Tools

Software tools

The software tools employed in this thesis are the following:

• Android SDK 33: development of Background Sensors, WearOS Sensors (see Data collection libraries) and TUG Test Smartwatch APP (see HAR in mHealth: TUG test using smartphones and smartwatches).
• Node v16.13, JDK 11 and NativeScript CLI v8.2.3: development of NativeScrit WearOS Sensors, AwarNS Phone Sensors and Wear OS (see Data collection libraries), TUG Test Smartphone APP (see HAR in mHealth: TUG test using smartphones and smartwatches).
• Espressif IoT Development Framework v5.1: development of CSI data collection tools employed in Looking into the future: Wi-Fi CSI based HAR.
• Python 3.9.15: tasks related to ML, DL and data analysis. The following Python modules have been employed for such tasks:

requirements.txt

alive_progress==3.0.1       # Animated progress bar employed for long-lasting processes
h5py==3.6.0                 # Numpy dependency
hampel==1.0.2               # Implementation of Hampel filter
itables==2.0.0              # Interactive displaying of DataFrames
keras==2.10.0               # API for building ML and DL models
keras-tuner==1.4.6          # API for tuning (e.g., Grid Search) ML and DL models
numpy==1.23.2               # Manipulation of multidimensional arrays 
pandas==1.5.2               # Manipulation of tabular data (i.e. DataFrames)
pingouin==0.5.3             # Statistical tests
plotly==5.14.0              # Interactive plotting library
python-dateutil==2.8.2      # Utils for manipulation of dates
PyWavelets==1.4.1           # Implementation of Wavelet transforms
scikit-learn==1.2.0         # ML and DL tools
scipy==1.10.0               # Algorithms for scienific computing (e.g., statistics, signal filters, etc.)
tensorflow==2.10.0          # Backend for running ML and DL models

Hardware tools

Following, the harware devices used during the development of this thesis are listed.

• Smart devices:
  • TicWatch Pro 3 GPS (WH12018): IMU data collection for Smartphone and smartwatch HAR dataset and system evaluation in HAR in mHealth: TUG test using smartphones and smartwatches.
  • Xiaomi Poco X3 Pro (M2102J20SG): IMU data collection for Smartphone and smartwatch HAR dataset and system evaluation in HAR in mHealth: TUG test using smartphones and smartwatches.
  • Xiaomi Poco F2 Pro (M2004J11G): video recordings for Smartphone and smartwatch HAR dataset and Localized HAR based on Wi-Fi CSI, and system evaluation in HAR in mHealth: TUG test using smartphones and smartwatches.
• Microcontrollers:
  • ESP32-S2 WROOM: CSI data collection for Localized HAR based on Wi-Fi CSI.
  • ESP32-S3 WROOM: CSI data collection for Temporal stability of Wi-Fi CSI data from ESP32 microcontrollers.
• Computers:
  • Windows 10 PC with i7-8700 CPU, NVIDIA GeForce GTX 750 GPU and 16 GB RAM: ML and DL models.
  • MacBook Air M1 16GB RAM: development and analysis tasks.
• Other:
  • TP-Link Archer C80 Router: CSI data collection for Localized HAR based on Wi-Fi CSI and Temporal stability of Wi-Fi CSI data from ESP32 microcontrollers.