Open Access
Add to BibSonomy
Abstract
Underwater geodesy is important for marine studies and seafloor mapping. These studies typically make use of the time of flight, for example SONAR, and collect a lot of data. Furthermore, high map densities require big data collection and transportation, and therefore require high bandwidth underwater networks. In this article, we propose and demonstrate an underwater optical network based on a directly-modulated laser at 650 nm that enables the deployment of underwater robotic systems which are capable of transferring the captured data to a base station and allow the synchronization of the clocking signals. As proof of concept, we demonstrated a unipolar NRZ data transfer from an Arduino at 2 Mbps through an underwater channel measuring about 1 meter in length. A bit error rate value of about 10−5 for underwater data transmission was realised by the designed optical network, showcasing its potential for use in underwater data transfer during robotic geodesy surveys. Recovery of the clock signal from the signal generator at 2 kHz was also shown. Phase noise floor values below −90 dBc/Hz were attained for the underwater clock signal transmission.
© 2023 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
In the last couple of years, different research communities have shown a great interest in the aquatic ecosystem for numerous applications. While many researchers are interested in the water environment for use as a wireless telecommunication channel for data communication, others may be interested in the aquatic ecosystem for studies such as oceanic climatic change, oceanic ecosystems, underwater marine surveillance [1], and seafloor geodetic studies for measurements of crustal deformation [2]. Geodesic studies are typically interested in the physical deformation of the seafloor, while at the same time making use of the aquatic environment as a wireless transmission channel of the captured data.
Current optical seafloor geodetic studies for measurements of crustal deformation and seafloor mapping use interferometry to extract the distance from ground to surface. In [3], underwater distance measurement was achieved by using a frequency comb laser at 518 nm (centre frequency) [3]. In this technique, the beam is guided into the water. A mirror is placed at the ground surface to reflect the beam back to the surface. In the time domain, a cross correlation technique is applied to determine the underground distance. Although accurate, this system is generally expensive to implement especially for large scale measurements. In this study, we propose a geodesic network which consist of robots deployed underwater as illustrated in Fig. 1. These underwater robots could be installed with sound navigation and ranging (SONAR) systems as well as clocking devices for seafloor mapping and time stamping of data, respectively. Clocking devices can also be used for sampling of the analog SONAR signals. This therefore implies that the underwater robots will require digital memories installed on them to store the captured geodesic data. The underwater robots can routinely navigate to the base station to upload the captured data and to synchronize their clocking devices. This will therefore require an underwater network for both clock and data transmission. Phase and/or amplitude of the clock signal could be deteriorated while the robots are carrying out their activities over a time period before returning to the base station. Clock synchronization is therefore important to restabilize the phase and/or amplitude of the clock signals.
There are different underwater communication technologies which can be used to transfer both the captured geodesic data and the clock signal for synchronization. These technologies include underwater acoustic waves [4], underwater radio frequency (RF) waves [5], and underwater optical waves [6]. Underwater acoustic wireless communication (UAWC) has attracted significant interest. This is attributed to their increased network transmission reach. However, acoustic underwater communication still has several limitations including scattering, increased delay due to low propagation speeds, high attenuation, and low bandwidth. Therefore, researchers have investigated the use of low- and high-frequency RF waves to realize underwater wireless networks [5]. RF frequencies such as microwaves and millimeter wave signals suffer considerable attenuation in water. For example, the attenuation of microwave signals in the ocean is about 169 dB/m for the Wi-Fi bandwidth of 2.4 GHz while the attenuation in freshwater is even higher at 189 dB/m [4]. Moreover, microwave-based underwater wireless communication will require high gain directional antennas which are expensive to manufacture. Cheap-scale RF antennas have been designed and is currently finding application in modern devices such smartphones, drones, and laptops. Their small size could allow them to be fitted on small underwater robots, making them suitable for this application. However, the main shortcoming and probably the most limiting fact with the RF technologies, is that they are limited in term of bandwidth and frequency. When large underwater map densities need to be surveyed, using RF technology on underwater robotic systems could limit the maximum data rate that can be achieved.
