Network Anomaly Detection System

A robust machine learning-based system for detecting anomalies in encrypted network traffic using Autoencoder, One-Class SVM, and K-Means clustering.

🚨 Recent Fixes

• Resolved Stuck Execution: Fixed by adding feature normalization in One-Class SVM to prevent kernel underflow.
• Fixed Visualization Errors: Updated visualizer to handle results properly with error checking.
• Improved Model Training: Ensured models train with normalized data and proper thresholds.
• Performance Note: TensorFlow.js Node backend installation may fail if project path has spaces. See installation section.

Quick Start

Prerequisites

• Node.js v18+ (v22 may require additional setup for native builds)
• NPM v9+
• Python 3.6+ (for native compilations if needed)

Installation

1. Clone the repository:

git clone https://github.com/yourusername/network-anomaly-detection.git
cd network-anomaly-detection

Important: If your project path has spaces (e.g., "Network Anomaly Detection"), rename folders to remove spaces before installing dependencies. Example:

mv "Network Anomaly Detection" Network-Anomaly-Detection
cd Network-Anomaly-Detection/network-anomaly-detection

This prevents build errors during native dependency installation.

2. Install dependencies:

npm install

3. (Optional) Install TensorFlow.js Node backend for faster performance:

npm install @tensorflow/tfjs-node

If installation fails (e.g., build errors), ensure no spaces in path and required build tools installed (see Troubleshooting).

📖 Usage

Run with Mock Data

npm start

Analyze PCAP File

npm start /path/to/capture.pcap

Run Tests

npm test

🔧 Troubleshooting

Application Stuck or Errors

• Model Training Errors: Ensure sufficient data; mock generator provides ~2000 packets.
• TypeError in Visualizer: Update to latest code with fixed visualizer.
• SVM Prediction Error: Fixed by normalization; pull latest changes.

Dependency Installation Issues

• node-pre-gyp build failures: Often due to spaces in path or missing build tools.
  • macOS: Install Xcode Command Line Tools: xcode-select --install
  • Also install: brew install pkg-config cairo pango libpng jpeg giflib librsvg
• No pre-built binaries: Try older Node.js version (e.g., v20) or build from source.
• Heap Out of Memory: Run with node --max-old-space-size=4096 src/index.js

Debug Logging

DEBUG=* npm start

📁 Project Structure

• src/capture/: Packet capturing and reading
• src/detection/: Anomaly models and detector
• src/features/: Feature extraction
• src/utils/: Helper functions
• src/visualization/: Results display
• src/index.js: Main entry

📄 License

MIT License - see LICENSE file.