Robust Cooperative Localization Technique for Wireless Sensor Networks

A novel cooperative localization algorithm for wireless sensor networks (WSNs) using robust estimation is proposed. We divide the WSN into different layers and each layer builds its local coordinates. In this context, we study how to build the local coordinates based on realistic assumptions and the geometrical dilution of precision (GDOP) of the network. We introduce a modified network adjustment (NA) technique which is commonly used in surveying applications to minimize the ranging error within the WSN layers. We propose combined estimators using L₁ norm and L₂ norm to mitigate large ranging errors (outliers) commonly found in weak non-line of sight (NLOS) signals. We derive a robust 2D coordinate transformation model to merge and transform the local coordinates to a global coordinate system considering imperfections of the local coordinates. Simulation results are shown to evaluate the proposed algorithm.