UWB 활용 DNN 기반 차량용 키 측위 알고리즘 (DNN-Based Car Key Localization Algorithm with UWB)

서지정보

· 발행기관 : 한국전자파학회
· 수록지 정보 : 한국전자파학회 논문지 / 35권 / 12호 / 985 ~ 989페이지
· 저자명 : 박수아, 박현우, 차경주, 오영균, 김선우

초록

최근다양한실내위치기반의서비스수요가증가함에따라무선통신기반측위연구가활발히진행중이다. 하지만, 이기술은장치간의LOS(line of sight) 경로를 차단하는 장애물에 의하여 성능이 저하되는 문제가 있다. 본 논문에서는UWB(ultra-wideband) 기반 차량용 측위 서비스의 성능을 향상시키기 위하여 딥러닝을 사용하는 알고리즘을제안한다. 차량환경을모사한환경에서실험을진행하여수집된UWB 신호세기및거리를DNN(deep neural network) 기반 딥러닝학습을통해모델을구축하고, 기존알고리즘과성능을비교분석한다. 통신경로내장애물이존재하는환경의시나리오에서실험을진행한결과, 기존알고리즘대비딥러닝기반알고리즘은향상된측위성능을보여준다. 이를통하여위치별신호 패턴의특징을중심으로학습하는딥러닝을적용하여기존기법대비측위정확도향상이가능함을확인했다.

영어초록

Recently, wireless communication-based localization has been extensively studied due to the increasing demand for various indoor location-based services. However, this technology experiences performance degradation due to obstacles that block the line-of-sight (LOS) path between devices. This study introduces a machine learning algorithm designed to enhance the performance of ultra-wideband (UWB)-based vehicle key localization services. Through experiments conducted in simulated vehicle environments using the collected UWB signal strengths, a deep neural network (DNN)-based deep learning model was developed, and its performance was evaluated against existing algorithms. In scenarios with obstacles in the communication path, the machine-learning-based algorithm demonstrated superior estimation accuracy compared to conventional algorithms. Thus, it was confirmed that machine learning, which focuses on rec ognizing the features of location-specific signal patterns, can significantly improve localization accuracy over traditional methods.

참고자료

· 없음