Adversary Post-Processing
Adversary Post-Processing is a technique where the threat actor (Eve), having surreptitiously obtained the raw key material during a quantum key distribution (QKD) session, mimics the legitimate post-processing steps performed by the original communicating parties to derive the same final encryption key.
Threat actors may exfiltrate data from the raw quantum data, such as keys from the QKD, through the application of post-processing (key distillation) after being gathered using obtained information during Reconnaissance.
The input she works from is that raw material combined with the basis information the parties announce, which together give her a sifted key that nearly matches the transmitter's. From there the step is cheap, because the two things she still needs are published. The transmitter (Alice) and the receiver (Bob) announce the error-correction syndrome and the privacy-amplification function on their public classical channel, so the threat actor reads both and applies them to her own copy [Pacher2016]. What she ends with is not a partial guess but the same key the two parties believe is theirs alone, which the source calls the strongest possible attack [Pacher2016].
| [ETSI008] | ETSI. "Quantum Key Distribution (QKD); QKD Module Security Specification". (2010) https://www.etsi.org/deliver/etsi_gs/QKD/001_099/008/01.01.01_60/gs_QKD008v010101p.pdf. |
| [Gerhardt2011] | I. Gerhardt, Q. Liu, A. Lamas-Linares, J. Skaar, C. Kurtsiefer, and V. Makarov. "Full-field implementation of a perfect eavesdropper on a quantum cryptography system" In: Nature Communications 2, 349. (2011) 10.1038/ncomms1348. |
| [Lorunser2023] | Thomas Lorünser, Stephan Krenn, Christoph Pacher, Bernhard Schrenk. "On the Security of Offloading Post-Processing for Quantum Key Distribution" (2023) 10.3390/e25020226. arXiv:2210.08977. https://www.mdpi.com/1099-4300/25/2/226. |
| [Lutkenhaus1999] | Norbert Lütkenhaus. "Estimates for practical quantum cryptography" In: Phys. Rev. A 59, 3301. (1999) 10.1103/PhysRevA.59.3301. |
| [Pacher2016] | C. Pacher, A. Abidin, T. Lorünser, M. Peev, R. Ursin, A. Zeilinger, J.-Å. Larsson. "Attacks on quantum key distribution protocols that employ non-ITS authentication" (2016) 10.1007/s11128-015-1160-4. arXiv:1209.0365. |
944bef51-de9e-44f4-ba7d-dca5709100e1
- Quantum Communication
Technique properties
Operational properties
| Description | Countermeasure |
|---|---|
|
Encrypting the classical post-processing exchange end to end removes this technique's input rather than detecting or containing the attack. The measure has a primary source: Lutkenhaus1999's error-correction design has the transmitter compute the extra parity bits, and then "encodes them with secret bits and sends them via the classical channel to Bob" [Lutkenhaus1999]; the countermeasure's own mitigation steps extend the same treatment to the parameter-estimation information, using pre-shared seed randomness on the first run and generated key thereafter. With the syndrome and the privacy-amplification function unreadable, the threat actor is left holding a sifted key she cannot distil [Pacher2016]. |
CM-0077 End-to-End Encryption for Classical Channels in Quantum Communication Systems |
| Description | Category | Impact Type |
|---|---|---|
|
Primary target. The technique replicates the classical post-processing and cryptographic layer: it reads the error-correction syndrome announced there and uses it to correct its own sifted key, then reads the privacy-amplification function and computes the same final key from it [Pacher2016]. Both steps sit wholly inside this layer, alongside the sifting, reconciliation, confirmation and privacy-amplification stages they consume [Lorunser2023]. Defeating this layer's authentication is a different technique, catalogued as A-0152, and is not claimed here. |
L5 — Post-Processing & Cryptographic | Primary Target |
|
Prerequisite. What has to come from upstream, whether through an operational-security failure or through reconnaissance, is the matching sifted key, plus one piece of configuration knowledge: which reconciliation code and which privacy-amplification family the deployment runs, so that the threat actor's own implementations match. The per-run parameters do not: the error-correction syndrome and the privacy-amplification function are announced on the public classical channel and simply read [Pacher2016], the security model treats that function as "part of Eve's knowledge in each realization" rather than as a leak [Lutkenhaus1999], and where it is not announced in its own step it is reconstructible from basis information the threat actor already holds [Pacher2016]. This is why the row carries target_recon_required low rather than high, and characterisation_dependency none. |
L7 — Operational Security | Prerequisite for Attack |
|
Secondary effect, and read only. The sifting and basis-reconciliation announcements traverse the system control and orchestration layer, and the threat actor reads them there: she needs complete knowledge of the basis information, which in one protocol variant is enough for her to reconstruct the privacy-amplification function without it ever being transmitted separately [Pacher2016]. She does not manipulate that traffic, which is why the row carries classical_channel_tamper none; active control of the exchange is A-0152's behaviour. What the exchange leaks beyond the syndrome itself depends on the reconciliation class: in Lutkenhaus1999's protocol, where the parity bits are sent encrypted, a unidirectional error-correction protocol gives no extra hints, while "The interactive error correction protocol, however, leaks some information to Eve about the position of bits which were received incorrectly by Bob during the quantum protocol" [Lutkenhaus1999]. |
L4 — System Control & Orchestration | Secondary Effect |
| Relation type | Description | Technique |
|---|---|---|
| Preceded by |
The published framework's blinding chain follows this step with post-processing of the intercepted data, which turns the sifted key the threat actor holds into the final key [Krelina2025]. |
A-0070 Blinding Attack with Faked States |
| Preceded by |
The published framework's PNS chain follows this step with post-processing of the intercepted data, which turns what the threat actor learned into the final key [Krelina2025]. |
A-0044 Photon-Number-Splitting (PNS) Attack |
| Preceded by |
The second and more common thing this foothold is used for. Where the deployment's authentication is sound, the classical traffic is still readable by design, and recording the sifting and reconciliation messages is what lets a threat actor mirror the legitimate post-processing and reduce her own raw outcomes to the same distilled key. Both worked kill chains in Krelina2025 use this step this way and follow it with post-processing in the Exploiting phase, which SQOUT carries as A-0040 [Krelina2025]. |
A-0032 Compromise and Eavesdrop Classical Channel in Quantum Systems |
|