From Crowds to Coordinates: A user density-distance dynamic transformation method for Telegram users geolocalization

Abstract

Telegram is a widely used instant messaging service with over 900 million monthly active users. Research on the geolocalization of Telegram users helps verify the platform’s privacy protection and aids in cybercrime investigations. Existing location methods struggle with Telegram’s characteristics, such as large granularity changes in reported distances and irregular user turnover, making direct geolocalization challenging. This paper presents a new geolocalization method for Telegram users based on user density-distance dynamic transformation (U3DT). Unlike traditional methods, U3DT integrates distance breaks caused by reported distance changes and varying user densities to assess actual user distance, dynamically deploying probes for precise positioning. By analyzing changes in reported distances and user densities, we establish a relationship between discontinuity points and user densities. The geolocalization process determines the target’s real distance using dynamically set probes and corresponding user densities. Finally, we address the selection of datum points in trilateration based on discontinuity points. Experimental results on the Telegram platform show that U3DT achieves high geolocalization accuracy, with an average deviation of 223 meters and a maximum error of 450 meters. Compared to existing methods like RRABG, ETBG, and HNBG, U3DT reduces the average error by 60.4 % to 71.2 %.