Indoor localization algorithm based on Bluetooth location fingerprinting and CNN

Abstract

To improve the accuracy of Bluetooth signal strength (RSSI)-based indoor localization algorithm, a localization algorithm based on location fingerprint localization fusing convolutional neural network and Kalman filter is proposed. The iBeacon is used as the AP node, the RSSI location fingerprint localization method is used, the offline fingerprint library is trained by convolutional neural network, and finally the observed location derived from convolutional neural network and the predicted location derived from Kalman filter are weighted and fused, and the KNN algorithm fused with Kalman filter is set as a control test, and it is concluded that the CNN-KF algorithm has more accurate localization accuracy and is closer to the actual trajectory.