Entanglement-assisted authenticated BB84 protocol

TL;DR

The paper tackles authenticated QKD by embedding BB84 within an entanglement-assisted framework that achieves two-factor information-theoretic security under specific adversary models. It presents a noiseless protocol and two noise-adaptation strategies (a rigorous QBER-based criterion and a DNN-based classifier) to maintain security in realistic conditions, including a key-recycling feature for pre-shared secrets. The authors provide a comprehensive simulation framework using photonic channels and cavity-enhanced AFC memories to assess performance, demonstrating secure operation for short to moderate distances (1–10 km) and storage times up to 150 μs with accuracy >0.80 in forgery detection. This work advances practical, provably secure QKD authentication by leveraging entanglement, pre-shared randomness, and modern ML techniques, potentially improving resilience against impersonation and MITM attacks in quantum networks.

Abstract

In this work, we present a novel authenticated Quantum Key Distribution (QKD) protocol employing maximally entangled qubit pairs. In the absence of noise, we securely authenticate the well-known BB84 QKD scheme under two assumptions: first, adversaries cannot simultaneously access preshared and non-pre-shared secret classical information, and second, adversaries cannot simultaneously access pre-shared secret classical information and quantum memories held by legitimate parties. The main strength of this noiseless result is that access to all secretly pre-shared classical information is insufficient for breaching our scheme. Additionally, our protocol desirably allows for pre-shared secrecy reusage, leading to secret key growing. In order to address noise, we simulate a photonic implementation of our scheme, together with a storage model that aims to replicate the performance of cavity-enhanced Atomic- Frequency Comb (AFC) memories. Two methods are used to distinguish authentic entities from forgery attempts: on the one hand, a statistical approach is used after calibration of its defining parameter $μ$. Alternatively, a Deep Neural Network (DNN) is designed and trained to learn the underlying different structure of that input data coming from adversaries in comparison to that one coming from legitimate parties. Both methods achieve a correct classification rate larger than 0.80 for memory storage time of 150 $μ$s and a 1 km distance between parties.

Figures (5)

• Figure 3: A spatially encoded optical circuit for generating the entangled pairs required in our protocol. $X$ or the identity operator $\mathbf{I}$ are applied according to the corresponding bit in $F_{\mathrm{Choices}}$. The phase-shifter and beam-splitter parameters, $\phi$ and $\theta$, respectively, control the optical transformations within the circuit.
• Figure 4: BB84 authentication performance, with the simulation parameters presented in Table. \ref{['parameter:table']} and $\lambda = 500$: Left)$1-\text{QBER}$ against transmission distance for multiple storage times. Right) Ratio $\lambda_{\mathrm{post}}/\lambda$ against distance for multiple storage times. Five samples, each generated with a different seed, were used to calculate the standard deviation.
• Figure 5: Comparison, between the static and the DNN methods, of the relative frequency obtained for correctly distinguishing between valid entities and attackers, for $\lambda = 100$. We set a distance between parties of $d= 1$ km, a memory storage time of $T = 150$$\mu\mathrm{s}$ and we calculate the standard deviation from 10 different simulations, each with distinct training and test set shuffling.
• Figure 6: ROC curve for $\lambda=100$.
• Figure 7: Accuracy and cross entropy for the training and test set of $\lambda=100$ for the DNN, for distance of 1 $\mathrm{km}$ and storage time of $150$$\mu \mathrm{s}$.

Theorems & Definitions (21)

• Definition 1
• Definition 2
• Remark 1
• Definition 3
• Definition 4
• Definition 5
• Remark 2
• Remark 3
• Remark 4
• Proposition 1