WASD

WASD: Wireless Anomaly Signal Dataset

WASD (Wireless Anomaly Signal Dataset) is a specialized dataset crafted for the detection of anomalies within wireless signals in urban environments. This dataset integrates real-world signal measurements from 19 LTE and 5G bands with simulated anomalous signals, enabling the effective training of various object detection networks and the development of robust deep learning models for accurate anomaly detection in practical settings.

Key Features

• Real-world Measurement Data: The dataset features authentic signal data sourced from 19 different LTE and 5G bands, providing a solid foundation for analysis and model development.

• Diverse Anomaly Signals: It includes various types of simulated anomalous signals such as tone, chirp, and pulse signals, enhancing the dataset's utility for extensive testing and validation purposes.

• Realistic Simulations: The dataset incorporates realistic random-based parameters and fading channels configured in accordance with 3GPP standards, simulating actual environmental conditions for more accurate training outcomes.

• Large-scale Dataset: Comprising a total of 85,500 samples across 19 bands, WASD is designed for extensive research and experimentation.

• Versatile Usage: Featuring both object detection and normal data, the dataset accommodates both supervised and unsupervised learning techniques, making it ideal for a wide range of applications.

Dataset Structure

The dataset is organized as follows:

|-- Dataset/
|   |-- IQ data.zip  
|   |   |-- Band name/  
|   |   |   |-- bin/  
|   |   |   |   |-- IQ_date.bin  
|   |   |   |   |-- IQ_date.json  
|   |   |   |   
|   |-- npy data.zip  
|   |   |-- Abnormal/  
|   |   |   |-- Abnormal_spectrogram.npy  
|   |   |-- Normal/  
|   |   |   |-- Normal_spectrogram.npy  
|   |   |-- Label/  
|   |   |   |-- Spectrum_label.csv  

• IQ Data: Contains raw IQ data in .bin and .json formats, measured for each band.
• Spectrogram Data: Features spectrogram data in .npy format, generated through the Short Time Fourier Transform (STFT) technique.
• Labels: Includes labels in .csv format that provide bounding box coordinates and Interference-to-Signal Ratio (ISR) details for anomaly signals.
• Data Loaders: Comes with example Jupyter Notebooks which facilitate the loading of IQ and spectrogram data into Python environments.
• Anomaly Signal Generation Examples: Provides additional Jupyter Notebooks that outline the process of generating abnormal wireless signals (tone, chirp, pulse) using the fading model, along with detailed explanations available in the associated research paper.

Dataset Access

The dataset can be accessed through the following links:

• Chunk 1
• Chunk 2
• Chunk 3
• Chunk 4
• Chunk 5
• Chunk 6
• Chunk 7
• Chunk 8
• Chunk 9
• Chunk 10
• Chunk 11
• Chunk 12
• Chunk 13
• Chunk 14
• Chunk 15
• Chunk 16

Additional Information

For further information on the dataset, refer to the associated research paper available at DOI:10.1109/ACCESS.2024.3521946. This work is essential for advancing anomaly detection methodologies in wireless communications.

Contributions

The WASD dataset has been developed by Jinha Kim, Hyeongwoo Kim, and Byungkwan Kim, with support from the MSIT/IITP through the ICT R&D program.