Address Localization Method Based on Data Fusion of Core Network and Radio Access Network

User address localization technologies have far-reaching implications for operators and governments. These technologies serve to optimize the planning and operation of communication networks and enhance the quality of network services, thereby providing powerful support for the provision of social public services, urban planning, and emergency management. The existing user address localization methods used by operators mainly rely on base stations locations and user-reported GPS. However, when the coverage area of a base station is too large, the localization accuracy will be poor even if triangulation techniques are used. Moreover, since the user-reported GPS is triggered by specific network events, its time continuity and volume are unstable. To address these limitations, this paper provides a user address localization method based on data fusion of signaling data from core network, measurement report (MR) data from radio access network, and operational data from carriers' business support system (BSS). The feature engineering is performed from three dimensions, including base station level analysis, GPS level analysis, and address level analysis, to construct a buildinglevel user address localization model. During data fusion, the signaling data compensates the weak time continuity of MR data, and the MR data compensates the poor localization accuracy of signaling data. In addition, through fusion with BSS data, it is able to convert users' GPS positions into building-level address locations, thereby further improving the practicality and accuracy of the model.