AN EFFICIENT METHOD FOR BLE INDOOR LOCALIZATION USING SIGNAL FINGERPRINT

Abstract

The emergence of Bluetooth Low Energy (BLE) technology has created many opportunities for indoor localization. However, extracting fingerprint features from the Received Signal Strength Indicator (RSSI) values of Bluetooth signals often yielded results with significant errors and instability. This study utilizes a Kalman filter to stabilize received RSSI values. It employs Autoencoder and Convolutional Autoencoder models to extract distinctive features and compares random test points with reference points in a database using normalized cross-correlation.