Multiple-Photon Absorption Attack on Entanglement-Based Quantum Key Distribution Protocols

TL;DR

This paper demonstrates a new attack method on entanglement-based quantum key distribution protocols, exploiting multiple-photon absorption to mimic secure statistics, challenging the assumption of security in such systems.

Contribution

It extends the multiple-photon absorption attack to various protocols and analyzes its effectiveness, revealing vulnerabilities in entanglement-based quantum key distribution.

Findings

The attack can reproduce secure-looking statistics with separable states.

Bell inequality violations increase with the order of photon absorption.

Quantum bit error rate decreases as the absorption order increases.

Abstract

In elaborating on the multiple-photon absorption attack on Ekert protocol proposed in arXiv:1011.4740, we show that it can be used in other entanglement-based protocols, in particular the BBM92 protocol. In this attack, the eavesdropper (Eve) is assumed to be in control of the source, and she sends pulses correlated in polarization (but not entangled) containing several photons at frequencies for which only multiple-photon absorptions are possible in Alice's and Bob's detectors. Whenever the photons stemming from one pulse are dispatched in such a way that the number of photons is insufficient to trigger a multiple-photon absorption in either channel, the pulse remains undetected. We show that this simple feature is enough to reproduce the type of statistics on the detected pulses that are considered as indicating a secure quantum key distribution, even though the source is actually a mixture of separable states. The violation of Bell inequalities measured by Alice and Bob increases with the order of the multiple-photon absorption that Eve can drive into their detectors, while the measured quantum bit error rate decreases as a function of the same variable. We show that the attack can be successful even in the simplest case of a two-photon absorption or three-photon absorption attack, and we discuss possible countermeasures, in particular the use of a fair sampling test.