인문학
사회과학
자연과학
공학
의약학
농수해양학
예술체육학
복합학
지원사업
학술연구/단체지원/교육 등 연구자 활동을 지속하도록 DBpia가 지원하고 있어요.
커뮤니티
연구자들이 자신의 연구와 전문성을 널리 알리고, 새로운 협력의 기회를 만들 수 있는 네트워킹 공간이에요.
논문 기본 정보
- 자료유형
- 학술저널
- 저자정보
- 발행연도
- 2020.12
- 수록면
- 1,116 - 1,121 (6page)
- DOI
- 10.5302/J.ICROS.2020.20.0151
이용수
초록· 키워드
Indoor positioning systems are attracting increasing attention due to the demand for accurate location awareness in areas where the Global Navigation Satellite System (GNSS) does not work. The Wi-Fi access points (APs) built in to many constructions can be used to develop a Wi-Fi-fingerprint-based indoor localization method. However, such a localization method needs large amounts of fingerprint data samples to achieve superior positioning performance, which increases the costs of data collection and calibration. To reduce these resource requirements, in this paper, we propose a new semi-supervised generative adversarial network (GAN) approach that can learn an accurate positioning model by using only a small number of training samples. The proposed semi-supervised GAN is extended from a general unsupervised GAN in a way to generate fake labeled samples and to involve a classification model for realizing localization without employing an additional positioning mechanism. Based on the results of indoor experiments conducted in multi-story buildings, the proposed method outperformed a supervised deep-learning-based localization method when room-size landmark positioning was conducted.
#Indoor positioning system
#Wi-Fi RSSI (received signal strength indicator)
#GAN (generative adversarial network)
#semi-supervised learning
#deep learning
상세정보 수정요청해당 페이지 내 제목·저자·목차·페이지정보가 잘못된 경우 알려주세요!
목차
- Abstract
- Ⅰ. 서론
- II. 문제 정의
- Ⅲ. 준 지도식 GAN
- Ⅳ. 실험 결과
- V. 결론 및 향후 연구계획
- REFERENCES
참고문헌
참고문헌 신청최근 본 자료
UCI(KEPA) : I410-ECN-0101-2021-003-000053053