Repository for the ECG variability database

Citation Author(s):
Pablo
Perez-Tirador
Universidad San Pablo-CEU
Abraham
Otero
Universidad San Pablo-CEU
Submitted by:
Pablo Perez Tirador
Last updated:
Mon, 10/02/2023 - 06:50
DOI:
10.21227/dwsg-sj50
License:
Creative Commons Attribution
326 Views
Categories:
Signal Processing
Wearable Sensing
Biophysiological Signals
Keywords:
Electrocardiogram (ECG), Cryptographic protocols; Software measurement; Information security; Cellular phones; Wearable computers
0
0 ratings - Please login to submit your rating.
ACCESS DATASET CITE SHARE/EMBED

Abstract 

Wearable and low power devices are vulnerable to side-channel attacks, which can retrieve private data (like sensitive data or the private key of a cryptographic algorithm) based on externally measured magnitudes, like power consumption. These attacks have a high dependence on the data being encrypted -- the more variable it is, the more information an attacker will have for performing it. This database contains ECG data measured with a wearable sensorized garment during different levels of activity. This provides different variability rates that can be used to test the effects of side-channel attacks on low power devices that execute cryptographic functions.

Instructions: 

1. Introduction

This repository contains a database of ECGs collected from a volunteer with different bit resolutions and activity rates. These files have been used to test the effects of data variability on side channel attacks to wearable sensors. All files have been obtained using a custom-made t-shirt with conductive pads and the data was collected using a Bitalino at maximum sample rate (1 kHz).

2. Files

All files are given in CSV format separated by tabs (\t). The files are divided among folders:

  • ecg_10bit: contains data acquired at 10 bit resolution for two scenarios:
    • rest_10bit: volunteer at rest, clean ECG
    • noisy_10bit: volunteer moving, noisy ECG
  • ecg_16_bit: contains data acquired at 10 bit resolution for four scenarios:
    • good_16bit: volunteer lying down, clean ECG
    • average_16bit: volunteer at rest, mostly clean ECG with some noise
    • noisy_16bit: volunteer moving around, ECG with some noise
    • verynoisy_16bit: volunteer moving around and using the stairs, ECG with noise, some segments clip at zero or maximum values
  • ecg_extra: other ECGs recorded with 16 bit resolution

Additionally some Python files are supplied to convert the data into NumPy arrays and filter saturating values.

 

Dataset Files

LOGIN TO ACCESS DATASET FILES

 

QUESTIONS?