Qiskit/QuNetSim Quantum Communication (QKD) Attack Simulation + GNN Detection & Secure Routing

Looking for Experienced Researcher Qiskit / QuNetSim based Quantum Communication and Quantum Network Security project. Tasks include: • Quantum Key Distribution (QKD) attack simulation (DV-QKD / CV-QKD) • Multi-party Quantum Network simulation (Qiskit + QuNetSim) • Attack Injection & Attack Detection using Machine Learning / GNN (GCN/GAT/GIN) • Implement Random Forest, SVM, PCA for DV/CV-QKD • Graph Neural Network for dynamic attack classification • Secure Routing and Dynamic Path Optimization based on detection results Deliverables: runnable code, dataset, trained models, results visualization, and full documentation. Keywords: Qiskit, QuNetSim, Quantum Communication, Quantum Network, QKD, Attack Simulation, Attack Detection, GNN, ML, SVM, Random Forest, Dynamic Routing, Secure Path Optimization, DV-QKD, CV-QKD, Quantum Replay Attack, MITM, Node Spoofing, Quantum Cryptography, Intrusion Detection. ———————————— Qiskit/QuNetSim Quantum Communication (QKD) Attack Simulation + GNN Detection & Secure Routing Overall Goal Build a quantum communication attack detection and defense experimental platform: Realistically simulate attacks in Qiskit/QuNetSim → Construct datasets → Use machine learning/graph neural networks to identify attacks → Finally achieve dynamic path optimization in multi-party quantum networks, forming a complete loop of “simulation—detection—defense.” ⸻ Experiment 1: Point-to-point QKD (small experiment) A. DV-QKD (week24) • Input features: System parameters s={d, e_d, N, η, L} + 32 groups of measurement counts (strong/medium × encoding state × basis × detection method) = 37-dimensional vector. • Task: Classification (identify device defects/attack types) + Regression (estimate errors or attack amplitudes). • Method: MultiOutput + Random Forest, PCA reduced to 11 dimensions still maintains ~98% accuracy, test latency ~2s, suitable for online detection. • Defects/attacks: Preparation error, phase modulation error, intensity fluctuation, detector efficiency mismatch, intercept-resend attack, etc. B. CV-QKD (week25) • Attack scenarios: 8 types (No attack, LO, Calibration, Saturation, Wavelength, Blinding, two hybrid types). • Features: 12-dimensional statistics (mean, variance, skewness, kurtosis, LO, intrinsic noise, etc.). • Method: MADS framework = DBSCAN (denoising) + multi-class SVM (Gaussian kernel), compare with “SVM only” accuracy. • Extension: Suggest trying GIN graph neural networks, classify 5 types of attacks, accuracy 86–99%. ⸻ Experiment 2: Multi-party quantum network (large experiment, priority) Content (week44) • Simulation environment: Qiskit/QuNetSim, simulate multi-node communication topology. • Attack types: Calibration, LO, Wavelength, Saturation, Blinding, Quantum Replay, MITM, Relay Gate Tampering, Interference, Entanglement Hijack, Node Spoofing, Collusion, Route Poisoning, Persistent Path, T-Gate Weakening, etc. • Graph modeling: • Node features: QBER, communication frequency, relay or not, attacked or not; • Edge features: Gate sequence similarity, channel loss, attack label; • Each round of communication forms a directed graph, multiple rounds form a dynamic graph sequence. • Detection model: GNN (GCN/GAT/GIN), identify attacked nodes/edges and classify attack types. • Secure routing: Calculate risk scores based on GNN detection results, perform dynamic path optimization, output before-and-after comparison of path selection and performance/security. ⸻ Deliverables (source code files packaged and runnable) 1. Runnable code: Simulation scripts (Qiskit/QuNetSim), data generation and labeling, ML/GNN training and inference code. 2. Data/models: Sample dataset or generation script + random seed, training logs and model weights. 3. Metrics and visualization: • Point-to-point: Confusion matrix, F1-score, PCA curve, online detection latency; • Multi-party: GNN classification performance, visualization of attacked nodes/edges, before-and-after comparison of routing optimization. 4. Documentation: Environment installation files ([login to view URL]), README one-click running guide, method and results explanation. Based mainly on the PPT file contents, we will communicate in a timely manner during the process to ensure everything goes smoothly.（Need to answer questions in a timely manner, provide guidance, and report progress promptly.)

Project ID: 39737176