Landmarks Screening Method Based on Elliptical Trajectories Constraint

Network landmarks are reference points for geolocating target IP addresses. The consistency of network characteristics between landmarks and target IP significantly impacts geolocation results. However, existing landmark evaluation methods do not screen landmarks based on the consistency of network characteristics between landmarks and target IP, which can lead to increased geolocation errors. Therefore, we propose a landmarks screening method based on elliptical trajectory constraints in this paper. The method constructs elliptical trajectories with the geographical location of landmarks as the focus and the network distance between landmarks as the major axis. The ellipse’s area represents the path’s localization degree between landmarks, and the ellipse’s eccentricity means the conversion relationship between the network distance and geographical distance between landmarks. We aggregated landmarks with a high localization degree and consistent conversion relationship between network and geographical distance into a training set. We filtered out landmark training sets compatible with the target IP’s network characteristics to geolocate the target IP. Experiments based on four geolocation methods on three different landmark sets show that the proposed method can effectively screen landmarks based on their network characteristics. During geolocation, the geolocation error can be reduced by up to 75.5% for targets that meet the conditions.