Maximum-Likelihood Shadow-Matching

In order to improve the availability and accuracy of GNSS position fixes we propose a novel maximum- likelihood algorithm for using a data base of 3D building models in dense urban canyons to augment the GNSS receiver. We derive the algorithm, including how to incorporate errors in the 3D building model. We use a simulation study with multiple satellite constellations to illustrate the improved performance over prior shadow-matching algorithms (with and without errors in the 3D model). We demonstrate that the availability of accurate localization using ML shadow matching converges to 100% over time for stationary users.