Network Traffic Headers from 12 Containerized Applications with Activity Labels

Twelve containerized applications, including Firefox, Slack, and VLC, had their network traffic headers captured during the first two minutes of execution. Each traffic sample is labeled with the specific activity the application was performing, such as downloading, browsing, or audio streaming. Mozhdeh Farhadi from the Centre National de la Recherche Scientifique created this dataset for analyzing application-specific network patterns.

Use Cases

• Train a classifier to identify application types based on network traffic headers.
• Analyze the network footprint of specific activities like video streaming or file downloading.
• Benchmark network traffic analysis tools against labeled, application-specific data.
• Study the differences in network behavior between similar applications (e.g., mpg123 vs. mplayer for audio streaming).

Strengths

• Data is labeled with specific application activities, providing clear ground truth for supervised learning.
• Includes traffic from 12 distinct, isolated applications, offering a variety of behavioral patterns.
• Traffic capture is standardized to the initial two minutes of application execution.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment for large-scale model training.
• Last update date is unknown; freshness unverified.

Provenance

Source

Mozhdeh Farhadi, Centre National de la Recherche Scientifique

Collection Method

Network traffic headers captured from 12 containerized applications during the first two minutes of execution.