Huber Estimator and Statistical Bootstrap Based Light-Weight Localization for IoT Systems

In this paper we study distributed cooperative localization algorithms for wireless sensor networks (WSN) and propose a novel approach that remarkably improves the accuracy of the localization in Internet of Things (IoT) systems. In our work, we focused on indoor localization problem in IoT infrastructures by investigating a two step position refinement method. We first utilize an iterative gradient descent method on the localization cost function based on Huber estimator. In the second step, we refine the position estimates given by the first step. We indeed use a statistical bootstrapping approach to deal with impairments due to Non-Line-Of-Sight (NLOS) disruptive effects and the wireless channel noise impact on range measurements; and once again, we run the gradient descent. The simulation results showup the robustness of our algorithm since it achieves better accuracy comparing to similar existing approaches.