Skip to main content
UCI Machine Learning Repository

Kitsune Network Attack: Nine Attacks on IP Cameras and IoT

IoT Security & Intrusion Detection IoT Security & Intrusion Detection
164 views
2 min read
27,170,754 rows · 115 columns
License
Catalog metadata: This page is a discovery record, not publisher documentation. Verify the description, schema, provenance, version, licence, and citation at the linked source before use.

Catalog Summary

"Nine attack captures for network intrusion detection, with a documented total of 27,170,754 packets, 115 statistical features per packet and separate binary labels."

Catalog Notes

Overview

Kitsune provides traffic from an IP surveillance network and an IoT network for evaluating intrusion detection. The original paper describes an eight-camera surveillance deployment and a separate network containing nine IoT devices and three PCs.

Attack coverage

  • OS Scan, Fuzzing, Video Injection, ARP MitM and Active Wiretap.
  • SSDP Flood, SYN DoS, SSL Renegotiation and Mirai.

Files and labels

Each capture has packet traces, extracted feature data and a separate label vector. Feature files are distributed as compressed CSVs. The documented total is 27,170,754 packets; the nine counts in the original paper sum to this value. UCI lists a 17.7 GB download.

Labels are 0 for benign and 1 for malicious traffic. In the MitM captures, packets passing through the attacker are labelled 1. Captures start with benign traffic.

Feature count and limitations

The documented feature matrix has 115 columns, computed from 23 statistics over five decaying time windows. This excludes the separate label; appending it produces 116 columns. UCI's headline value of 7 features conflicts with its detailed documentation. The stored column count refers to the feature matrix.

UCI reports no missing values and packet captures truncated to 200 bytes. Truncation alone does not establish complete anonymization. Full archives were not independently scanned for this catalog.

Citation and license

The 2019 UCI record lists no named creators, and its automatically generated citation begins with the dataset title. This catalog attributes the dataset to the four researchers who collected it, as identified in their 2018 paper and code repository: Yisroel Mirsky, Tomer Doitshman, Yuval Elovici and Asaf Shabtai.

Dataset DOI: 10.24432/C5D90Q. Paper: Kitsune: An Ensemble of Autoencoders for Online Network Intrusion Detection. The authors' repository requests citation of this paper when using the data or code.

The UCI record licenses the data under CC BY 4.0. Reuse requires appropriate attribution, a license link and an indication of changes. This catalog description is an independently prepared summary.

View Data Structure

To explore column names, data types, and sample rows, visit the official dataset page on UCI Machine Learning Repository.

Preview on UCI Machine Learning Repository

Cite This Dataset

Mirsky, Y., Doitshman, T., Elovici, Y., & Shabtai, A. (2019). Kitsune Network Attack [Data set]. UCI Machine Learning Repository. https://doi.org/10.24432/C5D90Q

Source metadata: UCI Machine Learning Repository (2019) · DOI: 10.24432/C5D90Q

Indexed by IoTDataset.com on Sep 22, 2026

Review the Source Record

Confirm the licence, version, access conditions, file format, and provenance at the source before use.

Open Source Page

Related Topics & Keywords

Browse all IoT Security & Intrusion Detection datasets

Share This Research

More in IoT Security & Intrusion Detection

View All
Network Security IoTSyn Generated

Synthetic IIoT Network Traffic Dataset — 21% Attacks

Free CC0 synthetic dataset: 4,000 rows of Modbus, OPC UA and DNP3 flows labelled for SCADA intrusion detection. 21% Attacks. Reproducible from its seed.

May 04, 2026
Network Security IoTSyn Generated

Synthetic IIoT Network Traffic Dataset — 29% Attacks

Free CC0 synthetic dataset: 10,000 rows of Modbus, OPC UA and DNP3 flows labelled for SCADA intrusion detection. 29% Attacks. Reproducible from its seed.

Apr 30, 2026
Cybersecurity / IoT Network Security University (Canadian Institute for Cybersecurity)

CICIDS2017 - Comprehensive Network Intrusion Detection Dataset

The most cited cybersecurity dataset worldwide with 2.8+ million network flows capturing 14 types of realistic attack scenarios including DDoS, brute force, botnet, and web attacks alongside benign traffic for advanced intrusion detection systems.

Jan 20, 2026
IoT Cybersecurity UCI Machine Learning Repository

RT-IoT2022

Real-time network traffic dataset from diverse IoT devices including normal behavior and various attacks (DDoS, brute-force, scans) for developing intrusion detection systems.

Jan 26, 2026
Cybersecurity University

MedBIoT — Medium-Sized IoT Botnet IDS Dataset [83 devices]

IoT botnet IDS dataset using 83 real and emulated devices with Mirai, BashLite, and Torii traffic. Raw PCAP files support botnet and anomaly detection research.

Jun 02, 2026
Cybersecurity IoTSyn Generated

Synthetic IoT Intrusion Detection Dataset — 18% Attacks

Free CC0 synthetic dataset: 500 rows of labelled network flows covering DoS, DDoS, botnet and reconnaissance traffic. 18% Attacks.

Apr 04, 2026

Explore other topics

All topics →