Apart from the acoustic and RF technologies, an alternative solution is to use optical waves which can provide high-speed underwater wireless communications. Underwater propagation of optical waves has different characteristics at different wavelengths. Research on underwater optical wireless communication (UOWC) mainly focuses on improving the transmission range and the data rate of such networks for different water types. In [6], it was found that attenuation within the range of 450–550 nm wavelengths (blue and green lights) is much smaller compared to other wavelengths in the visible spectrum. As a result, most reported UOWC networks are experimented at the 450-550 nm wavelength range. A study in [7] demonstrated that the transmission property of red light at 650 nm in water in terms of extinction coefficient and channel bandwidth via Monte Carlo simulation outperforms blue-green light in highly turbid water. For the application proposed in this article, their conclusion motivated us to investigate our study at the 650 nm wavelength. This is because, as a lot of robots are expected to be deployed underwater in order to achieve geodesic studies over larger areas, the water channel could become turbulent and will therefore require systems with optimal operation under such channel conditions.
Although there has been a lot of reports on underwater optical networks, most of these used bulk and high-cost optical transceivers. Additionally, the reported underwater optical wireless networks, although demonstrating high-speed data transmission rates, could be difficult to be adopted for underwater robotic systems for geodesic studies as they can make the overall robots bulky and costly. To reduce the size and cost of these underwater robotic systems, cheap and small-scale optical transceivers are needed. Additionally, there is a novel need to study the clock transmission performance of underwater optical wireless systems. Therefore, the primary focus of this study is to demonstrate a cheap and compact underwater robotic system for marine geodesic studies which require both clock and data transmission. UOWC networks at high data rates have been studied in different wavelengths of light. Regardless of the channel wavelength, laser diodes are manly used to achieve longer transmission distance and higher data rates [7–9] compared to using light emitting diodes (LEDs) as light sources. In this study, we demonstrated a simple, compact, and low-cost underwater robotic system using readily-available red laser source, pattern generators, and optical detector.
2. Theoretical optical propagation in aquatic medium
Optical light propagating in a water channel is primarily affected by absorption and scattering. Generally, the spectral beam attenuation coefficient is used to describe light propagation in an aquatic channel under low scattering regimes. The attenuation coefficient, $c(\lambda )$, is a wavelength-dependent parameter. The attenuation coefficient is a linear combination of the scattering, $a(\lambda )$ and the absorption coefficients, $b(\lambda )$ as shown in Eq. (1) [10].
Numerous factors contribute to the values of $a(\lambda )$ and $b(\lambda )$. These factors include water molecules, bubble content of the water channel, presence of organic substances such as sodium chloride in sea water, and colouring content present in the water [11]. In addition to the wavelength dependency of the attenuation coefficient, $a(\lambda )$ and $b(\lambda )$ also vary with water type. Usually, for simplicity, different values of chlorophyll concentration, C, are used to distinguish between different water types [12]. As a result, $a(\lambda )$ and $b(\lambda )$ are expressed as functions of wavelength, λ, and the concentration, C, as [13]
In Eq. (2), ${a_w}$ is the absorption coefficient of pure water and ${a_c}$ is a statistically-derived, non-dimensional number, denoting the chlorophyll concentration. The effects of absorption and scattering is such that light is lost and deviated, respectively, as it passes through the water channel.
3. Experimental setup
Figure 2 shows the experimental setup. A compact ACCULASE-LC-650-5-S red laser source was used for the experiment. The laser emits at 650 nm. The maximum output power from the laser was 5 mW. In order to attain this power, i.e., for maximum transmission through the medium, the laser was biased at 5 V using an Agilent E3646A dual source power supply. During initial investigations, a 2-kHz sinusoidal wave generated by a Beckman Industrial low-frequency function generator was used as the clock signal. In a practical geodesic application such as that using an underwater robot to measure depth from the robot to the seafloor, this clock signal will be used to determine the distance between the robot and the seafloor. For example, if we consider the speed of sound in water to be 1480 m/s, this clock frequency implies a distance measuring resolution of about 74 cm.
In the second study, a unipolar non-return to zero (NRZ) pseudo random binary sequence (PRBS) generated by an Arduino Uno was used. The unipolar NRZ PRBS had a bit period of about 0.5 milliseconds, implying an underwater upload speed of approximately 2 Mbps between the robot and the base station, which is partially submerged above the water channel as shown in Fig. 1. The low-speed data rate of 2 Mbps was limited by the used Arduino hardware. The Arduino was chosen considering the application of this network to serve as a compact pulse pattern generator for digital to analog conversion (DAC) of the digital data stored on the robots. Since most robotic systems are preferred to be small in size, the choice of an Arduino as a DAC is suitable for this application. When the size of the robot can be compromised, a high-speed but large size pulse pattern generator can be used.
