Entanglement-assisted authenticated BB84 protocol

Installation of required libraries

install -r requirements.txt

Necessary library modifications!

In the folder "Modified_libraries" are the updated files for perceval and qutip (changes explained in Modified_libraries/comment.txt). In source code you can find the original files at:

• perceval/components/unitary_components.py (adds the loss of beam splitters and phase shifters according to: arXiv:2311.10613v3)
• qutip/measurements.py (creates extra protection with the choice of probabilities during Z-basis measurement)

Update them by creating a development directory, see more at: https://stackoverflow.com/questions/23075397/python-how-to-edit-an-installed-package

Architecture

One directional CX/CZX BB84 embedded authentication, QM defines the Atomic frequency comb cavity enhanced memory which stores the states at the server-user end. In the output measurement (Z-basis) Alice verifies Bob by measuring [0,1,1,0..]. For noiseless authentication 0s would correspond to CX and 1s to CZX.

Usage

IBM Token:

• You should in main.py update token=' ' with your IBM token API to use the IBM backend for the decoherence and dephasing parameters.

You either run the code independently (main.py and DNN_binary_class.py) or sequentially using snakemake by running ( "--cores x" means the amount of cores for multiple file processing to create the attacker and user dataset from main.py):

Command example:

snakemake --cores 1

The authentication protocol code is divided into two parts:

main.py

Generates the simulation with CPU multiprocessing for user (attack0) or attacker (attack1);

Command example: (10 shots, distance 1km, wait 1000ns, amount of authentications 1, cores in multiprocessing of shots 1, user is using the protocol)

python main.py shots10 dist1000 wait1000 users1 cores1 attack0

Input:

• dist: Between Alice and Bob [m].
• wait: Time to retain qubits in a quantum memory [ns].
• cores: Amount of cores used in multiprocessing for the simulation (its different from the file multiprocessing present in snakemake --cores x).
• users: Amount of users - repeats the trials for each user, the simulator iterates from (1,nusers,step).
• shots: λ, lenght of the authentication string.
• attack: 0 (user authenticating) or 1 (attacker authenticating).

Output:

• Vec: probability vector of correct authentication.
• Store_V: Output vector of measurements for authentication.
• Authen_x.txt: Output file with the Store_V in it

DNN_binary_class.py

Generates the DNN binary classification for the input data;

Input:

• At: Output vector of measurements for authentication for Attacker.

• E: Output vector of measurements for authentication for User.

  or

• Authen_1.txt: Output file from main.py with vector of measurements for authentication for Attacker.

• Authen_0.txt: Output file from main.py with vector of measurements for authentication for User.

Output:

• Roc curve plot.
• Accuracy, cross entropy plot.
• Att: Probability of correctly predicted authentication.
• DataDNN.txt: Att data in a file.

Contents of requirements.txt

keras==2.9.0
matplotlib==3.5.2
numpy==2.1.2
pandas==1.4.3
perceval_quandela==0.11.1
qiskit==1.2.4
qiskit_aer==0.15.1
qiskit_experiments==0.7.0
qiskit_ibm_runtime==0.30.0
qiskit_ibmq_provider==0.19.2
qiskit_ignis==0.7.1
qiskit_terra==0.25.2.1
qutip==5.0.4
scikit_learn==1.1.1
scipy==1.14.1
sdeint==0.3.0
tensorflow==2.9.0
tensorflow_intel==2.16.1
torch==1.12.0+cu116
qiskit_terra==0.22.3
torch==1.12.0
snakemake==7.32.4

Authentication of QKD background:

An assumption is made during QKD protocols that both parties are to be trusted, what if that's not the case? A realistic AFC memory + noise simulation for one-way authentication of QKD is proposed in this work. This repository combines all the codes to produce the plots and results from the following article: arXiv:2407.03119.

Please cite it if the code is used.