yxiao92 / wifi-localisation

Using WiFi signals (RSSI values) to predict indoor locations

Indoor WiFi Positioning

This study uses user-generated WiFi fingerprints to derive locations at different granularity levels.

For optimal results, each room location should have a distinct set of repeated access points with varying signal intensity.

File structure

• data/: UJIIndoorLoc data (original and subsets)
• notebooks/: Jupyter notebooks for classification
• results/: K-Means clustering results
• utils/: helper functions
• visualisation/: visualisation plots

Results

• Building classification: 99.88% of test accuracy
• Floor classification: 88.56% to 93.44% of test accuracy
• Room classification: varied accuracy scores