SLIDE 1 Tuan Anh Nguyen, Marco Aiello
University of Groningen, The Netherlands {t.a.nguyen, m.aiello}@rug.nl
A Self-healing Framework for Online Sensor Data
Kenji Tei
National Institute of Informatics, Japan tei@nii.ac.jp
Takuro Yonezawa
Keio University, Japan takuro@ht.sfc.keio.ac.jp
SLIDE 2
SLIDE 3
SLIDE 4
Self-healing process of natural systems
SLIDE 5
MAPE-K architecture
SLIDE 6
The framework
SLIDE 7 The framework
Models:
model
Pre-processing:
- Historical data
- Neighbours
Fault detection and classification Execute:
- Correct faults
- Notify users
Fault correction
value
- Model learning: f(x) = vf
SLIDE 8
SLIDE 9
Fault model
SLIDE 10
SLIDE 11
Fault detection and classification
SLIDE 12
Ground Truth Actual readings of a Temperature sensor Median of Neighbour readings Forecast value with ARMA
SLIDE 13
Ground Truth Actual readings of a Temperature sensor Median of Neighbour readings Forecast value with ARMA Intersection Result
SLIDE 14
Experiment: Intel Lab Dataset
SLIDE 15
Environment Model
SLIDE 16
Neighbouring: k-means++ with Dynamic Time Warping (DTW) as a distance
K = 2 K = 3
SLIDE 17
First results with Intel Lab Dataset
SLIDE 18
City Data Process
SLIDE 19
The City Data Processing architecture
SLIDE 20
Santander sensors
SLIDE 21 K = 2 K = 3
Neighbouring: k-means++ with Dynamic Time Warping (DTW) as a distance
SLIDE 22
K = 2 K = 3
SLIDE 25 Self-healing for Santander sensors
http://sox.ht.sfc.keio.ac.jp:54380/show/9652237040c8e344a2d553773f5feea0
SLIDE 26 Next steps: Fault correction
Model learning with statistical pattern recognition
- 1. expectations of correct behaviour established at the
calibration phase
- 2. historical sensor data.
SLIDE 27
Real-world implementation
SLIDE 28 Tuan Anh Nguyen
t.a.nguyen@rug.nl
Thank you very much for your attention!
A Self-healing Framework for Online Sensor Data
