COMBINED TECHNIQUES OF INDOOR POSITIONING SYSTEM USING BLUETOOTH LOW ENERGY
Authors/Creators
- 1. Computer Science Department, BINUS Graduate Program - Master of Computer Science, Bina Nusantara University, Jakarta, Indonesia, 11480
Description
This study tried a combination of several methods for the Indoor Positioning System covering all stages, RSSI Filtering, BLE Selection, and Location Estimation, where the fingerprinting approach used by most other studies is partial. In the RSSI Filtering stage, Kalman Filter and Autoencoder methods are used, BLE Selection stage using Strongest Signal and Fisher Criterion. Position estimation uses the WKNN algorithm and the Kalman Filter. Two static and dynamic datasets (including the movement of an object) are used to measure the accuracy of these combined methods. Overall, the best combination on the static dataset is the Kalman filter and the Strongest Signal Criterion which results in an accuracy of 2.44m, and for the dynamic dataset is the Strongest Signal and WKNN + Kalman Filter with an accuracy of 1.91m.