BLE Baseband Power Spectrum Frames for RF Fingerprinting, 40 Transmitters
by Hyeon Park·Updated 1mo ago
6.3 GB1files
Available on 1 platform
Sign in to view source links and access this dataset
Description
Power spectrum frame data derived from Bluetooth Low Energy baseband signals, collected for RF fingerprint-based IoT device authentication. The dataset contains 20,000 NumPy array files (500 per module) from 40 distinct BLE transmitter modules, sampled at 20 MHz in the 2.4 GHz band. It was created by Hyeon Park, G. Cho, and T. Kim from Korea University and last updated on 2026-04 16.
Use Cases
Reproducing experimental results for RF fingerprint-based authentication based on the described framework.
Applying various filter bank configurations (Linear, Gammatone, Mel, Inverse Mel) to the provided spectral frames.
Testing frequency-domain emphasis strategies on pre-processed BLE signal data.
Developing machine learning models for IoT device identification based on per-signal power spectrum representations.
Strengths
Contains 20,000 signal captures (500 per module) from 40 distinct BLE transmitter modules, enabling device differentiation studies.
Data is provided as processed power spectrum frames (1,025 frequency bins per frame), ready for feature extraction.
Accompanied by a specific research paper detailing the collection and processing methodology.
Released under a permissive CC-BY-4.0 license.
Limitations
Column-level documentation is absent; field semantics must be inferred after download.
Row count is unknown, which may limit suitability assessment.
The dataset appears to be from a controlled lab environment with a single receiver, which may limit generalizability.
Provenance
Source
Korea University (H. Park, G. Cho, T. Kim)
Collection Method
BLE advertising signals captured using a HackRF One SDR receiver, processed through pre-emphasis, framing, windowing, and FFT.
Time Range
null
Freshness
Last updated 2026-04-16 03:29:39; freshness should be verified.
Geography
null
Files are in NumPy binary format (.npy) within a 6.3 GB ZIP archive; requires Python and NumPy for use.