Dark Mode
SAGE Feature Wireless Anomaly Dataset
Data Science and Analytics
Tags and Keywords
Trusted By
"No reviews yet"
Free
About
This resource provides channel measurement data specifically engineered for machine learning applications concerning anomaly detection within wireless communication networks. The data reflects simulations based on the rigorous 3GPP TR 38.901 standard channel models. The features were meticulously extracted using the Space-Alternating Generalized Expectation-Maximisation (SAGE) method, incorporating characteristics derived from real-world propagation dynamics and the QuaDRiGa simulator. It is an essential tool for developing AI solutions aimed at identifying unusual behaviour in wireless environments.
Columns
The dataset contains parameters related to channel measurements, appearing as numerical floating-point data.
The data samples observed indicate seven distinct feature columns. Since detailed headers are unavailable, these represent various extracted channel state information or measurement parameters crucial for subsequent anomaly detection model training.
Distribution
The product is structured into four separate CSV files, corresponding to different environmental conditions. The total size is approximately 113.52 kB.
The files include:
- v1_RMALOS.csv (Rural Macro, Line-of-Sight)
- v1_RMANLOS.csv (Rural Macro, Non-Line-of-Sight)
- v1_UMALOS.csv (Urban Macro, Line-of-Sight)
- v1_UMANLOS.csv (Urban Macro, Non-Line-of-Sight) Specific numbers detailing the exact row or record count for each file are not available.
Usage
This dataset is highly valuable for training and testing machine learning models focused on wireless anomaly detection. It is ideally suited for researchers and engineers developing algorithms that differentiate normal wireless communication behaviour from unusual or malicious activity. It enables the precise modelling of various propagation conditions, aiding in the robust evaluation of AI detection mechanisms.
Coverage
The data covers two principal propagation environments: Rural Macro (RMa) and Urban Macro (UMa) scenarios. These environments are further differentiated by Line-of-Sight (LoS) and Non-Line-of-Sight (NLoS) conditions. The data is generated via simulation based on established channel standards, ensuring accurate representation of these scenarios. The dataset does not contain specific geographic locations, time ranges, or demographic data, as it models physics-based channel measurements. The expected update frequency for this version of the dataset is stated as never.
License
Attribution 4.0 International (CC BY 4.0)
Who Can Use It
- Machine Learning Engineers: For developing and tuning deep learning or statistical models for network intrusion and anomaly detection.
- Academics and Researchers: For studying the impact of various propagation conditions on communication integrity and model performance.
- Cybersecurity Analysts: For establishing baseline models of 'normal' wireless activity to better identify deviations that signal security threats.
Dataset Name Suggestions
- AI-Driven Wireless Network Anomaly Detection Dataset
- 3GPP Channel Measurement Anomaly Data
- SAGE Feature Wireless Anomaly Dataset
- RMa and UMa Wireless Anomaly Detection Data
Attributes
Original Data Source: SAGE Feature Wireless Anomaly Dataset
QUALITYLoading...