Underwater localization system based on visible-light communications using neural networks

Alzahraa M. Ghonim; Wessam M. Salama; Abd El-Rahman A. El-Fikky; Ashraf A. M. Khalaf; Hossam M. H. Shalaby; Hossam M. H. Shalaby

doi:10.1364/AO.419494

Applied Optics
Vol. 60,
Issue 13,
pp. 3977-3988
(2021)
•https://doi.org/10.1364/AO.419494

Underwater localization system based on visible-light communications using neural networks

Alzahraa M. Ghonim, Wessam M. Salama, Abd El-Rahman A. El-Fikky, Ashraf A. M. Khalaf, and Hossam M. H. Shalaby

Not Accessible

Your library or personal account may give you access

Get PDF
Email
Share
Get Citation
Copy Citation Text
Alzahraa M. Ghonim, Wessam M. Salama, Abd El-Rahman A. El-Fikky, Ashraf A. M. Khalaf, and Hossam M. H. Shalaby, "Underwater localization system based on visible-light communications using neural networks," Appl. Opt. 60, 3977-3988 (2021)

Export Citation
- BibTex
- Endnote (RIS)
- HTML
- Plain Text
Citation alert
Save article

Check for updates

Related Topics
Table of Contents Category
- Fiber Optics, Fiber Sensors, and Optical Communications
Optics & Photonics Topics
?

The topics in this list come from the Optics and Photonics Topics applied to this article.

About this Article
History
- Original Manuscript: January 13, 2021
- Revised Manuscript: March 27, 2021
- Manuscript Accepted: April 6, 2021
- Published: April 29, 2021

Abstract

Underwater localization using visible-light communications is proposed based on neural networks (NNs) estimation of received signal strength (RSS). Our proposed work compromises two steps: data collection and NN training. First, data are collected with the aid of Zemax OpticStudio Monte Carlo ray tracing software, where we configure 40,000 receivers in a $100\;{\rm m} \times 100\;{\rm m}$ area in order to measure the channel gain for each detector in seawater. The channel gains represent the input data set to the NN, while the output of the NN is the coordinates of each detector based on the RSS intensity technique. Next, an NN system is built and trained with the aid of Orange data mining software. Several trials for NN implementations are performed, and the best training algorithms, activation functions, and number of neurons are determined. In addition, several performance measures are considered in order to evaluate the robustness of the proposed network. Specifically, we evaluate the following parameters: classification accuracy (CA), area under the curve (AUC), training time, testing time, F1, precision, recall, logloss, and specificity. The corresponding measures are as follows: 99.1% for AUC and 98.7% for CA, F1, precision, and recall. Further, the performance results of logloss and specificity are 7.3% and 99.3% respectively.

Full Article | PDF Article

More Like This

Gaussian kernel-aided deep neural network equalizer utilized in underwater PAM8 visible light communication system

Nan Chi, Yiheng Zhao, Meng Shi, Peng Zou, and Xingyu Lu
Opt. Express 26(20) 26700-26712 (2018)

Experimental investigation of environmental interference mitigation and blocked LEDs using a memory-artificial neural network in 3D indoor visible light positioning systems

Zicai Cao, Mengfan Cheng, Qi Yang, Ming Tang, Deming Liu, and Lei Deng
Opt. Express 29(21) 33937-33953 (2021)

Two tributaries heterogeneous neural network based channel emulator for underwater visible light communication systems

Yiheng Zhao, Peng Zou, Weixiang Yu, and Nan Chi
Opt. Express 27(16) 22532-22541 (2019)

Previous Article Next Article

Data Availability

Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.

Cited By

You do not have subscription access to this journal. Cited by links are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

Figures (9)

You do not have subscription access to this journal. Figure files are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

Tables (15)

You do not have subscription access to this journal. Article tables are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

Equations (13)

You do not have subscription access to this journal. Equations are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

Transmitter specifications	Source: Rectangular array with 4 LEDs
	Cree white LED filtered in cyan color [8,15]
	Power: 1 W
	Viewing angle: $120^{\circ}$ [8]
	Number of rays per LED chip: 500,000 rays [8]
Receiver specifications	Total area: $100 m^{2}$
	Aperture area: $5 {c m}^{2}$ [8]
	Field of view: $180^{\circ}$ [8]
Type of water	Seawater

