Abstract: Architecture that obtains and utilizes collections of geographically-tagged data to discover optimal vantage points for viewsheds of entities of interest such as physical entities and conceptual entities such as landmarks, sunset, skyline, etc. The disclosed architecture discloses the utilization of at least geo-tagged image data to discover relationships between a combination of concrete entities and/or abstract concepts, and techniques for surfacing such relationships to users. The data can be crowd-sourced geo-tagged image data that are mined from social content and which can be observed or experienced from a certain location/area.

Abstract: An RF fingerprinting methodology is generalized to include non-RF related factors. For each fingerprinted tile, there is an associated distance function between two fingerprints (the training fingerprint and the test fingerprint) from within that tile which may be a linear or non-linear combination of the deltas between multiple factors of the two fingerprints. The distance function for each tile is derived from a training dataset corresponding to that specific tile, and optimized to minimize the total difference between real distances and predicted distances. Upon receipt of an inference request, a result is derived from a combination of the fingerprints from the training dataset having the least distance per application of the distance function. Likely error for the tile is also determined to ascertain whether to rely on other location methods.

Abstract: An RF fingerprinting methodology is generalized to include non-RF related factors. For each fingerprinted tile, there is an associated distance function between two fingerprints (the training fingerprint and the test fingerprint) from within that tile which may be a linear or non-linear combination of the deltas between multiple factors of the two fingerprints. The distance function for each tile is derived from a training dataset corresponding to that specific tile, and optimized to minimize the total difference between real distances and predicted distances. Upon receipt of an inference request, a result is derived from a combination of the fingerprints from the training dataset having the least distance per application of the distance function. Likely error for the tile is also determined to ascertain whether to rely on other location methods.

Abstract: An RF fingerprinting methodology is generalized to include non-RF related factors. For each fingerprinted tile, there is an associated distance function between two fingerprints (the training fingerprint and the test fingerprint) from within that tile which may be a linear or non-linear combination of the deltas between multiple factors of the two fingerprints. The distance function for each tile is derived from a training dataset corresponding to that specific tile, and optimized to minimize the total difference between real distances and predicted distances. Upon receipt of an inference request, a result is derived from a combination of the fingerprints from the training dataset having the least distance per application of the distance function. Likely error for the tile is also determined to ascertain whether to rely on other location methods.

Abstract: Assessing the accuracy of location estimation systems. A mobile computing device provides location information including a device location (e.g., via GPS) and one or more wireless network beacons accessible by the computing device at the device location. The wireless network beacons accessible by the computing device are compared to stored post information including a plurality of beacon lists. An estimated device location is determined based on the comparison. The estimated device location is compared to the known device location. A difference between the estimated device location and the received device location is determined based on the comparison. An analysis of the determined difference is performed to generate accuracy maps and other insight into the relationship between accuracy and geographic area for the location estimation systems.

Abstract: An RF fingerprinting methodology is generalized to include non-RF related factors. For each fingerprinted tile, there is an associated distance function between two fingerprints (the training fingerprint and the test fingerprint) from within that tile which may be a linear or non-linear combination of the deltas between multiple factors of the two fingerprints. The distance function for each tile is derived from a training dataset corresponding to that specific tile, and optimized to minimize the total difference between real distances and predicted distances. Upon receipt of an inference request, a result is derived from a combination of the fingerprints from the training dataset having the least distance per application of the distance function. Likely error for the tile is also determined to ascertain whether to rely on other location methods.

Abstract: Estimating positions of beacons based on spatial relationships among neighboring beacons. Beacon reference data defining positions of beacons is stored from beacon fingerprints observed by devices (e.g., enabled with global positioning system receivers). For a received beacon fingerprint having at least one beacon for which the beacon reference data is missing (e.g., from a device without a GPS receiver), beacons in the received beacon fingerprint for which beacon reference data is available are identified. Based on these identified beacons, the missing beacon reference data is calculated. In some embodiments, a set of spatially diverse beacons is selected from the identified beacons prior to calculating the beacon reference data.

Abstract: Estimating positions of beacons based on spatial relationships among neighboring beacons. Beacon reference data defining positions of beacons is stored from beacon fingerprints observed by devices (e.g., enabled with global positioning system receivers). For a received beacon fingerprint having at least one beacon for which the beacon reference data is missing (e.g., from a device without a GPS receiver), beacons in the received beacon fingerprint for which beacon reference data is available are identified. Based on these identified beacons, the missing beacon reference data is calculated. In some embodiments, a set of spatially diverse beacons is selected from the identified beacons prior to calculating the beacon reference data.

Abstract: A middleware system is provided that is situated between the user applications and the various content databases that are to be searched in order to simplify the creation of user applications for mobile devices that use location-based services that employ ontology-based search systems. The middleware system exposes one or more services to the user application. For example, a service exposes a service that allows the user to annotate and/or tag known semantic locations. As another example, a service provides a list of suggested semantic POIs to user applications in response to user queries. The suggested semantic POIs are selected based on a user's location and possibly context-dependent information. The suggested semantic POIs also may be based on user-dependent information obtained from a user-profile or the like and the suggested semantic locations that are provided to the user applications may be ranked and presented in an order beginning with those semantic locations that may be of greatest interest.

Abstract: Assessing the accuracy of location estimation systems. A mobile computing device provides location information including a device location (e.g., via GPS) and one or more wireless network beacons accessible by the computing device at the device location. The wireless network beacons accessible by the computing device are compared to stored post information including a plurality of beacon lists. An estimated device location is determined based on the comparison. The estimated device location is compared to the known device location. A difference between the estimated device location and the received device location is determined based on the comparison. An analysis of the determined difference is performed to generate accuracy maps and other insight into the relationship between accuracy and geographic area for the location estimation systems.