Spatial Misalignment Exploitation in QKD
A free-space transmitter (Alice) that encodes each quantum state with its own laser diode cannot point all of those diodes in exactly the same direction. The resulting angular misalignment makes the signal states spatially distinguishable: each diode's beam leaves at a slightly different angle, so a threat actor (Eve) who collects the beams and focuses them with her own optics sees each state land on a slightly different spot. She divides that focal plane into two regions and measures each region in a different basis, which means she measures every state in the basis it was prepared in and therefore identifies it outright instead of guessing. She then re-creates the state she has just identified and sends it on to the receiver (Bob), who sees a link that is behaving normally [ArteagaDiaz2022].
Where the beams are fully separated at her focal plane she learns the whole key without raising the error rate. With less separation she learns less and causes errors [ArteagaDiaz2022]. Distance works for her rather than against her: the beams spread as they travel while the angle between them stays the same, so a longer path makes the states easier to tell apart. That makes satellite and long-range terrestrial free-space links the exposed case [ArteagaDiaz2022]. A second, related difference needs no misalignment at all, because each diode's output beam is already slightly oval and tilted in line with the polarisation it carries [Rau2015].
| [ArteagaDiaz2022] | P. Arteaga-Dıaz, D. Cano, and V. Fernandez. "Practical Side-Channel Attack on Free-Space QKD Systems With Misaligned Sources and Countermeasures" In: IEEE Access 10, 82697–82705. (2022) 10.1109/ACCESS.2022.3196677. |
| [ETSI016PP] | ETSI. "Quantum Key Distribution (QKD); Common Criteria Protection Profile - Pair of Prepare and Measure Quantum Key Distribution Modules". In: ETSI GS QKD 016. (2024) https://www.etsi.org/deliver/etsi_gs/QKD/001_099/016/02.01.01_60/gs_QKD016v020101p.pdf. |
| [Huang2019a] | W. Huang, W. Zhang, and Y. Huang. "Elimination of Spatial Side-Channel Information for Compact Quantum Key Distribution Senders" In: J. Electron. Sci. Technol. 17.3, 195–203. (2018) 10.11989/JEST.1674-862X.90416014. |
| [ISOIEC23837P2] | ISO/IEC JTC 1/SC 27. "ISO/IEC DIS 23837-2:2022 Information security - Security requirements, test and evaluation methods for quantum key distribution - Part 2: Evaluation and testing methods". (2022) |
| [Miller2025] | Alexander Miller. "Micius, the world's first quantum communication satellite, was hackable". (2025) arXiv:2505.06532. |
| [Nauerth2009] | S. Nauerth, M. F¨urst, T. Schmitt-Manderbach, H. Weier, and H. Weinfurter. "Information leakage via side channels in freespace BB84 quantum cryptography" In: New J. Phys. 11, 065001. (2009) 10.1088/1367-2630/11/6/065001. |
| [Rau2015] | M. Rau, T. Vogl, G. Corrielli, G. Vest, L. Fuchs, S. Nauerth, and H. Weinfurter. "Spatial Mode Side Channels in Free-Space QKD Implementations" In: IEEE Journal of Selected Topics in Quantum Electronics 21, 187–191. (2015) 10.1109/JSTQE.2014.2372008. |
927ba25e-040c-4128-a37d-0035a4434450
- Quantum Communication
Technique properties
Operational properties
| Description | Countermeasure |
|---|---|
|
Generating the optical pulses with one laser and selecting the polarisation state with a polarisation modulator removes the thing this attack reads, rather than shrinking it or watching for it. The primary states the reason in full: with a single source there are no differences in the wavelength spectrum, spatial mode, propagation direction, or temporal profile among the different states, which eliminates the possibility of side-channel attacks that take advantage of the spectral, spatial, and temporal distinguishability of the different sources [ArteagaDiaz2022]. It is the only control on this row that reaches all three degrees of freedom at once, and the only one that defeats both of this technique's routes, since neither a far-field angular difference nor a near-field source separation exists when there is one emitter. It is also the control whose absence is the row's first precondition, a transmitter using multiple sources. |
CM-0056 Single Laser Diode Utilization |
|
Coupling every quantum state into one single-mode optical fibre before launching it into the free-space channel forces a single spatial mode of propagation, so the states leave the transmitter (Alice) with nothing left to tell them apart geometrically [ArteagaDiaz2022]. This is the row's primary control and the only one whose effectiveness has been measured on real hardware. [Nauerth2009] reports the implemented version on a working eight-diode free-space BB84 transmitter, where a custom design of conical and pyramidal mirrors combines the light of the eight diodes into a short piece of single mode fibre acting as a spatial filter, and measures what survives it: scanning the collimated beam 40 cm after the fibre, no correlation between the bit value and the detection place could be found above the noise level, with the information leakage of the order of 1e-5 bits per pulse, two orders of magnitude below the residual spectral leakage on the same device. [Rau2015] gives the same measure as its sender-side conclusion, that spatial indistinguishability can be ensured for multi-laser diode sources by single mode filtering at the output. |
CM-0055 Single-Mode Fiber Coupling |
|
