Deep Learning-based Localization Using Spatial Information in the NLoS Scenarios

In this paper, we propose a new type of deep learning (DL)-based localization for the urban NLoS scenarios, termed Intelligent Localization via Spatial Information Embedding (I-LOSIE). In I-LOSIE, we train the deep neural network (DNN) to identify the location of the wireless device by using the combination of the uplink measurements and the position and size of the obstacles (we call these spatial information). From the extensive numerical evaluations, we show that the proposed I-LOSIE achieves the high-resolution localization in the wireless environments with high density of obstacles..