BSLoc: Base Station ID-Based Telco Outdoor Localization

Telecommunication (Telco) localization is an important complementary technique of Global Position System (GPS). Traditional Telco localization approaches requires radio signal strength indicator (RSSI) of mobile devices with the connected base stations (BSs). Unfortunately, many of real-world signal measurement could miss RSSI values, and Telco operators typically will not record RSSI information, e.g., due to the major departure from current operational practices of Telco operators [6]. To address this problem, we design a novel BS ID-based coarse-to-fine Telco localization model, namely BSLoc, which requires only the connected BS IDs, time and speed information of mobile devices. BSLoc consists of two layers: (1) a sequence localization model via Hidden Markov Model (HMM) to localize the mobile devices with coarse-grained locations, and (2) a machine learning regression model with engineered features to acquire the fine-grained locations of mobile devices. Our experiments verify that, on a 2G dataset, BSLoc achieves a median error 26.0 m, which is almost comparable with two state-of-art RSSI-based techniques [9] 17.0 m and [20] 20.3 m.