A Novel Learning-Based Indoor Localization Method Combined with Room Features

The requirement for location information to enable mass of mobile applications in indoor environments has motivated research efforts in the past decades. Indoor localization method using fingerprint has turned out to be more considerable recently. However, the complex indoor radio propagation environment affects the positioning accuracy. In order to improve accuracy, we propose the ANN-XGBoost model, which combine Artificial Neural Network (ANN) and XGBoost, to fuse Received Signal Strength (RSS) and room features for positioning. The experiment results of the new model are promising, which perform better than traditional algorithms.