z-a-h-r-a/Esprit-PI-4DATA-2026-SmartCityIoT6G

AI-Driven Attack Defense for IoT Smart City in 6G Networks – Developed at Esprit School of Engineering – Tunisia | Academic Year 2025–2026 | AI, IoT, 6G, Cybersecurity # AI-Driven Attack Defense for IoT Smart City in 6G Networks ## Overview This project was developed as part of the PI – 4th Year Engineering Program at **Esprit School of Engineering** (Academic Year 2025–2026). It presents an AI-powered intrusion detection system designed to automatically detect and defend against cyber-attacks (DDoS, MITM, Botnet) targeting IoT devices within Smart City infrastructures operating over 6G networks. The project follows the **CRISP-DM methodology** and evaluates detection models across four network slicing datasets: mMTC, URLLC, eMBB, and G6. ## Features - AI-based cyber-attack detection across 6G network slices (mMTC, URLLC, eMBB, G6) - Multi-model comparison: Logistic Regression, KNN, Random Forest, Gradient Boosting - Slice-specific preprocessing pipelines tailored to each network category - Feature selection with 62–71% dimensionality reduction while preserving detection accuracy - Class imbalance handling via stratified sampling and class-weighted approaches - Model evaluation using Accuracy, Precision, Recall, F1-score, ROC-AUC - Permutation-based feature importance analysis ## Tech Stack ### Language - Python 3.x ### Data Manipulation & Computing - Pandas - NumPy ### Data Visualization - Matplotlib - Seaborn ### Preprocessing & Feature Engineering - Scikit-learn (StandardScaler, RobustScaler, OneHotEncoder, SimpleImputer, KNNImputer, ColumnTransformer, Pipeline) ### Machine Learning Models - Logistic Regression (baseline) - K-Nearest Neighbors (KNN) - Random Forest Classifier - Gradient Boosting Classifier ### Model Evaluation - classification_report, confusion_matrix - roc_auc_score, RocCurveDisplay - Permutation Feature Importance ### Project Organization - Pathlib ## Architecture The project follows the CRISP-DM methodology across 6 phases: ``` Business Understanding → Data Understanding → Data Preparation → Modeling → Evaluation → Deployment ``` ### Datasets Used | Dataset | Rows | Features | Category …