Street-level Landmark Mining Algorithm Based on Radar Search

Abstract

Street-level landmarks are the important foundation for high-precision IP geolocation, which is of significant value in network security. As online map-based landmark mining algorithms are constrained by the online map service itself, a street-level landmark mining algorithm based on radar search is proposed in this study. Initially, the region in which landmarks will be mined is divided into smaller sub-regions. Then, the radar search service of an online map is used to perform a recursive query request for each sub-region, and street-level candidate landmarks in the sub-region are obtained. Finally, IP geolocation databases and the street-level geolocation algorithm are used to evaluate the landmarks and retain reliable ones. Landmark mining experiments of two groups were conducted to verify the algorithm. Experimental results show that the number of obtained reliable landmarks of the proposed method increases by 4.1 times, the landmark coverage area increases by 59% and the average geolocation error is reduced from 9.94 km to 4.33 km compared with the existing online maps-based method, and the proposed algorithm can also obtain more candidate landmarks as well as reliable landmarks and got lower mean error of geolocation results than the state-of-art landmark mining algorithms based on other web resources.