Article
Recovery of Ionospheric Signals Using Fully Convolutional
DenseNet and Its Challenges
Merlin M. Mendoza
1
, Yu-Chi Chang
1
, Alexei V. Dmitriev
1,2,
* , Chia-Hsien Lin
1,
, Lung-Chih Tsai
3
,
Yung-Hui Li
4
, Mon-Chai Hsieh
1
, Hao-Wei Hsu
1
, Guan-Han Huang
1
, Yu-Ciang Lin
1
and Enkhtuya Tsogtbaatar
5
Citation: Mendoza, M.M.; Chang,
Y.-C.; Dmitriev, A.V.; Lin, C.-H.; Tsai,
L.-C.; Li, Y.-H.; Hsieh, M.-C.; Hsu,
H.-W.; Huang, G.-H.; Lin, Y.-C.; et al.
Recovery of Ionospheric Signals
Using Fully Convolutional DenseNet
and Its Challenges. Sensors 2021, 21,
6482. https://doi.org/10.3390/
s21196482
Academic Editor: Nunzio Cennamo
Received: 8 August 2021
Accepted: 24 September 2021
Published: 28 September 2021
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2021 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
1
Department of Space Science and Engineering, National Central University, Taoyuan City 320317, Taiwan;
mmmendoza@g.ncu.edu.tw (M.M.M.); jason0010125@g.ncu.edu.tw (Y.-C.C.);
chlin@jupiter.ss.ncu.edu.tw (C.-H.L.); davidhsieh@g.ncu.edu.tw (M.-C.H.);
willy108623016@g.ncu.edu.tw (H.-W.H.); enter468@g.ncu.edu.tw (G.-H.H.);
alterjohnnylife@g.ncu.edu.tw (Y.-C.L.)
2
Skobeltsyn Institute of Nuclear Physics, Lomonosov Moscow State University, 119899 Moscow, Russia
3
Center for Space and Remote Sensing Research, National Central University, Taoyuan City 320317, Taiwan;
lctsai@csrsr.ncu.edu.tw
4
AI Research Center, Hon Hai Research Institute, Taipei 114699, Taiwan; yunghui.li@foxconn.com
5
Department of Computer Science and Information Engineering, National Central University,
Taoyuan City 320317, Taiwan; enkhtuya@g.ncu.edu.tw
* Correspondence: dalex@jupiter.ss.ncu.edu.tw
Abstract:
The technique of active ionospheric sounding by ionosondes requires sophisticated meth-
ods for the recovery of experimental data on ionograms. In this work, we applied an advanced
algorithm of deep learning for the identification and classification of signals from different iono-
spheric layers. We collected a dataset of 6131 manually labeled ionograms acquired from low-latitude
ionosondes in Taiwan. In the ionograms, we distinguished 11 different classes of the signals according
to their ionospheric layers. We developed an artificial neural network, FC-DenseNet24, based on
the FC-DenseNet convolutional neural network. We also developed a double-filtering algorithm
to reduce incorrectly classified signals. That made it possible to successfully recover the sporadic
E layer and the F2 layer from highly noise-contaminated ionograms whose mean signal-to-noise
ratio was low, SNR = 1.43. The Intersection over Union (IoU) of the recovery of these two signal
classes was greater than 0.6, which was higher than the previous models reported. We also identified
three factors that can lower the recovery accuracy: (1) smaller statistics of samples; (2) mixing and
overlapping of different signals; (3) the compact shape of signals.
Keywords:
ionospheric sounding; space weather; artificial intelligence; fully convolutional DenseNet
1. Introduction
The task of monitoring the ionosphere above the Taiwan–Korea region is important
for the control and prediction of the conditions for the propagation and distortions of
signals from the Global Navigation Satellite Signal (GNSS) constellation. The low-latitude
ionosphere is highly variable due to the daily, seasonal, and solar cycle variations, as well
as due to the impacts of the external and internal drivers. These external drivers are
related to space weather effects such as solar flares [
1
], geomagnetic storms [
2
], and particle
precipitations from the radiation belts [
3
]. The internal sources are related to tidal effects [
4
],
earthquakes [
5
], and tsunamis [
6
]. Operative identification of the ionospheric disturbances
of different origins is required for the development of nowcast and forecast models of
the regional ionosphere. The wide variety of disturbances and the complex behaviors of
the ionosphere require the development of advanced methods of ionospheric observation
and diagnostics.
Sensors 2021, 21, 6482. https://doi.org/10.3390/s21196482 https://www.mdpi.com/journal/sensors
