How shophine/unobtrusive-authentication-of-smartphone-users-by-context-aware-behavioural-biometrics works
This repository combines an Android behavioral-biometric authentication app, Node.js endpoints, and MATLAB/libsvm tooling. The intended user journey is sensor-based gait capture, feature and activity analysis, then identity/trust assessment used to control app access. The README documents processing stages and security capabilities; sampled code confirms SVM training/prediction adapters and server file-based endpoints. Android implementation and most MATLAB analysis stages were not sampled, so their internal links are shown only where documented or left unverified.
Main entry points: nodeserver.js Request API · MainActivity.java App entry · Matlab Activity recognition · svmtrain.c SVM training
Components 19
Android app
- App entry
Android/app/src/main/java/com/example/myapplication/MainActivity.java - Sensor capture
Android/app/src/main/java/com/example/myapplication - App access control
Android/app/src/main/java/com/example/myapplication/AppRestrictionHomeScreen.java - Geofencing
Android/app/src/main/java/com/example/myapplication
Biometric analysis
- Noise elimination
Matlab - Gait segmentation
Matlab - Gait features
Matlab - PCA
Matlab - Activity recognition
Matlab
Backend API
- Request API
Server/nodeserver.js - Log file
- Prediction files
SVM model
- SVM training
Matlab/svmtrain.c - SVM prediction
Matlab/svmpredict.c - Model conversion
Matlab/svm_model_matlab.c - Dataset reader
Matlab/libsvmread.c - Dataset writer
Matlab/libsvmwrite.c
Other components
- Phone user
- Android device
Connections 19
- Phone user → App entry: uses
- App entry → Sensor capture: collects sensors
- Sensor capture → Noise elimination: sensor data
- Noise elimination → Gait segmentation: cleaned data
- Gait segmentation → Gait features: gait cycles
- Gait features → PCA: gait patterns
- PCA → Activity recognition: features
- Activity recognition → SVM prediction: classification input
Matlab/svmpredict.c - SVM training → Model conversion: converts model
Matlab/svmtrain.c - SVM prediction → Model conversion: reads model
Matlab/svmpredict.c - Dataset reader → SVM training: loads samples
- SVM training → Dataset writer: exports samples
- Android device → Request API: sends requests
- Request API → Log file: writes logs
Server/nodeserver.js - Request API → Prediction files: reads prediction
Server/nodeserver.js - Request API → Prediction files: writes request input
Server/nodeserver.js - App entry → App access control: shows access
- Activity recognition → App access control: informs trust
- Geofencing → App access control: context for access