System and method for localization and tracking using GNSS location estimates, satellite SNR data and 3D maps

1. A method of determining location of a user device, the method comprising:

receiving global navigation satellite system (GNSS) fix data that represents a GNSS calculated position of the user device;

receiving signal strength data associated with each satellite communicating with the user device;

receiving map information regarding an environment surrounding the user device;

providing the received GNSS fix data, signal strength data, and a sampled particle set generated based on a previous particle set output to a particle filter, wherein the particle filter updates particle weights associated with the previous particle set output by fusing the GNSS fix data and the signal strength data, wherein the particle filter utilizes GNSS fix matching based on the GNSS fix data and probabilistic shadow matching estimates that represent a likelihood of received signal strength data as a function of hypothesized user device locations within the environment described by the received map information to update the particle weights and generate an output particle set estimate;

applying a motion model to the output particle set estimate, wherein the motion model generates a predicted particle set that for each particle location comprises a distribution of possible locations and a distribution of possible velocities in a future time step;

applying Rao-Blackwell sampling to the predicted particle set to generate a sampled particle set, wherein Rao-Blackwell sampling restricts the distribution of possible locations to a point mass, and wherein the sampled particle set is provided in feedback to the particle filter to be updated based on the received GNSS fix data and signal strength data; and

providing a corrected device location output based on the output particle set estimate generated by the particle filter.

2. The method of claim 1 , wherein the probabilistic shadow matching applies a signal-to-noise ratio (SNR) model to received signal strength data to determine a probability of whether the received signal is line-of-sight (LOS) or non-line-of-sight (NLOS), and further includes utilizing the map information to determine a probability of a signal received from each satellite being blocked, wherein the LOS/NLOS probability and blockage probability are combined to generate the probabilistic shadow matching estimate.

3. The method of claim 2 , wherein the map information includes a 3D occupancy map, wherein the blockage probability is calculated utilizing ray-tracing between hypothesized user device locations and each satellite.

4. The method of claim 2 , wherein the map information regarding the environment includes information regarding street locations and coarse building height statistics, wherein the blockage probability is calculated utilizing street assignments for each hypothesized user device location and coarse building height statistics.

5. The method of claim 2 , wherein the map information regarding the environment includes at least one of 2D maps, road network maps, statistical information on building heights, and a 2.5D map based on building footprints.

6. The method of claim 1 , wherein the non-linear filter utilizes a motion model to predict user device locations in a subsequent time step, wherein the predicted user device locations are provided in feedback to be fused with current GNSS fix data and signal strength data.

7. The method of claim 6 , further including:

generating a likelihood surface based on the GNSS position fix measurement and the predicted user device locations generated by the motion model, wherein the likelihood surface defines the hypothesized user device locations.

8. The method of claim 7 , wherein the likelihood surface is generated using kernelized estimates with kernel centers selected as an ellipse around the GNSS position fix and ellipses around predicted user device locations generated by the motion model.

9. The method of claim 1 , further comprising:

receiving a road network map that identifies the location of roads within an area surrounding the user;

computing road matching likelihoods for particles in the output particle set estimate based on the proximity of each particle to the location of roads identified in the road network map; and

multiplying the computed likelihoods computed for each particle onto the output particle set estimate to modify the weights associated with each particle to generate an updated output particle set estimate.

10. The method of claim 9 , further comprising:

assigning particles in the output particle set estimate to one of the roads identified in the road network map to ensure that all particles in the output particle set estimate are assigned to a road location.

11. The method of claim 2 , further comprising:

computing a probability that all received signals for each particle in the output particle set is NLOS; and

generating an output indicating whether the user is located indoors based on the computed probability that all received signals for each particle in the output particle set is NLOS.