学术文献库

In this work, we investigate user equipment (UE) positioning assisted by deep learning (DL) in 5G and beyond networks. As compared to state of the art positioning algorithms used in today's networks, radio signal fingerprinting and machine learning (ML) assisted positioning requires smaller additional feedback overhead; and the positioning estimates are made directly inside the radio access network (RAN), thereby assisting in radio resource management. The conventional positioning algorithms will be used as back-up for the environments with high variability in conditions; but ML-assisted positioning serves as more efficient and simpler technique to provide better or similar positioning accuracy. In this regard, we study ML-assisted positioning methods and evaluate their performance using system level simulations for an outdoor scenario in Lincoln park Chicago. The study is based on the use of raytracing tools, a 3GPP 5G NR compliant system level simulator and DL framework to estimate positioning accuracy of the UE. The use of raytracing tool and system level simulator helps avoid expensive drive test measurements in practical scenarios. Our proposed mechanism is a first step towards more proactive mobility management in future networks. We evaluate and compare performance of various DL models and show mean positioning error in the range of 1-1.5m for the best DL configuration with appropriate system feature-modeling.