A Deep Neural Network-Based LOS Classification for Triangulation Positioning

We propose a deep neural network based a line-of-sight classification for triangulation positioning. The accuracy of triangulation decreases in the non-line-of-sight environments but finding the channel which has a line-of-sight path is difficult in the mathematical method. A proposed scheme can solve this problem by taking advantage of deep neural network. Performance evaluation is carried out by the system level simulator of Long Term Evolution-Advanced with Spatial Channel Model. Performance results indicate that the proposed scheme can effectively increase the positioning accuracy and robustness of the inaccurate received signal strength measurements.