PROJECT TITLE :

Context-aware Telco Outdoor Localization

ABSTRACT:

In recent years, we have seen a rapid expansion in the techniques used for telecommunication (Telco), from 2G to the soon-to-be-developed 5G. A precise outdoor localization is essential for the management, operation, and optimization of telecommunications networks by operators of telecommunications companies. In contrast to GPS, Telco localization is a method used by telecommunications service providers to locate mobile devices used in outdoor environments by making use of measurement report (MR) data. In situations where MR samples are provided that contain noisy signals (for example, as a result of Telco signal interference and attenuation), Telco localization frequently experiences high error rates. To achieve this goal, the primary focus of this paper is on how to improve the accuracy of Telco localization through the use of algorithms to detect and repair outlier positions that have high error rates. Specifically, we propose a context-aware Telco localization technique called RLoc. It is comprised of the following three primary components: a machine-learning-based localization algorithm, a detection algorithm to find flawed samples, and a repair algorithm to replace outlier localization results with better ones (ideally ground truth positions). We take into account the spatio-temporal locality of MR locations and exploit trajectory context in order to detect and repair flawed positions. This is in contrast to the majority of existing works, which attempt to detect and repair every flawed MR sample independently. Through the use of real MR data sets obtained from 2G GSM and 4G LTE Telco networks, we were able to demonstrate that our work, RLoc, can significantly improve the accuracy of Telco location services. For instance, RLoc can achieve 32.2 meters of median errors when applied to a large 4G MR data set, which is approximately 17.4 percent better than the state-of-the-art.