A security proof that absorbs this leak does not stop the threat actor (Eve) reading the spatial mode; it measures what she learns and shortens the key until that knowledge is worthless. Both sides of this edge already name the same imperfection: CM-0014's own text lists emission of multimode signals, meaning signals that inadvertently encode information in multiple or undesired modes, among the device imperfections a proof must incorporate, and [Huang2019a] supplies the mechanism, stating that the upper bound on the mutual information required to call the spatial side channel eliminated is determined by the QKD security requirement, which depends on the privacy-amplification algorithm. The primary states the matching position from the attack side and makes it a requirement rather than an option: if technological limitations are not to be assumed, increasing the precision of the alignment between the laser sources is not enough, and countermeasures that prevent the attack altogether must be designed [ArteagaDiaz2022]. The practical reading is that this control is how a deployment accounts for whatever spatial residual its other controls leave, for instance the 1e-5 bits per pulse measured after single-mode filtering [Nauerth2009]. |
CM-0014 Full Device Imperfections Incorporation in Scientific Security Proof |
| Description | Category | Impact Type |
|---|---|---|
|
Physical layer, quantum and optical, and the primary target. Everything this attack touches is bulk optics and beam geometry: the per-state emitters' pointing directions and their elliptical mode shapes at the transmitter [Rau2015], the free-space propagation that grows the beam radius while leaving the misalignment untouched, and the threat actor's own collection and focusing optics with a wedge mirror splitting the focal plane into two basis measurements [ArteagaDiaz2022]. No electronics, firmware or control-plane element is involved on either side, which is what separates this row from its spectral sibling A-0082, where the diodes' thermal control loop is a secondary target. |
L1 — Physical Layer (Quantum & Optical) | Primary Target |
|
Post-processing and cryptographic layer, as a secondary effect rather than the target. The attack acts optically at the physical layer, but whether it ends in a compromised key is decided here, in two ways. The threat actor needs the public basis announcement to turn the states she identified into key positions, since Eve will learn about the basis during sifting [Nauerth2009]. And the residual leakage left by any spatial control is absorbed, or not, by privacy amplification, since [Huang2019a] states that the upper bound on the mutual information required to call the spatial side channel eliminated is determined by the QKD security requirement, which depends on the privacy-amplification algorithm. A deployment that measures its spatial residual and prices it here keeps a shorter key; one that does not, claims a key the threat actor may already hold. |
L5 — Post-Processing & Cryptographic | Secondary Effect |
|
Operational security, as a prerequisite the threat actor must already enjoy rather than a layer the attack acts on. She has to occupy the free-space beam path with a large collecting aperture [ArteagaDiaz2022], or else insert optics that lengthen the path between the two parties. ETSI GS QKD 016 places exactly that stretch outside the evaluated boundary, stating that the QKD link can pass through uncontrolled environment without physical protection and does not provide any security services [ETSI016PP], so whether anyone is watching it is an operational matter rather than a module property. It is also the layer the one detect-mode control on this row operates at, since CM-0143 Free-Space Beam-Path Surveillance is itself mapped to L7. |
L7 — Operational Security | Prerequisite for Attack |
| Relation type | Description | Technique |
|---|---|---|
| Refines |
This technique is the conventional intercept-and-resend attack with the basis guess replaced by a side-channel reading; the primary compares its zero-separation limit with conventional intercept-and-resend. At zero separation between the beams at the focal plane the probability of measuring in the correct basis is 50 percent, which agrees with that obtained in a conventional intercept-and-resend attack in which a measurement base is randomly chosen each time [ArteagaDiaz2022]. A-0045 is that attack, and its own description gives the textbook consequence, that measuring a state the threat actor cannot identify disturbs it so the errors appear as soon as the parties compare a sample. What this row adds is the identification: because the spatial mode tells her which basis to use, her error cost falls from the roughly 25 percent of the naive attack towards zero as the separation grows past about three beam radii. The two rows are therefore the same action at two different levels of knowledge, which is worth navigating between because the countermeasure sets differ completely, A-0045's being photon-number and loss-monitoring controls and this row's being source-indistinguishability controls. |
A-0045 Intercept-and-Resend Attack |
| Relation type | Description | Technique |
|---|---|---|
| Alternative to |
A-0081 Spatial Misalignment Exploitation and this technique are two routes to the same outcome against the same architecture, separated by degree of freedom and by whether the difference is found or made. Both attack a free-space transmitter that encodes with several laser diodes, both end in full knowledge of the key, and both carry cat03. A-0081 reads a spatial difference that is already there, caused by angular misalignment of the diodes; this technique induces a spectral difference that was not. [Rau2015] analyses both degrees of freedom in one fielded sender and keeps their countermeasures separate, single-mode filtering for the spatial route and matched-spectra selection with an output filter for the spectral one, which is why the two rows share CM-0056 but not much else. |
A-0082 Wavelength Manipulation Attack in Free-Space |
|