An LTE Fingerprint Localization Method based on Combined KNN and LSTM Prediction
Haosen Wang, Lijie Zhang · 2024
With the comprehensive coverage of the LTE network, LTE fingerprint localization has become one of the research hotspots of indoor localization. In the offline phase, this paper uses the performance metrics of LTE Cell Reference Signal (CRS) as location fingerprint features to build a fingerprint database. In the online stage, the fingerprint matching algorithm is a key factor affecting the localization accuracy, to make full use of the structural information and temporal correlation of sequential fingerprint features, this paper proposes a LTE fingerprint localization method based on the combination of KNN and LSTM prediction. The method adopts the K-Nearest Neighbor algorithm (KNN) and Long Short Term Memory (LSTM) to predict the moving target location simultaneously and fuses the prediction results with the residual inverse method. The experimental results show that the localization accuracy of the proposed algorithm in this paper is improved by 9.02% and 33.71% over the KNN and LSTM alone for LTE fingerprint localization.