No. of Neurons	Training Time [s]	Test Time [s]	AUC	CA	F1	Precision	Recall	Logloss	Specificity
1	34.073	0.044	0.989	0.987	0.987	0.987	0.987	0.149	0.993
2	23.972	0.052	0.991	0.987	0.987	0.987	0.987	0.073	0.993
3	24.078	0.036	0.991	0.987	0.987	0.987	0.987	0.073	0.993
4	16.497	0.036	0.991	0.987	0.987	0.987	0.987	0.072	0.993
5	17.387	0.035	0.991	0.987	0.987	0.987	0.987	0.072	0.993
$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$
10	16.844	0.036	0.991	0.987	0.987	0.987	0.987	0.073	0.993

No. of Neurons	Training Time [s]	Test Time [s]	AUC	CA	F1	Precision	Recall	Logloss	Specificity
1	31.495	0.04	0.991	0.987	0.987	0.987	0.987	0.071	0.993
2	25.247	0.034	0.991	0.987	0.987	0.987	0.987	0.071	0.993
3	25.885	0.034	0.991	0.987	0.987	0.987	0.987	0.073	0.993
4	21.267	0.039	0.991	0.987	0.987	0.987	0.987	0.073	0.993
5	18.731	0.045	0.991	0.987	0.987	0.987	0.987	0.073	0.993
$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$
10	19.654	0.039	0.991	0.987	0.987	0.987	0.987	0.073	0.993

No. of Neurons	Training Time [s]	Test Time [s]	AUC	CA	F1	Precision	Recall	Logloss	Specificity
1	2.137	0.047	0.989	0.987	0.987	0.987	0.987	0.149	0.993
2	3.184	0.035	0.991	0.987	0.987	0.987	0.987	0.071	0.993
3	2.603	0.033	0.991	0.987	0.987	0.987	0.987	0.071	0.993
4	2.624	0.04	0.991	0.987	0.987	0.987	0.987	0.071	0.993
5	3.116	0.058	0.991	0.987	0.987	0.987	0.987	0.071	0.993
$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$
10	3.057	0.037	0.991	0.987	0.987	0.987	0.987	0.071	0.993

No. of Neurons	Training Time [s]	Test Time [s]	AUC	CA	F1	Precision	Recall	Logloss	Specificity
1	125.467	0.045	0.992	0.987	0.987	0.987	0.987	0.075	0.994
2	58.662	0.06	0.992	0.987	0.987	0.987	0.987	0.073	0.994
3	53.329	0.067	0.991	0.987	0.987	0.987	0.987	0.072	0.994
4	40.75	0.058	0.991	0.987	0.987	0.987	0.987	0.072	0.994
5	34.917	0.061	0.991	0.987	0.987	0.987	0.987	0.073	0.994
$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$
10	31.466	0.077	0.991	0.987	0.987	0.987	0.987	0.072	0.994

No. of Neurons	Training Time [s]	Test Time [s]	AUC	CA	F1	Precision	Recall	Logloss	Specificity
1	160.016	0.043	0.988	0.987	0.987	0.987	0.987	0.209	0.993
2	111.396	0.042	0.991	0.987	0.987	0.987	0.987	0.078	0.993
3	109.075	0.047	0.991	0.987	0.987	0.987	0.987	0.078	0.993
4	82.641	0.052	0.991	0.987	0.987	0.987	0.987	0.075	0.993
5	76.674	0.048	0.991	0.987	0.987	0.987	0.987	0.076	0.993
$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$
10	66.228	0.065	0.991	0.987	0.987	0.987	0.987	0.074	0.993

No. of Neurons	Training Time [s]	Test Time [s]	AUC	CA	F1	Precision	Recall	Logloss	Specificity
1	3.178	0.041	0.989	0.987	0.987	0.987	0.987	0.149	0.993
2	3.609	0.055	0.991	0.987	0.987	0.987	0.987	0.071	0.993
3	3.6	0.042	0.991	0.987	0.987	0.987	0.987	0.071	0.993
4	2.444	0.04	0.991	0.987	0.987	0.987	0.987	0.071	0.993
5	3.8	0.059	0.991	0.987	0.987	0.987	0.987	0.071	0.993
$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$
10	5.415	0.053	0.991	0.987	0.987	0.987	0.987	0.071	0.993

