---
dataset_info:
  features:
  - name: features
    sequence: float32
    length: 115
  - name: attack
    dtype:
      class_label:
        names:
          '0': benign_traffic
          '1': combo
          '2': junk
          '3': mirai-ack
          '4': mirai-scan
          '5': mirai-syn
          '6': mirai-udp
          '7': mirai-udpplain
          '8': scan
          '9': tcp
          '10': udp
  - name: device
    dtype:
      class_label:
        names:
          '0': Danmini_Doorbell
          '1': Ecobee_Thermostat
          '2': Ennio_Doorbell
          '3': Philips_B120N10_Baby_Monitor
          '4': Provision_PT_737E_Security_Camera
          '5': Provision_PT_838_Security_Camera
          '6': Samsung_SNH_1011_N_Webcam
          '7': SimpleHome_XCS7_1002_WHT_Security_Camera
          '8': SimpleHome_XCS7_1003_WHT_Security_Camera
  splits:
  - name: train
    num_bytes: 2857231888
    num_examples: 6002588
  - name: test
    num_bytes: 504568568
    num_examples: 1060018
  download_size: 1772922927
  dataset_size: 3361800456
license: cc-by-4.0
pretty_name: nbaiot
---

# Dataset Card for N-BAIoT

*From https://archive.ics.uci.edu/dataset/442/detection+of+iot+botnet+attacks+n+baiot:* This dataset addresses the lack of public botnet datasets, especially for the IoT. It suggests *real* traffic data, gathered from 9 commercial IoT devices authentically infected by Mirai and BASHLITE.

## Dataset Details

### Dataset Description

*From https://archive.ics.uci.edu/dataset/442/detection+of+iot+botnet+attacks+n+baiot:*
(a) Attribute being predicted:
-- Originally we aimed at distinguishing between benign and Malicious traffic data by means of anomaly detection techniques.
-- However, as the malicious data can be divided into 10 attacks carried by 2 botnets, the dataset can also be used for multi-class classification: 10 classes of attacks, plus 1 class of 'benign'.


(b) The study's results:
-- For each of the 9 IoT devices we trained and optimized a deep autoencoder on 2/3 of its benign data (i.e., the training set of each device). This was done to capture normal network traffic patterns.
-- The test data of each device comprised of the remaining 1/3 of benign data plus all the malicious data. On each test set we applied the respective trained (deep) autoencoder as an anomaly detector. The detection of anomalies (i.e., the cyberattacks launched from each of the above IoT devices) concluded with 100% TPR.



- **Curated by:** Meidan, Yair, Bohadana, Michael, Mathov, Yael, Mirsky, Yisroel, Breitenbacher, Dominik, , Asaf, and Shabtai, Asaf
- **License:** [Creative Commons Attribution 4.0 International (CC BY 4.0)](https://creativecommons.org/licenses/by/4.0/legalcode)

### Dataset Sources

- **Repository:** https://archive.ics.uci.edu/dataset/442/detection+of+iot+botnet+attacks+n+baiot
- **Paper:** https://arxiv.org/abs/1805.03409

## Citation

**BibTeX:**

@misc{misc_detection_of_iot_botnet_attacks_n_baiot_442,
  author       = {Meidan,Yair, Bohadana,Michael, Mathov,Yael, Mirsky,Yisroel, Breitenbacher,Dominik, ,Asaf, and Shabtai,Asaf},
  title        = {{N-BaIoT Dataset to Detect IoT Botnet Attacks}},
  year         = {2018},
  howpublished = {UCI Machine Learning Repository},
  note         = {{DOI}: https://doi.org/10.24432/C5RC8J}
}

**APA:**

Meidan, Yair, Bohadana, Michael, Mathov, Yael, Mirsky, Yisroel, Breitenbacher, Dominik, ,Asaf, and Shabtai, Asaf. (2018). N-BaIoT Dataset to Detect IoT Botnet Attacks. UCI Machine Learning Repository. https://doi.org/10.24432/C5RC8J.

## Glossary [optional]

- **IoT**: Internet of Things
- **Botnet**: A collection of devices that are maliciously controlled via malware