QNN-QRL: Quantum Neural Network Integrated with Quantum Reinforcement Learning for Quantum Key Distribution

Bikash K. Behera, Saif Al-Kuwari, Ahmed Farouk · 2025

Read QNN-QRL: Quantum Neural Network Integrated with Quantum Reinforcement Learning for Quantum Key Distribution by Bikash K. Behera, Saif Al-Kuwari, Ahmed Farouk (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-18188.txt.

Introduction

Bikash Behera, Saif Al-Kuwari, and Ahmed Farouk blend, in a 2025 paper, quantum learning methods into the protocols that share encryption keys with quantum security. Quantum key distribution promises secrecy guaranteed by physics. Its real-world rates and robustness still need tuning.

The authors add quantum neural networks to two classic protocols, and reinforcement learning agents that adapt the process. Combined versions merge both learners. The goal is faster key generation without weakening the security claims. Machine learning that lives on quantum hardware itself is young. Trials like this test whether it can manage the very links it may one day protect.

Catalog record

CREATORBikash K. Behera, Saif Al-Kuwari, Ahmed Farouk
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE50 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2501-18188

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

QNN-QRL: Quantum Neural Network Integrated with Quantum Reinforcement Learning for Quantum Key Distribution Bikash K. Behera, Saif Al-Kuwari, Ahmed Farouk Abstract Quantum key distribution (QKD) has emerged as a critical component of secure communication in the quantum era, ensuring information-theoretic security. Despite its potential, there are practical issues in optimizing key generation rates, enhancing security, and incorporating QKD into deployable implementations. In this paper, we introduce a unique framework for incorporating quantum machine learning (QML) algorithms, notably quantum reinforcement learning (QRL) and quantum neural networks (QNN), into QKD protocols to improve key generation performance. We evaluate our proposed algorithms using standard evaluation metrics, including accuracy, precision, recall, F1 score, confusion matrices, and ROC curves. Our results demonstrate considerable improvements in key generation quality. Furthermore, the existing and proposed models are investigated in the presence of different noisy channels to evaluate their robustness. The proposed integration of QML algorithms into QKD protocols and their noisy analysis creates a new paradigm for efficient key generation, which advances the practical implementation of QKD systems. I Introduction Quantum key distribution (QKD) is a promising technology to realize secure communication in future networks. Fig. 1 : Schematic diagram showing the workings of the proposed QNN-QRL algorithm. However, QKD faces numerous challenges, including limited key generation rates, restricted communication distances, practical implementation complexities, and security vulnerabilities arising from imperfect devices, all of which hinder its widespread adoption in quantum-safe communications [ 1 ] . Key QKD protocols, such as BB84, B92, SARG04, and E91, have been developed to address some of these concerns [ 2 ] , but the practical use of QKD remains largely speculative. Quantum machine learning (QML) approaches, such as quantum reinforcement learning (QRL) and quantum neural network (QNN) learning algorithms, offer promising solutions to the problem of QKD. QRL can improve resource allocation and routing in QKD networks, enabling efficient key generation and secure communication over long distances while decreasing vulnerabilities associated with trusted relay nodes [ 3 , 4 ] . QNNs can detect and prevent side-channel attacks and device faults by learning intricate patterns in noisy quantum signals and forecasting probable vulnerabilities [ 5 , 6 ] . QRL-based frameworks can dynamically react to ambient conditions and device failures, improving scalability and robustness [ 7 ] . Integrating QNNs into authentication systems for conventional post-processing can improve resilience against denial-of-service attacks and reduce key consumption [ 8 ] . QML methods use quantum computing for real-time optimization and prediction, overcoming QKD’s limitations in speed, scalability, and integration with existing networks [ 9 ] . These enhancements demonstrate QML’s transformative potential in overcoming QKD restrictions and paving the way for quantum-safe communication systems. I-A Related Work This work integrates machine learning (ML) with a cascade error correction protocol that efficiently predicts quantum bit error correction and final key length [ 10 ] . This uses deep learning models to detect and mitigate eavesdropping attacks [ 11 ] , and reinforcement learning (RL) to adaptively alter system settings in real-time to account for the noise profiles of different quantum channels. This introduces QKD networks enabled by software-defined networking (SDN) as a promising technology [ 12 ] . This explores the integration of IoT-enabled medical devices with QKD to enhance data security and energy efficiency in healthcare systems, particularly during pandemics like COVID-19 [ 13 ] . Three QKD protocols are demonstrated using IBM Q platforms [ 14 ] . This reviews recent developments, in-field deployments, and standardization advancements of QKD networks, highlighting their potential and challenges for large-scale commercialization [ 15 ] . This introduces the principles and requirements of QKD in practical scenarios and applications while highlighting advancements in its technical and architectural domains [ 16 ] . This shows that RL can efficiently allocate resources in QKD networks [ 17 ] . This study investigates QKD as a use case for QML, enhancing eavesdropping attacks on the BB84 protocol, developing a unique cloning mechanism for noisy environments, and building a collective attack utilizing QKD post-processing [ 18 ] . This study uses a Markov decision process and deep reinforcement learning (DRL) to optimize memory cutoff strategies for entanglement distribution over multi-segment repeater chains [ 19 ] . This proposes a DRL-based routing and resource assignment (RRA) scheme for QKD-secured optical networks (QKD-ONs), which outperforms traditional methods like deep Q-network (DQN), first-fit (FF), and random-fit (RF) across national science foundation network (NSFNET) and US backbone network (UBN24) [ 20 ] . This framework combines QKD, Multi-Objective Optimization (MOO), and RL to ensure safe and efficient data transmission in Smart Grid Advanced Metering Infrastructure (SGAMI) [ 21 ] . This demonstrates how RL optimizes resource allocation for efficient key distribution in QKD networks [ 22 ] . I-B Contribution Fig. 1 represents the schematic of the overall proposed quantum algorithms for the quantum key generation process. Although there has been considerable previous work on improving the key generation process, QML algorithms such as QRL and QNN have not been integrated into them to investigate the quality of the key generation process. In this paper, we proposed two QRL-based QKD protocols and two QNN-based QKD protocols. The existing and proposed algorithms are run against six different noise channels to evaluate their performance. The contributions of the paper are summarized as follows. 1. Two novel QRL-based algorithms are proposed for the quantum key generation process, namely QRL-V1 and QRL-V2. 2. QNN-based enthronements are added to the popular BB84 and B92 protoocls are introuced to produce two novel algorithms , QNN-BB84, QNN-B92, 3. Two integrated QNN-QRL algorithms are proposed, namely QNN-QRL-V.1, and QNN-QRL-V.2 4. A comparative analysis between the existing BB84, B92, and the proposed QRL-based and QNN-based algorithms is performed 5. An extensive noise model analysis for the existing and proposed algorithms is made. I-C Organization The rest of the paper is organized as follows: Section II briefly discusses the BB84, B92, QRL, and QNN algorithms. Section III presents the general formulation of the problem and the proposed algorithms in detail. In Section IV , the settings, datasets, preprocessing, hyperparameters, and overall results are presented. Finally, the paper is concluded in Section V . II Preliminaries II-A BB84 The BB84 protocol, a cornerstone of QKD leverages the principles of quantum mechanics to securely share a cryptographic key between two parties, traditionally called Alice (sender) and Bob (receiver), even in the presence of an eavesdropper, Eve [ 23 ] . The BB84 protocol exploits two fundamental properties of quantum mechanics, such as superposition and measurement, and the no-cloning theorem: Alice selects a random sequence of bits, b=\{b_{1},b_{2},\dots,b_{n}\} , and a random sequence of bases, r=\{r_{1},r_{2},\dots,r_{n}\} , for encoding: • Rectilinear basis ( + ): \ket{0},\ket{1} • Diagonal basis ( \times ): \ket{+}=\frac{1}{\sqrt{2}}(\ket{0}+\ket{1}),\;\ket{-}=\frac{1}{\sqrt{2}}(\ket{0}-\ket{1}) For each bit b_{i} and basis r_{i} : • If r_{i}=+ : encode b_{i}=0 as \ket{0} , b_{i}=1 as \ket{1} . • If r_{i}=\times : encode b_{i}=0 as \ket{+} , b_{i}=1 as \ket{-} . Alice transmits the encoded qubits through a quantum channel to Bob. Bob randomly selects bases r^{\prime}=\{r^{\prime}_{1},r^{\prime}_{2},\dots,r^{\prime}_{n}\} to measure the incoming qubits: • If r^{\prime}_{i}=r_{i} : Bob obtains the correct bit value b_{i} . • If r^{\prime}_{i}\neq r_{i} : Bob’s measurement results are random. Alice and Bob publicly announce their

Read the complete text