📡 Intelligent QoS Prediction for 5G Network Slices using Machine Learning

🚀 Overview

This project presents an AI-powered system for predicting and optimizing Quality of Service (QoS) in 5G network slices. It leverages machine learning models to forecast network performance metrics and provide actionable recommendations for improving QoS in real-time. The system is deployed as an interactive web application using Streamlit, enabling users to simulate network conditions and analyze their impact on QoS.

🎯 Problem Statement Traditional network management systems are reactive and fail to detect subtle performance degradations (e.g., jitter, latency spikes) in dynamic 5G environments. This project addresses: Lack of predictive QoS monitoring Inability to detect “grey failures” No intelligent decision-support system

💡 Solution We propose a predictive + advisory system that: Predicts QoS using machine learning Classifies network performance (High / Medium / Low) Identifies 5G slice types (eMBB, URLLC, mMTC) Provides optimization recommendations Enables real-time simulation via web app

🧠 Machine Learning Models Used Linear Regression Random Forest (Best performing) Gradient Boosting 📊 Model Performance

Model R² Score Linear Regression: 0.7587 Random Forest: 0.9460 Gradient Boosting: 0.9398 Random Forest achieved the highest accuracy and was selected for deployment.

⚙️ Features 🔮 QoS Prediction Predicts network QoS based on input parameters: Speed SNR (Signal-to-Noise Ratio) MCS (Modulation Coding Scheme) Resource Blocks 📡 5G Slice Classification Automatically identifies slice type: eMBB → High throughput URLLC → Low latency mMTC → IoT devices ⚡ QoS Categorization HIGH → Optimal performance MEDIUM → Acceptable performance LOW → Needs improvement 💡 AI-Based Recommendations Provides actionable suggestions such as: Increase signal strength (SNR) Allocate more resource blocks Improve modulation scheme 📈 Real-Time Simulation Interactive dashboard allows: Parameter tuning via sliders Live QoS updates Sensitivity analysis graphs 🔍 What-If Simulator Simulates improvements: Compare current vs improved QoS Visualize performance gain Decision-support system

🌐 Web Application The system is deployed using: Streamlit Run Locally Bash pip install -r requirements.txt python -m streamlit run app.py 📁 Project Structure

5g_qos_prediction_system/ │ ├── app.py # Streamlit web app ├── rf_model.pkl # Trained ML model ├── requirements.txt # Dependencies ├── train.csv # Dataset (optional) └── README.md # Project documentation

🛠️ Tech Stack Python Pandas NumPy scikit-learn Matplotlib Joblib Streamlit

📊 Key Insights QoS is highly influenced by: Signal strength (SNR) Resource allocation (RBs) Modulation scheme (MCS) Tree-based models outperform linear models due to non-linear network behavior Different 5G slices exhibit distinct QoS characteristics

🏆 Results Achieved R² score of 0.94 using Random Forest Successfully built a real-time QoS prediction system Implemented interactive AI dashboard Enabled proactive network optimization

🔮 Future Scope Integration with real-time telecom data Deployment in edge computing environments Advanced explainability using SHAP Integration with network orchestration systems

📌 Conclusion This project demonstrates how AI can transform 5G network management from reactive to proactive by predicting QoS and enabling intelligent decision-making.