Exploiting Hardware Imperfections in Practical Quantum Key Distribution

27 concepts11 papers1 proposal

Theoretical security proofs for Quantum Key Distribution (QKD), such as the foundational work by Biham et al. (2006), provide mathematical guarantees of absolute security against arbitrary eavesdropping attacks. However, these proofs rely on idealized assumptions: the sender (Alice) transmits perfect single photons, and the receiver (Bob) uses flawless, wavelength-independent optical components.

In practice, building a true single-photon source remains a significant engineering challenge. Instead, real-world implementations often substitute single photons with attenuated laser pulses, known as weak coherent states, or rely on commercial optical components that exhibit subtle physical imperfections. This gap between theoretical ideals and physical hardware introduces critical vulnerabilities that an eavesdropper (Eve) can exploit to steal secret keys without detection.

The Intuition Behind Photon Number Splitting

When Alice uses an attenuated laser pulse to simulate a single photon, the number of photons in each pulse follows a Poisson distribution. While most pulses are empty or contain exactly one photon, a small fraction inevitably contain two or more identical photons. As noted by Waks and Yamamoto (2002) and Curty et al. (2007), this multi-photon emission opens the door to the Photon Number Splitting (PNS) attack.

In a standard intercept-and-resend attack on a true single photon, Eve is forced to measure the photon in a guessed basis. If her guess is wrong, she alters the quantum state, introducing a measurable error rate (such as the 25% Quantum Bit Error Rate in the BB84 protocol) that alerts Alice and Bob to her presence.

The PNS attack circumvents this fundamental quantum defense by exploiting multi-photon pulses:

  1. Non-Demolition Measurement: When Alice sends a pulse, Eve uses a quantum non-demolition measurement to count the number of photons in the pulse without disturbing their polarization or phase.
  2. Splitting the Pulse: If Eve finds only a single photon, she blocks it to avoid risking detection. If she detects a multi-photon pulse (e.g., containing two photons), she splits it. She diverts one photon into a quantum memory buffer and allows the remaining photon to travel to Bob.
  3. Zero-Disturbance Eavesdropping: Because the photon Bob receives is completely unaltered, he detects no anomaly or increased error rate.
  4. Delayed Measurement: Later, during the public basis reconciliation phase, Alice and Bob openly announce which measurement bases they used. Eve simply retrieves her stored photon from her quantum memory and measures it in the correct basis.

By utilizing this strategy, Eve gains full information about that key bit with zero risk of altering the signal or revealing her presence. This vulnerability drastically limits the secure transmission distance and key generation rates of practical systems (Curty et al., 2007; Duo et al., 2021; Lu et al., 2019).

Defending Against PNS Attacks

To counteract the PNS vulnerability, researchers have developed two primary strategies: engineering alternative quantum light sources or implementing decoy states.

One approach is to replace weak coherent pulses with quantum states that suppress multi-photon emissions. For instance, Wang et al. (2015) and Zhu et al. (2018) explore the use of Single-Photon-Added Coherent Sources (SPACS), which eliminate the vacuum component and lower multi-photon probabilities. Alternatively, Waks and Yamamoto (2002) demonstrate that using entangled-photon pairs inherently protects against individual PNS attacks because the source physics prevents Eve from cleanly splitting off a redundant photon without disrupting the shared entanglement.

When weak coherent sources must be used, the standard defense is the decoy-state method (Zhu et al., 2018; Cao et al., 2020). By randomly varying the intensity of her laser pulses, Alice creates "signal" pulses and "decoy" pulses. Because a PNS attack selectively targets multi-photon pulses, Eve's tampering alters the transmission statistics of the different intensity classes in a way that Alice and Bob can mathematically detect, exposing the eavesdropper.

Exploiting Component Side-Channels

Eavesdropping strategies are not limited to manipulating photon numbers; they can also exploit the physical properties of optical components inside the receiver's terminal.

Remote Basis Control via Wavelength Manipulation

In many QKD setups, Bob selects his measurement basis passively using a 50:50 beam splitter that randomly routes incoming photons to different measurement arms. However, commercial beam splitters are only balanced for a specific design wavelength.

As demonstrated by Li et al. (2011), Eve can exploit this hardware imperfection by intercepting Alice's photons and resending them at completely different, carefully chosen "trigger" wavelengths. At these alternative wavelengths, the beam splitter's routing ratio shifts from 50:50 to nearly 100:0 or 0:100.

[Eve Intercepts] ---> [Resends at Trigger Wavelength] ---> [Bob's Beam Splitter]
                                                                  |
                                      (Splitting ratio shifts from 50:50 to 100:0)
                                                                  |
                                                                  v
                                                   [Forced into Eve's Chosen Basis]

By shifting the wavelength, Eve can remotely dictate which path the photon takes, forcing Bob to measure in the exact basis Eve used during her interception. This eliminates the basis mismatch errors that normally expose an intercept-and-resend attack, keeping the error rate below the detection threshold. Feng et al. (2024) extended this concept to underwater QKD systems, showing that wavelength manipulation can trick legitimate users into underestimating channel noise, masking the presence of an intruder.

Hiding Noise via Polarization Manipulation

Hardware vulnerabilities also manifest in Continuous-Variable QKD (CV-QKD) systems that utilize a Local Local Oscillator (LLO) for phase reference. To compensate for physical polarization drift in the fiber channel, the system must periodically measure the polarization of reference pulses.

However, to maintain high data rates, practical systems often measure only a subset of these pulses. Shao et al. (2022) showed that this incomplete monitoring creates a loophole. Eve can deliberately manipulate the polarization of the unmeasured phase reference pulses. This targeted manipulation distorts the receiver's phase noise estimation, causing Alice and Bob to underestimate the actual system noise and mistakenly overestimate their secure key rate, effectively blinding them to the ongoing eavesdropping.

Key Foundational Work

Security of quantum key distribution with entangled photons against individual attacks (2002) — contributed 1 concept elements to this direction.
A Proof of the Security of Quantum Key Distribution (2006) — contributed 1 concept elements to this direction.
History & citationVersion 1 · 2026-10-04

Current · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_photon_number_splitting_vulnerability_bffc1d3a,
  title = {photon-number-splitting vulnerability},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/photon-number-splitting-vulnerability/rdr_bffc1d3a63854b139af39fc2b8c150e1},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-20},
}

This exact version (version 1, frozen text):

@misc{lacuna_photon_number_splitting_vulnerability_bffc1d3a_v1,
  title = {photon-number-splitting vulnerability},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/photon-number-splitting-vulnerability/rdr_bffc1d3a63854b139af39fc2b8c150e1/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-20},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake