Prediction of TOA-Based Localization Accuracy Using CRLB and 3D Buildings with Field Trial Validation

In Low Power Wide Area (LPWA) networks, radio localization based on Time of Arrival (TOA) measurements collected from gateways synchronized on GPS time is an appealing technology for Internet of Things (IoT). However, it is hard to predict the actual accuracy that could be expected from a given deployment without taking into account the propagation channel complexity. This paper proposes a new approach to generate a localization accuracy map (LAM) based on Cramér-Rao lower bound (CRLB) and 3D buildings models to predict the propagation conditions. A comparison with measurements from LPWA field trials conducted in the city of Grenoble (France) is presented, comprising more than 300000 LPWA transmissions collected. We show that for an infrastructure composed of 6 LPWA gateways, our LAM is able to predict the performance obtained with an maximum likelihood estimator (MLE) localization algorithm with only 12% of mismatch.