For both clock and PRBS signals, direct modulation of the laser was implemented. A water tank measuring $0.7 \times 0.1 \times 0.1$, i.e., length, width, and height, respectively, in meters was used to emulate an aquatic channel. The water tank had a focus lens installed at its input side for optimum coupling of the modulated laser light into the water channel. At the output, the received modulated optical signal was detected using a Hawkeye-LD-488-1386 laser detector. The laser detector was also biased at 5 V. The laser detector has two output pins. The first output supports signal frequencies from DC to 1 kHz [14]. For this case, it was used to receive the PRBS signal. The second output port can support signal frequencies between 1 kHz and 750 kHz, and it was used to receive the 2 kHz clock [14]. For this port, the output is DC coupled with a 2.5 V offset about which the output is centred [14]. The received signals were measured by a Tektronix MSO 54 oscilloscope for offline analysis. Furthermore, a Rohde & Schwarz R&S FSV3000 spectrum analyser was used to quantify the short-term stability of the received clock signal by means of phase noise measurements.
4. Results and discussions
The electrical clock and PRBS signals used in this experiment are given in Figs. 3 and 4, respectively. The used laser required a signal of 1 V peak-to-peak (Vpp) maximum. The input clock and PRBS signals used were therefore of approximately 1 Vpp as shown in both Figs. 3 and 4. During the first part of the experiment, we investigated the effects of electrical to optical conversion and the free space transmission.
4.1. Effects of electrical to optical conversion and the free space transmission
To determine the effect of electrical to optical conversion and the free space transmission channel on the modulating electrical clock signal, we compare the modulating electrical clock signal with the received optical clock signal after free space transmission in term of voltage, frequency, and phase noise performance. Since the free space transmission performance was analysed against the electrical signal which is at 1 Vpp as shown in Fig. 3, the laser was biased at around 1.2 V so that the received free space signal at the photodetector was also around 1 Vpp (comparable to the Vpp of the electrical signal) for accurate analysis. When both signals have approximately equal amplitudes, any difference in these signals in term of phase noise and spectral purity can be attributed to the free space channel or the used network devices. This also gave a way of characterizing the used optical devices. It is important to note that the free space distance between the laser and the receiver was about 0.7 meters, the length of the used water tank. The results for this analysis are given in Figs. 5, 6, and 7. In the spectral domain of Fig. 6, the noise floor of the signal after free space transmission is comparable to that of the electrical signal. At the nominal frequency of 2 kHz, the signal powers are also comparable at about −42 dBm. This shows that free space transmission did not affect the transmitted electrical signal as far as its centre frequency and noise floor are concerned. However, the effect due to free space transmission and electrical to optical conversion can be seen in Fig. 5 (amplitude noise) despite both signals having comparable amplitude values (1 Vpp for the modulating electrical clock signal and 1.1 Vpp for the recovered signal after free space transmission). The phase noise performance of the two signals as given in Fig. 7 also reveals that electrical to optical conversion and free space transmission introduced some phase noise on the modulating electrical clock. This noise is visible as jitter in the time domain signal of Fig. 5. For example, at the offset frequency of about 650 Hz, the received free space clock had a phase noise value of about −97 dBc/Hz compared to the modulating electrical signal whose phase noise was about −105 dBc/Hz. Moreover, the phase noise floor of the received signal after free space transmission is high compared to the modulating electrical clock signal, i.e., −95 dBc/Hz for free space clock and −100 dBc/Hz for the modulating electrical clock signal. Overall, the effects due to free space transmission was not very severe as can clearly be seen on both Figs. 5 to 7. This means that this optical network can be applied for long distance free space transfer of clocking signals.
