Double-Click Attack
The Double-click attack in quantum key distribution systems exploits simultaneous detection events in both detectors of a chosen basis at the receiver's (Bob's) end [Lutkenhaus1999,Lutkenhaus2000]. By employing an intercept-and-resend strategy with re-sent states as intense multi-photon pulses, the threat actor (Eve) can induce double-click events when Bob's basis selection is incompatible with the re-sent state. Properly leveraging these events allows Eve to gain knowledge of the key, as the double-clicks, often discarded by QKD protocols, filter out non-manipulated detection events, leaving only those controlled by Eve.
Because the events that would have exposed Eve are exactly the ones thrown away, the key material that survives carries almost no error, so she can learn it without raising the alarm that normally reveals an eavesdropper. The cited sources state the loophole [Lutkenhaus2000]; the size of the yield is inferred from the mechanism. What keeps a receiver safe is that the ambiguous rounds stay inside the security accounting, and the usual fix is a post-processing rule rather than new hardware. The receiver can assign each double click a random bit value, which is the rule the primary sources name [Lutkenhaus2000,Beaudry2008]. Since 2011 a universal squash covers essentially every protocol in scope; it discards double-click key bits and counts their rate as possible errors [Fung2011]. Or it can let the ambiguous rounds enter the disturbance calculation instead, either by weighting them [Lutkenhaus1999] or by discarding them while charging their rate against the key, as proved for prepare-and-measure protocols [Fung2011] and for an entanglement-based link [Koashi2008]. A system exposed to this attack is one that drops the ambiguous rounds without letting them affect either the key bits or the key rate.
| [Beaudry2008] | N. J. Beaudry, T. Moroder, and N. Lütkehaus. "Squashing Models for Optical Measurements in Quantum Communication" In: Phys. Rev. Lett. 101, 093601. (2008) 10.1103/PhysRevLett.101.093601. |
| [Fung2011] | C.-H. F. Fung, H. F. Chau, and H.-K. Lo. "Universal squash model for optical communications using linear optics and threshold detectors" In: Phys. Rev. A 84, 020303(R). (2011) 10.1103/PhysRevA.84.020303. |
| [Gittsovich2014] | O. Gittsovich, N. J. Beaudry, V. Narasimhachar, R. Romero Alvarez, T. Moroder, N. Lutkenhaus. "Squashing model for detectors and applications to quantum key distribution protocols" (2014) 10.1103/PhysRevA.89.012325. arXiv:1310.5059. |
| [Kamin2024] | Lars Kamin, Norbert Lutkenhaus. "Improved decoy-state and flag-state squashing methods" (2024) 10.1103/PhysRevResearch.6.043223. arXiv:2405.05069. |
| [Kato2016] | Go Kato, Kiyoshi Tamaki. "Security of six-state quantum key distribution protocol with threshold detectors" (2016) 10.1038/srep30044. arXiv:1008.4663. |
| [Koashi2008] | M. Koashi, Y. Adachi, T. Yamamoto, and N. Imoto. "Security of entanglement-based quantum key distribution with practical detectors" In: Phys. Rev. A 90, 032320. (2008) 10.48550/arXiv.0804.0891. arXiv:0804.0891. |
| [Lutkenhaus1999] | Norbert Lütkenhaus. "Estimates for practical quantum cryptography" In: Phys. Rev. A 59, 3301. (1999) 10.1103/PhysRevA.59.3301. |
| [Lutkenhaus2000] | N. Lutkenhaus. "Security against individual attacks for realistic quantum key distribution" In: Phys. Rev. A 61 (2000), p. 052304. (2000) 10.1103/PhysRevA.61.052304. |
| [Tsurumaru2008] | Tsurumaru, Toyohiro and Tamaki, Kiyoshi. "Security proof for quantum-key-distribution systems with threshold detectors" In: Phys. Rev. A 78, 032302. (2008) 10.1103/PhysRevA.78.032302. |
249d8fe1-41a0-49a2-876e-a8b35d836cfb
- Quantum Communication
Technique properties
Operational properties
| Description | Countermeasure |
|---|---|
|
A threshold detector that cannot resolve photon number is exactly the device imperfection a security proof has to incorporate rather than assume away. Lutkenhaus1999 does it concretely: the ambiguous double-click outcome is admitted into the generalized sifted key with an adjustable weight , fixed at for a passive detection unit, so that any multi-photon eavesdropping strategy is charged more disturbance than a single-photon one and stops being profitable [Lutkenhaus1999]. A proof that instead models Bob as an ideal qubit measurement has no double-click outcome at all, which is the gap this attack lives in [Beaudry2008]. Fung2011 is the modern instance of doing exactly this: it takes the real weak-coherent-source and threshold-detector implementation into the proof rather than idealising it away, and shows that once the requirement is relaxed from exact statistics to bounds a universal squash exists [Fung2011]. |
CM-0014 Full Device Imperfections Incorporation in Scientific Security Proof |
| Description | Category | Impact Type |
|---|---|---|
|
Post-Processing & Cryptographic: the exploit depends on how the sifting rule treats double-clicks, and the fix lives at this layer and nowhere else. This is the layer that carries the defect: changing that one rule, and nothing else, ends the attack [Lutkenhaus2000]. More than one rule works. The receiver can apply a squashing model, assigning each double click at random to one of the two single-click outcomes [Beaudry2008]. It can keep the ambiguous rounds in the disturbance parameter at weight instead [Lutkenhaus1999]. Or it can discard the double clicks and carry their rate into the key-rate calculation, which Koashi2008 proves sufficient for unconditional security on BBM92, concluding that for almost all practical purposes the random-bit assignment is unnecessary [Koashi2008]. All three are the same move at bottom, keeping the ambiguous rounds inside the security accounting. What fails is letting them affect neither the key bits nor the key rate. The fix is also no longer protocol-limited, which is the single largest thing this entry was missing. Beaudry2008's 2008 result left six-state without a squashing map, but Fung2011 showed a universal squash exists once the requirement is relaxed from exact statistics to bounds [Fung2011], and Kato2016 proved six-state secure with threshold detectors at a bit-error threshold of 12.611 percent against 12.619 percent for the idealised case [Kato2016]. A receiver whose double clicks reach neither the key bits nor the key rate is in that state by implementation choice, not because it lacks a valid model. |
L5 — Post-Processing & Cryptographic | Primary Target |
|
Physical Layer (Quantum & Optical): the attack drives the receiver's threshold single-photon detectors into double-click events with bright multi-photon pulses [Beaudry2008]. This is the layer the attack passes through rather than the layer whose defect makes it pay. The detectors behave exactly as designed: a threshold detector cannot resolve photon number, so firing both arms on a multi-photon input is correct behaviour, not a fault [Fung2011]. Nothing at L1 has to be changed to close the attack. |
L1 — Physical Layer (Quantum & Optical) | Secondary Effect |
| Relation type | Description | Technique |
|---|---|---|
| Refines |
A-0045 Intercept-and-Resend is the delivery vehicle; this technique refines it by choosing the resend intensity so that basis-mismatched events become ambiguous double clicks the receiver discards, rather than errors it counts [Lutkenhaus2000]. |
A-0045 Intercept-and-Resend Attack |
|