IoT-23

Dataset Overview

Dataset Name

• IoT-23

Source

• Provider: Stratosphere Laboratory

Content

• Labeled IoT network traffic, including both benign and malicious flows.
• The lightweight version excludes PCAP files and contains only the labeled flows.

Size

• Compressed download: 8.8 GB
• Extracted size: ~44 GB

Use Cases

• Research on network‑traffic anomaly detection and classification.
• After download and extraction, data are processed with specific Python scripts.

Processing Steps

1. Data Extraction: Extract traffic from scenario files, producing separate benign and attack flow files.
2. Data Shuffling: Split large files and shuffle randomly to improve sample reliability.

Dependencies

• Programming language: Python 3.8.8
• ML tools: scikit‑learn 0.24.1
• Scientific computing: NumPy 1.19.5
• Data analysis: pandas 1.2.2
• Visualization: matplotlib 3.3.4, seaborn 0.11.1
• System info: psutil 5.8.0
• Model serialization: pickle

Configuration

• Config file: config.py (set dataset and experiment paths).
• Config check: run the provided script to verify correctness.

Experiments

• Demo: Quick verification with 10 000 records.
• Full: Process >20 million records (≈24 h).
• Custom: TODO