Abstract | ||
---|---|---|
Direction of arrival (DOA) estimation of radio waves is demanded in many situations. In addition to MUSIC and ESPRIT, which are well-known traditional algorithms, compressed sensing has been recently applied to DOA estimation. If a large computational load as seen in some of compressed sensing algorithms is acceptable, it may be possible to apply deep learning to DOA estimation. In this paper, we propose estimating DOAs using deep learning and discuss training data preparation and designing for a specific scenario. The simulation results show reasonably-high estimation accuracy, performance dependency on training data preparation, and effectivity of specialized deep neural network. |
Year | DOI | Venue |
---|---|---|
2018 | 10.1109/WPNC.2018.8555814 | 2018 15th Workshop on Positioning, Navigation and Communications (WPNC) |
Keywords | Field | DocType |
compressed sensing algorithms,deep learning,DOA estimation,training data preparation,reasonably-high estimation accuracy,direction of arrival estimation,performance dependency | Training set,Radio wave,Computer science,Direction of arrival,Electronic engineering,Artificial intelligence,Deep learning,Artificial neural network,Computer engineering,Compressed sensing | Conference |
ISSN | ISBN | Citations |
2164-9758 | 978-1-5386-6437-7 | 0 |
PageRank | References | Authors |
0.34 | 3 | 6 |
Name | Order | Citations | PageRank |
---|---|---|---|
Yuya Kase | 1 | 0 | 0.34 |
Toshihiko Nishimura | 2 | 76 | 18.80 |
Takeo Ohgane | 3 | 116 | 31.39 |
Yasutaka Ogawa | 4 | 105 | 29.38 |
Daisuke Kitayama | 5 | 60 | 19.42 |
Yoshihisa Kishiyama | 6 | 1185 | 140.34 |