No. of Neurons	Training Time [s]	Test Time [s]	AUC	CA	F1	Precision	Recall	Logloss	Specificity
1	79.98	0.045	0.991	0.987	0.987	0.987	0.987	0.072	0.993
2	37.067	0.057	0.991	0.987	0.987	0.987	0.987	0.072	0.993
3	39.432	0.051	0.991	0.987	0.987	0.987	0.987	0.072	0.993
4	23.947	0.037	0.991	0.987	0.987	0.987	0.987	0.072	0.993
5	27.558	0.039	0.991	0.987	0.987	0.987	0.987	0.072	0.993
$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$
10	24.009	0.039	0.991	0.987	0.987	0.987	0.987	0.072	0.993

No. of Neurons	Training Time [s]	Test Time [s]	AUC	CA	F1	Precision	Recall	Logloss	Specificity
1	2.888	0.026	0.989	0.987	0.987	0.987	0.987	0.149	0.993
2	3.393	0.055	0.991	0.987	0.987	0.987	0.987	0.071	0.993
3	2.512	0.055	0.991	0.987	0.987	0.987	0.987	0.071	0.993
4	2.673	0.038	0.991	0.987	0.987	0.987	0.987	0.071	0.993
5	3.031	0.053	0.991	0.987	0.987	0.987	0.987	0.071	0.993
$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$
10	4.229	0.061	0.991	0.987	0.987	0.987	0.987	0.071	0.993

No. of Neurons	Training Time [s]	Test Time [s]	AUC	CA	F1	Precision	Recall	Logloss	Specificity
1	117.86	0.045	0.989	0.987	0.987	0.987	0.987	0.154	0.993
2	44.525	0.059	0.991	0.987	0.987	0.987	0.987	0.073	0.993
3	42.86	0.046	0.991	0.987	0.987	0.987	0.987	0.073	0.993
4	31.303	0.052	0.991	0.987	0.987	0.987	0.987	0.073	0.993
5	34.427	0.085	0.991	0.987	0.987	0.987	0.987	0.073	0.993
$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$
10	25.94	0.041	0.991	0.987	0.987	0.987	0.987	0.073	0.993

No. of Neurons	Training Time [s]	Test Time [s]	AUC	CA	F1	Precision	Recall	Logloss	Specificity
1	30.214	0.035	0.823	0.657	0.643	0.657	0.657	0.529	0.828
2	43.687	0.052	0.828	0.662	0.629	0.661	0.662	0.497	0.831
3	42.731	0.045	0.991	0.987	0.987	0.987	0.987	0.072	0.993
4	34.239	0.055	0.991	0.987	0.987	0.987	0.987	0.072	0.993
5	22.885	0.04	0.991	0.987	0.987	0.987	0.987	0.072	0.993
$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$
10	23.675	0.045	0.992	0.987	0.987	0.987	0.987	0.072	0.994

No. of Neurons	Training Time [s]	Test Time [s]	AUC	CA	F1	Precision	Recall	Logloss	Specificity
1	1.752	0.034	0.822	0.657	0.547	0.492	0.657	0.529	0.828
2	2.644	0.04	0.827	0.663	0.552	0.496	0.663	0.497	0.831
3	3.97	0.057	0.991	0.987	0.987	0.987	0.987	0.071	0.993
4	3.96	0.039	0.991	0.987	0.987	0.987	0.987	0.071	0.993
5	3.691	0.044	0.991	0.987	0.987	0.987	0.987	0.071	0.993
$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$
10	4.385	0.04	0.991	0.987	0.987	0.987	0.987	0.071	0.993

No. of Neurons	Training Time [s]	Test Time [s]	AUC	CA	F1	Precision	Recall	Logloss	Specificity
1	42.72	0.043	0.822	0.657	0.547	0.492	0.657	0.53	0.828
2	53.042	0.049	0.828	0.663	0.552	0.496	0.663	0.499	0.831
3	66.112	0.047	0.989	0.987	0.987	0.987	0.987	0.076	0.993
4	46.973	0.051	0.991	0.987	0.987	0.987	0.987	0.073	0.993
5	21.926	0.04	0.991	0.987	0.987	0.987	0.987	0.073	0.993
$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$	$\dots$
10	29.786	0.039	0.991	0.987	0.987	0.987	0.987	0.073	0.993

Reference	Training Algorithm	Percentage Accuracy
This work	Neural networks	98.7%
[12]	Enhanced J48 tree	98.15%
[13]	SCG	99.47%
[36]	ICT-Net	97.14%

Abstract

Data Availability

Cited By

Figures (9)

Tables (15)

Equations (13)

Applied Optics