attacker faces, ahead of any optical or electronic difficulty. The layer does not reach the other four classical children, A-0124, A-0125, A-0126 and A-0130, because none of them needs location knowledge at all and A-0126's adversary is already inside the site boundary; nor does it reach A-0119, A-0121 or A-0122, which map no operational-security layer of their own.

L7 — Operational Security Prerequisite for Attack

System Control & Orchestration is a secondary target: denial of the classical command-and-control link and of quantum-network service (routing manipulation, costly rerouting in low-density networks) [Krelina2023].

This layer is not only where some of the damage lands, it is where the family's effects are read and acted on, and that reading matters most.

The automation is driven by exactly the two quantities this family moves. The routing control function also performs the rerouting of key relay depending on the status of the key management layer and/or the quantum layer, ensuring the continued operation of the key supply/key relay [ITUY3804], and it fires on two conditions and only two: a quantum bit error rate increasing beyond a threshold in a certain QKD link connected to the relay node, and the residual amount of keys in the KM(s) in the relay node(s) running short below a threshold [ITUY3804]. Those two conditions partition exactly onto this family's two halves, the error-rate children producing the first and the key-exhaustion children producing the second.

And the control plane reads the first of them as an eavesdropper. Alarms and failure diagnosis on QKD links include the increase of QBER implying that eavesdropping is launched against a quantum channel, or the loss of quantum channel is increased [ITUY3804]. Both observables in that sentence are moved by this family: the error-rate children raise the first and A-0120 raises the second. So the defender's automated response is triggered by an adversary who, in every one of these twelve techniques, is stealing nothing.

The response itself is exploitable. Faulty key managers, key-management links, QKD modules and QKD links are recommended to be de-activated, and reconfiguration is made by controlling an optical switch or splitter, or by bringing a backup quantum channel into service [ITUY3804]. An adversary who can induce the alarm can therefore drive the network into taking its own components out of service and into consuming its own backup capacity, which is a second-order denial produced by the defence rather than by the attack. The library copy read is the 09/2020 first edition and later Recommendations cite a 2025 edition, so this note names the Recommendation rather than relying on its clause numbering.

L4 — System Control & Orchestration Secondary Effect

Physical Layer (Quantum & Optical) is the primary target, and it is where seven of this family's twelve children act. Three kinds of act land here. Light is added: laser dazzling or bright-light flooding of the receiver's single-photon detector, which [Krelina2023] identifies as the most susceptible target in a quantum node. Light is removed: the fibre is bent until no light passes or the beam is obstructed, or the insertion loss is quietly raised so the link stays up and starves. Or the light is left alone and a property of the medium is moved, as when the transmittance of a controlled span is varied, or the optical synchronisation channel is injected into so the recovered clock no longer triggers the detector at the right instant.

Where this layer's involvement stops. Permanent damage to the detector is not part of this technique: every child of A-0047 carries reversible effect persistence, and [Krelina2023] places dazzling in the band that has no permanent effect, with blinding, meaning temporary or permanent damage to the detector, and structural damage in the bands above it. Those belong to the Destruction subtree, A-0038 and the laser-damage rows, not here. The same source warns that the threshold between dazzling and blinding is almost impossible to predict, so an operator raising the power leaves this layer's mapping and this family at the same moment.

L1 — Physical Layer (Quantum & Optical) Primary Target

Post-Processing and Cryptographic is where eight of the twelve children land. For five of them the damage is done here rather than on the wire: the abort decision that ends the round is taken in parameter estimation, the authentication tag that an impostor's occupation consumes covers the whole post-processing transcript, the distillation pipeline that a saturation attack overruns is a post-processing function, the key-relay re-encryption a held relay corrupts is another, and so is the key-identifier exchange that A-0130 alters. Those five, A-0124, A-0125, A-0126, A-0127 and A-0130, map this layer as their primary target rather than as a knock-on, and they are also the five that sit on the third of the three attack vectors [Krelina2023] names, the classical command-and-control link, which is why the layer is where they act rather than where they are felt. The other three, A-0121, A-0122 and A-0123, reach it as a secondary effect of an act performed on the quantum layer. It is recorded as secondary at the parent because the majority of routes act physically and reach this layer through the input they feed it, which is the relation [ITUY3815] describes when it separates a quantum-layer failure from a key-management-layer one and treats them as different problems with different controls.

L5 — Post-Processing & Cryptographic Secondary Effect
Techniques Relating to This One 3
Relation type Description Technique
Alternative to

Same weapon, different power band, opposite persistence. The three-band effect model separates dazzling, which is considered more as a countermeasure than a weapon because it has no permanent effect, from blinding, where the damage is permanent [Krelina2023]. A-0039 owns the first band and this row owns the second and third. The boundary is not reliably controllable, since the threshold between dazzling and blinding is almost impossible to predict, so an operator or attacker raising power for a reversible purpose can cross into this row by accident [Krelina2023].

A-0037 Non-Kinetic Physical Attack
From parent: Denial
Alternative to

Denial (A-0047) is the sibling family in the same tactic and produces the same immediate picture, a link that makes no key. The two are told apart by what happens next: a Denial route ends when the attack ends and the service returns by itself, while a Destruction route leaves damage that outlasts the threat actor, so the service returns only when the operator repairs, replaces or restores something. The relation is recorded as an alternative because a threat actor who wants an outage can choose either family, but the sources describe a continuum rather than two clean options, and the crossing cannot be predicted from either side: "the threshold between dazzling and blinding is almost impossible to predict" [Krelina2023], and for a damage attempt the outcome "is thus impossible to predict prior to testing" [Makarov2016]. An operator should therefore treat a Denial event at the top of its power band as a possible Destruction event until the hardware has been re-characterised.

A-0048 Destruction
Preceded by

This acquisition step precedes the Denial technique that consumes its output. The laser weapon system, tracking and targeting capability and deployment platform acquired here are exactly the equipment A-0047 requires, and both rows draw that inventory from [Krelina2023].

A-0022 Acquire Infrastructure
See Also
A-0037 Non-Kinetic Physical Attack
A-0038 Direct Physical Attack
A-0040 Adversary Post-Processing
A-0044 Photon-Number-Splitting (PNS) Attack
A-0045 Intercept-and-Resend Attack