콘텐츠 본문
논문 해외 국제일반학술지 Multi-output Convolutional Neural Network Based Distance and Velocity Estimation Technique for Orthogonal Frequency Division Multiplexing Radar Systems
- 학술지 구분 국제일반학술지
- 게재년월 2022-01
- 저자명 정의림공동(교신),최재웅
- 학술지명 Webology
- 발행처명 Webology center
- 발행국가 해외
- 논문언어 외국어
- 전체저자수 2
논문 초록 (Abstract)
The objective of this work is to propose a new method of estimating velocity and distance based on multi-output convolutional neural network (CNN) for orthogonal frequency division multiplexing (OFDM) radars. The two-dimensional (2D) periodogram is extracted from the received reflected waveforms through radar signal processing of received OFDM symbols. Conventionally, constant false alarm rate (CFAR) algorithm is used to estimate distance and velocity of targets. In contrast, this paper proposes a novel deep-learning based approach for the estimation of the targets in OFDM radar systems. The proposed multi-output CNN-based target detector estimates the distance and velocity of the target simultaneously. The proposed technique is verified through computer simulation. The results show that the proposed multi-output CNN-based method demonstrates more accurate distance and speed estimates than the conventional CFAR. Specifically, the distance and speed estimates of the proposed method are 9.8 and 12.3 times accurate, respectively, than those of the conventional CFAR.
