Dataset Viewer

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.

This dataset contains a comprehensive collection of 2D amplitude-phase histograms derived from raw IEEE 802.11 (Wi-Fi) IQ signal captures. It is intended to support research in Automatic Modulation Classification (AMC) under realistic and challenging indoor wireless propagation environments.

The dataset is particularly useful for:

  • modulation recognition and classification
  • machine learning and deep learning for RF sensing
  • channel robustness studies
  • SNR-aware signal analysis
  • indoor Wi-Fi signal characterization

There are 4 distinct datasets: (1) a synthetic dataset incorporating multipath fading and noise under perfect synchronization, (2) a synthetic dataset including fading, noise, and synchronization impairments, (3) a conducted hardware dataset acquired through cable-based measurements, and (4) an OTA dataset.


Dataset Structure

The dataset follows a hierarchical directory structure that enables easy filtering by:

  • channel model / physical scenario
  • SNR level (dB)

Path Format

[Channel_Model]/[SNR_dB]/[Filename].mat

File Naming Convention

Each file name encodes important metadata about the captured signal.

Filename Pattern

H_{Format}_{Channel}_{Modulation}_{SNR}_{SampleID}.mat

Component Breakdown

Component Description
H Static prefix indicating histogram data
{Format} HT: High Throughput (IEEE 802.11n)
NHT: Non-High Throughput
{Channel} Single-letter identifier for the propagation scenario
{Modulation} Integer label representing the modulation type
{SNR} Signal-to-Noise Ratio in dB
{SampleID} Unique sample index

Modulation Label Mapping

The modulation field in the filename maps directly to the number of bits per symbol.

Index Modulation Bits per Symbol
1 BPSK 1
2 QPSK 2
4 16-QAM 4
6 64-QAM 6

Notes

  • All samples are stored in MATLAB .mat format.

  • Histograms are generated from raw IQ captures.

  • The nested folder structure makes it easy to create:

    • train/validation/test splits by channel
    • SNR-specific subsets
    • modulation-balanced experiments

Citation / Attribution

If you use this dataset in academic work, please cite the associated publication or repository and include attribution under the CC BY-NC 4.0 license. DOI will be available soon

Downloads last month
21