Modeling urban extent and density using radio artifacts from cellular and WiFi beacons

Abstract

Skyhook provides location to more than 100 million mobile devices, generating over a quarter billion location requests daily. This provides us with a large set of radio artifacts logged with high accuracy both temporally and spatially. The signal information returned with each of these location requests allows us to extend, refine and self heal the Skyhook beacon database across thousands of urban areas worldwide. Beacon density may be measured and discretized to produce accurate representations of underlying real world features. Since Wi-Fi beacons have become ubiquitous in all major cities worldwide, mapping their distribution provides a dependable signal for delineating urban boundaries. In addition to defining general urban polygons, features such as lakes, parks and other signal-starved areas may be captured as isopleths derived from beacon density raster surfaces. This is especially useful in parts of the world where high quality polygonal data is scarce or non-existent. And because of the ubiquitous sampling method, areas of new development tend to self-register nearly instantly. Quality of feature extraction can be assessed and validated against aerial imagery as well as existing geospatial layers. We find that Skyhook's radio artifact-derived features generally provide superior positional accuracy to existing geospatial resources from both public sources and commercial vendors for determining urban boundaries as well as sub-urban features.