Privacy-Preserving Filtering: Application to Traffic Estimation Jerome Le Ny École Polytechnique de Montréal Road traffic information systems rely on data streams provided by an increasingly large number and variety of sensors, such as loop detectors, cameras, or GPS. These data streams and even the aggregate information ultimately published by these systems contain potentially sensitive location information about private users. In this talk we discuss some recent developments in privacy-preserving data analysis, with a focus on the notion of differential privacy and the processing of dynamic data streams with formal privacy guarantees. We present techniques to design filters and dynamic estimators satisfying differential privacy constraints, and show how tools from control theory and signal processing can help with this task. In the context of traffic estimation, we discuss how macroscopic hydrodynamic models of the aggregated traffic can help limit the impact on estimation performance of the privacy-preserving mechanisms.