부산대학교 도서관

Location awareness is fundamental for several applications operating in fifth generation (5G) and beyond wireless networks. To provide accurate localization in 5G networks, time difference-of-arrival (TDOA) measurements are commonly employed. However, the quality of TDOA measurements significantly impacts the localization accuracy and heavily depends on the selection of the reference base station (RBS). Selecting the best RBS is particularly challenging in cluttered environments, such as Industrial Internet-of-Things (IIoT) environments, due to non-line-of-sight conditions and multipath propagation. This paper proposes a machine learning-based method for RBS selection, leveraging the rich information encapsulated in the received signals. The localization performance gain provided by the proposed approach is quantified in the 3rd Generation Partnership Project indoor factory scenario. Results show that the proposed RBS selection method provides a new level of location awareness for applications operating in 5G and beyond environments.