LANL2 Network Traffic
The LANL2 tensor was formed from publicly available anonymized NetFlow data collected over 58 days, published by Los Alamos National Laboratories (LANL). The modes represent timestamp-source-destination-destination port-bytes where the timestamp is binned by 10 minute intervals and the number of bytes transferred is binned in logarithmic scale. The non-zero values are the number of incidents. More detailed information is available at the LANL Cyber Security Research Data Sets website.
Tensor Statistics
| Non-zeros | 69,082,467 |
| Order | 5 |
| Dimensions | 3,761 x 11,154 x 8,711 x 75,147 x 9 |
| Tags | count , network |
Downloadable Files
| File | Description |
|---|---|
| lanl2.tns.gz | LANL2 NetFlow tensor |
| map_mode_0.map.gz | Timestamp |
| map_mode_1.map.gz | Source device |
| map_mode_2.map.gz | Destination device |
| map_mode_3.map.gz | Destination port |
| map_mode_4.map.gz | Bytes transferred |
Citation
@INPROCEEDINGS{9622828,
author={Ranadive, Teresa M. and Baskaran, Muthu M.},
booktitle={2021 IEEE High Performance Extreme Computing Conference (HPEC)},
title={An All–at–Once CP Decomposition Method for Count Tensors},
year={2021},
volume={},
number={},
pages={1-8},
keywords={Tensors;Conferences;Telecommunication traffic;Big Data;Optimization;Big Data Analytics;Count Tensor Decomposition;Generalized Gauss-Newton;High Performance Computing},
doi={10.1109/HPEC49654.2021.9622828}}
}
License
Public Domain
To the extent possible under law, Los Alamos National Laboratory has waived all copyright and related or neighboring rights to Comprehensive, Multi-Source Cyber-Security Events. This work is published from: United States.