Robust Distributed Estimation for Localization in Lossy Sensor Networks

In this paper we address the problem of fault resilient estimation for large-scale systems, where the measurements are possibly corrupted due to faults of low-cost sensors. As a toy application, we consider the problem of localization in Sensor Networks (SN). We propose a distributed solution based on a recently developed generalized descent algorithm. To cope with real-world applications, the algorithm we propose is suitable for an asynchronous implementation and is numerically robust to non ideal communications, i.e., packet-losses. Under mild assumptions, theoretical convergence of the algorithm is shown. The algorithm is compared with a recently developed ADMM-based algorithm for robust state estimation.