Solution for estimating position from data collected in mobile communication network
Deep Neural Network Learning
Estimate the location of a terminal using a deep neural network model trained on data preprocessed with a pair of features and labels |
Using RF Data
If location information is missing or inaccurate, return location information based on learning by inputting RF information |
Use for various businesses
Based on accurate user location information, it can be used for various businesses such as regional service quality and commercial area analysis |
Learning Data
Smartphones RF characteristics and location information data |
Learning Type
Supervised learning |
Development/Utilization Model
DNN |
Technology Application/API
TensorFlow, Keras |
Technology applied to
CLAIR, Emergency Rescue Location Services |
MDT data-based learning model
* Dataset: MDT data, Data within a 1km2 radius of downtown
* Feature: ECGI, RSRQ, N1_PCI, N1_RSRP, N2_PCI, N2_RSRP
| Training Data | 355,252 |
| Test Data | 39,473 |
| Average differences | 50m |
| CEP 90 | 114m |
