Fingerprinting based localization with channel state information features and spatio-temporal deep learning in long term evolution femtocell network: An experimental approach

Abstract

The need for accurate indoor localization for related services and radio resource management have imparted increased research attention on fingerprinting based positioning. In this work, we present a novel spatio-temporal deep learning (STDL) based fingerprinting method for indoor localization utilizing Channel State Information (CSI). The system works in two phases, acquisition and training and tracking. Experimental evaluation of the method for LTE femtocell signal with Software Defined Radio (SDR) hardware found to outperform spatial and temporal Received Signal Strength (RSS), spatial CSI and temporal CSI based methods.