The dataset viewer is not available because its heuristics could not detect any supported data files. You can try uploading some data files, or configuring the data files location manually.
YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
RF Signal Classification Dataset
This dataset contains raw IQ samples across four signal classes used to train and evaluate the RF Signal Classification models.
Classes
| Class | Description |
|---|---|
| ADS_B | Aircraft transponder signals at 1090 MHz |
| FM_broadcast | Commercial FM radio signals (91.1, 93.5, 95.0, 98.3, 104.8, 106.4 MHz) |
| ISM_sensors | ISM-band device signals at 433 MHz |
| noise | Background RF noise captured across 50, 300, 470, 800, 1200 MHz |
160 recordings per class, 640 total.
Sources
The dataset is a combination of three sources:
- TrevTron/rtl-ml-dataset — publicly available RTL-SDR IQ captures
- Simulink simulations — synthetically generated IQ signals simulated in MATLAB Simulink
- Live RTL-SDR captures — signals captured using a Nooelec NESDR SMArt v5 SDR across multiple capture sessions
Format
Each recording is a .npy file containing a Python dict with the following fields:
{
"samples": np.ndarray, # complex64 IQ samples
"center_freq": float, # center frequency in Hz
"sample_rate": float, # sample rate in Hz (default: 1,024,000)
"timestamp": str, # capture timestamp
"label": str # signal class name
}
Structure
datasets_validated/
├── ADS_B/ — 160 recordings
├── FM_broadcast/ — 160 recordings
├── ISM_sensors/ — 160 recordings
└── noise/ — 160 recordings
Split
Files are split at the recording level (no leakage between splits):
| Split | Ratio |
|---|---|
| Train | 70% |
| Validation | 10% |
| Test | 20% |
Hardware
Live captures were recorded using:
Nooelec NESDR SMArt v5 — HF/VHF/UHF (100 kHz–1.75 GHz), RTL2832U & R820T2
Default sample rate: 1,024,000 samples/second
- Downloads last month
- 67