Timing Attack Exploiting Detector Response Times
A quantum key distribution receiver uses several single-photon detectors, and which one of them fires is what carries the secret bit. The detectors do not all answer at the same instant: each has its own small, fixed delay between the photon arriving and the detector reporting a click. Because the two parties routinely publish their detection times over the ordinary public channel so they can match their events, the published time quietly carries the identity of the detector, and with it part of the key.
The threat actor (Eve) therefore needs no access to the quantum channel at all. She listens to traffic the protocol publishes on purpose, separates the announced times into the groups belonging to each detector, and reads off part of the secret key. She sends nothing, measures no photons and adds no errors, so the transmitter (Alice) and the receiver (Bob) see a link that behaves completely normally and have no way to notice [LamasLinares2007].
The measuring paper attributes the delay differences mainly to the receiver's readout electronics, from the length of each detector's own circuit-board track and from small differences between the diodes, rather than from the optics: in the device that was measured, the optical paths to the four detectors were matched far too closely to account for the effect. The repair is to equalise the delays and to control what the system publishes [LamasLinares2007,Pahali2022].
| [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. |
| [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) |
| [LamasLinares2007] | A. Lamas-Linares and C. Kurtsiefer. "Breaking a quantum key distribution system through a timing side channel" In: Opt. Express 15, 9388-9393. (2007) 10.1364/OE.15.009388. |
| [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. |
| [Pahali2022] | M. Pahali, U. Tefek, K. Durak. "Cryptographic Security Concerns on Timestamp Sharing via Public Channel in Quantum Key Distribution Systems" (2022) 10.1103/PhysRevA.106.012611. arXiv:2203.03250. |
2857f98a-8dee-4bb8-ab3e-d6b5fbd1e2c9
- Quantum Communication
Technique properties
Operational properties
| Description | Countermeasure |
|---|---|
|
Privacy amplification is the control the primary names as the final step of its prescribed fix: the timing information should be characterised and the delays equalized, randomized or the precision truncated so that the leakage falls below a chosen threshold, after which the protocol deals with the residue in the same way it deals with errors, by applying an appropriate amount of privacy amplification [LamasLinares2007]. It neutralises rather than prevents, because it does not stop the threat actor from learning the bits; it compresses the final key so that what she learned is no longer about it. |
CM-0001 Privacy amplification |
|
This is the control written against this attack: CM-0062's own first measure is characterisation and equalisation of the timing delays associated with each detector, and the row already cites this primary. Its three measures are not equivalent and the difference decides whether an operator is safe. Equalise the offsets. The measure with a measured result. For the characterised device the dominant distinguishing feature is the time offset, and compensating it so that it is identical for all detectors reduces the leakage from per cent to around 0.3 per cent [LamasLinares2007]. Restrict who publishes. [Pahali2022] gives a concrete procedure: only one party publishes its timestamps and basis information, the other measures the distinguishability of the coincidence peaks in the cross-correlogram and eliminates the offset delay in one detector so that the peaks overlap completely, and only then shares its own timestamps, so the threat actor never learns the relative delay. Truncate the published precision. The weakest of the three and the one most likely to be mistaken for sufficient. The primary already warned that discretising into 0.5 ns and 1 ns bins reduces the eavesdropper's information but that somewhat counterintuitively there is still a strong leakage even at bin sizes comparable to the width of the distribution, with a penalty in increased background, and that a typical experimental precision of about 150 ps differs negligibly from the continuous case [LamasLinares2007]. [Pahali2022] computed the failure: increasing the bin width does not monotonically reduce the mutual information between the key bits and the eavesdropper's observation, with the values at 1.0, 1.6 and 3.2 ns almost the same, and the leak is periodic in where the bins start with period equal to the bin width, so a defensible bin width can still sit at a maximum. |
CM-0062 Management of Published Information In Classical Channel Communication |
| Description | Category | Impact Type |
|---|---|---|
|
Electronics and Analog Components is the primary layer, because the property the attack reads is created there and repaired there. The 240 ps spread comes from a few centimetres of difference in per-detector circuit-board trace together with different parasitic capacitances acting through a fixed comparator threshold, and equalising it is a board-layout and delay-calibration task [LamasLinares2007]. The catalogue's nearest precedent agrees: CM-0139, whose measures are the same layout decisions, maps to L2 as its primary layer. |
L2 — Electronics & Analog Components | Primary Target |
|
Post-Processing and Cryptographic is a secondary layer and the one the threat actor actually touches: she reads the public exchange in which detection times are announced so that coincident events can be matched [LamasLinares2007], and two of the three available repairs act here rather than at the detector, namely limiting the published precision and absorbing the residue with additional privacy amplification. It is secondary rather than primary because a post-processing change can conceal the defect but cannot remove it, and because the successor paper shows the concealment is unreliable [Pahali2022]. |
L5 — Post-Processing & Cryptographic | Secondary Effect |
|
Physical Layer, Quantum and Optical, is affected but only in a bounded way, and recording it with that bound is the point. The optical distance from the entrance of the module to the four detectors differs by less than 1 mm [LamasLinares2007], which the paper does not claim to be zero, so the optics contribute a real but small share of a 240 ps spread. The paper bounds this contribution rather than quantifying it, and the layer is secondary because an optical-bench correction alone would not close the channel. |
L1 — Physical Layer (Quantum & Optical) | Secondary Effect |
| Relation type | Description | Technique |
|---|---|---|
| Alternative to |
The same distinguishability analysis applied at the other end of the link. A-0086 exploits imperfections that make a receiver's detectors distinguishable from one another; A-0083 exploits imperfections that make a transmitter's emitted states distinguishable. The primary itself draws the connection in its closing paragraph, noting that if the states to be sent are prepared by different physical devices their temporal response needs to be characterised and the possible information leakage evaluated with a similar analysis [LamasLinares2007]. That analysis was later carried out on an eight-diode free-space transmitter, which measured bits per pulse of temporal leakage after the emitters had been overlapped with programmable delay lines [Nauerth2009]. Both are passive leaks of partial key that add no errors. |
A-0083 Side Channel Exploitation in Transmitter Imperfections |
| Relation type | Description | Technique |
|---|---|---|
| Alternative to |
The closest structural relative in the catalogue, and the pairing a defender needs. Both techniques recover detector identity from publicly announced detection timing with no action on the quantum channel, no equipment beyond a classical-channel tap and no error signature, since in both cases "she knows when bits are detected and in which basis they are sifted" is the whole mechanism [Rogers2007]. They differ in which timing property carries the identity: A-0086 uses the fixed per-detector click-latency difference, so it works at any repetition rate and leaks on every detection, while this one uses the dead-time-forced alternation, so it needs the link clocked above the inverse dead time but then leaks a whole run from one bit. |
A-0085 Dead Time Correlation Attack |
|