Enhanced LoRaWAN RSSI Indoor Localization Based on BP Neural Network

Abstract

Received Signal Strength Indicator (RSSI) technique is often used to estimate locations. However, this technique often leads to considerable errors in the real-world environment. This paper proposes a filter to remove outliers according to the Log-normal shadowing model and leverage a Back-Propagation (BP) neural network to predict the unknown positions. The results of our work prove that the BP neural network with the proper outlier filter can enhance the accuracy of indoor localization based on RSSI. We achieved an average distance error of 0.5971 meters in the test set, which has higher accuracy and more robustness.