Table of Contents
Fetching ...

Optical-pumping attack on a quantum key distribution laser source

Maxim Fadeev, Anastasiya Ponosova, Qingquan Peng, Anqi Huang, Roman Shakhovoy, Vadim Makarov

TL;DR

The paper identifies a new attack on practical QKD transmitters where sub-band optical pumping at 1310 nm increases the pulse energy of a 1550 nm gain-switched laser, potentially biasing security. Using a controlled experimental setup that injects 1310 nm light into a surrogate Alice laser, the authors measure increases in pulse energy (up to ~10%) and average power (up to ~21.7%), with a pumping threshold around 140 µW and modest, transient changes in pulse shape. They discuss the security implications, showing that sufficient optical isolation can mitigate the attack, but note that spectral dependencies of passive countermeasures may leave some systems vulnerable. A risk evaluation on an industrial QKD transmitter suggests resilience if isolation meets a threshold, but broader spectral testing and certification are necessary to ensure protection across architectures and absorption bands.

Abstract

We report a new type of vulnerability in practical implementations of quantum key distribution systems. We show that it is possible to increase the pulse energy of a source laser diode not only by injection-locking it by external light near its emission wavelength of 1550 nm, but also by optically pumping it at a much shorter wavelength. We demonstrate 10% increase in pulse energy when exposing the laser diode to 1310-nm, 1.6-mW cw laser light via its fiber pigtail. This may allow an eavesdropper to steal the secret key. A possible countermeasure is to install broadband optical filters and isolators at the source's output and characterise them during the security certification.

Optical-pumping attack on a quantum key distribution laser source

TL;DR

The paper identifies a new attack on practical QKD transmitters where sub-band optical pumping at 1310 nm increases the pulse energy of a 1550 nm gain-switched laser, potentially biasing security. Using a controlled experimental setup that injects 1310 nm light into a surrogate Alice laser, the authors measure increases in pulse energy (up to ~10%) and average power (up to ~21.7%), with a pumping threshold around 140 µW and modest, transient changes in pulse shape. They discuss the security implications, showing that sufficient optical isolation can mitigate the attack, but note that spectral dependencies of passive countermeasures may leave some systems vulnerable. A risk evaluation on an industrial QKD transmitter suggests resilience if isolation meets a threshold, but broader spectral testing and certification are necessary to ensure protection across architectures and absorption bands.

Abstract

We report a new type of vulnerability in practical implementations of quantum key distribution systems. We show that it is possible to increase the pulse energy of a source laser diode not only by injection-locking it by external light near its emission wavelength of 1550 nm, but also by optically pumping it at a much shorter wavelength. We demonstrate 10% increase in pulse energy when exposing the laser diode to 1310-nm, 1.6-mW cw laser light via its fiber pigtail. This may allow an eavesdropper to steal the secret key. A possible countermeasure is to install broadband optical filters and isolators at the source's output and characterise them during the security certification.

Paper Structure

This paper contains 7 sections, 6 equations, 8 figures, 2 tables.

Figures (8)

  • Figure 1: Scheme of experiment. LD, laser diode; PG, pulse generator; OC, optical circulator; PD, photodiode; Osc, oscilloscope; PwM, power meter; PC, polarisation controller; VOA, variable optical attenuator. External electrical connections of LD1550 for pulsed operation are shown. Port 3 of OC is interchangeably connected to either PD or PwM.
  • Figure 2: Dependence of differential quantum efficiency on the injected cw power of Eve. "No pumping after attack" shows the level of differential quantum efficiency immediately after exposure. The inset shows measured light--current characteristics of LD1550 in cw regime.
  • Figure 3: Average output power of pulsed LD1550 under exposure to $1310$-$\nano\meter$ light.
  • Figure 4: Pulse energy of LD1550's pulses under Eve's illumination, normalised to their energy without pumping. Error bars present the standard deviation. Inset shows typical single-shot pulse shapes before exposure and under exposure to the minimum and maximum 1310-nm laser powers. The pulses change shape. Their mean timing, however, does not change; the time shift visible in the plot is the result of a random jitter of individual pulses.
  • Figure 5: Spectral characteristics near $1310~\nano\meter$ of typical $1550$-$\nano\meter$ fiber-optic isolators in backward direction and a dense-wavelength-division multiplexer (DWDM; Prointech DWDM-1T-MOD777-34) measured from its common port to the port of channel 34.
  • ...and 3 more figures