4.2 Free Space vs Underwater Clock Transmission
Next, we investigated the underwater clock transmission performance of our proposed simple and low-cost optical network which can find application in immerging underwater robotic systems for geodesic studies. For this investigation, the transmitter and receiver parameters were kept constant. Most importantly, the laser bias voltage was kept at 5 volts, and only the transmitting channel was changed from free space to underwater. The underwater clock transmission performance was determined by comparing its amplitude, frequency, and phase noise characteristics to that of the received clock signals after free space transmission. The free space and underwater clock transmission results are given in Figs. 8–10. As can be seen in all figures, and as expected, free space transmission had better performance compared to the recovered clock signal after underwater transmission. In Fig. 8 for example, the aquatic channel introduced amplitude noise to the recovered clock. Also visible in Fig. 8 is the reduction in the clock’s Vpp value after aquatic transmission. Underwater transmission reduced the amplitude to approximately 0.8 Vpp compared with a Vpp of about 4 V for the recovered optical clock signal after free space transmission. We notice in Fig. 9 however that, the two clock signals had the same spectral profile apart from a power reduction at the nominal frequency. This means that the effects of underwater transmission were only to reduce the power of the clock but did not affect its spectral composition. This is important for this application as the linear reduction in power can be recovered by using electrical amplifiers after the photodiode. Also noticeable in Fig. 9 is that the noise floor of the clock signal after underwater transmission remained comparable to that after free space transmission. Another clear effect of underwater transmission is visible in Fig. 10. The recovered underwater clock had high phase noise relative to the received clock after free space transmission as shown in Fig. 10. For example, at an offset frequency of about 650 Hz, the free-space channel had a phase noise value of about −110 dBc/Hz compared to the underwater case which had a phase noise value of about −95 dBc/Hz. This represents a 15-dB underwater transmission penalty, a difference that cannot be ignored.
Nevertheless, the phase noise performance of the received clock signal after underwater transmission is sufficient to be used as a timing source for robotic-based underwater geodesy application. The clock signal can be used by the base station so that it can be synchronised to allow its use by the deployed underwater robots for measuring seafloor depth.
4.3 Free space vs Underwater Data Transmission
During the final part of the experiment, we investigated the underwater unipolar NRZ PRBS data transmission performance of the network. As it was the case with clock signal transmission, the PRBS data was initially transmitted through free space before underwater. First, the free space transmission performance of the system was determined qualitatively by plotting the bit patterns and the eye diagrams of the recovered data. For this part, we reduced the laser bias voltage to about 1 V. This is because, at 5 V bias voltage, the signal after free space transmission was high at about 4 Vpp and it was difficult to trace any channel effects on the signal. We also chose to show the qualitative results for free space transmission at 1 V biasing because it was at about the same bias voltage where a minimum BER of 10−9 was achieved for free space transmission. This can be seen on the BER curve for free space transmission on Fig. 15. Figures 11 and 12 show the recovered free space data pattern and eye diagram, respectively. The free space signal was received at about 0.8 Vpp. When comparing the electrical patterns of Fig. 4 with the received free space signal in Fig. 11, the recovered free space data had noticeable amplitude noise. However, this noise had little effect on the degradation of the received data as supported by the opening of the eye diagram in Fig. 12. Next, the free space channel was replaced with an aquatic channel. The results are given in Figs. 13 and 14, respectively. At this point, the laser biasing voltage was at 5 V because of the strong signal attenuation in water channel. As Fig. 13 shows, the signal was reduced from about 4 Vpp before underwater to approximately 0.8 Vpp.
When comparing the free space pattern of Fig. 11 with the underwater pattern of Fig. 13, both signals are comparable in term of performance. Their respective eye diagrams in Figs. 12 and 14 also gives similar conclusion. This is reasonable since both signals were taken at similar amplitude values. To give a clear indication about the effects of free-space and underwater transmission channels on the PRBS pattern, we performed a quantitative analysis by calculating the bit error rate (BER) values of the signal at different laser bias voltage. Starting with the free space transmission channel, the laser power was attenuated by varying the bias voltage of the laser from 4 V to 0.3 V while recording the signals at each bias voltage for offline BER calculations. The results are given in Fig. 15. For free-space transmission, the minimum BER was 10−13 at 4 V. At this very same biasing voltage, the underwater BER was approximately 10−5. This value corresponds to the minimum BER achieved for underwater transmission. A very low bias voltage of about 0.6 V was needed to achieve a bit error rate of 10−5 for free-space transmission. Figure 15 also shows that even though the qualitative results in Figs. 11 to 14 show a similar performance between free space transmission and underwater transmission at about 0.8 V bias voltage, the calculated BER values show that free space transmission had better performance compared to underwater transmission. This proof of concept shows that it is possible to use the red laser at 650 nm to transmit data signals up to 2 Mbps in an underwater channel of about 0.7 meters. For geodesy applications, the laser-based system can be used for upstream and downstream transmission of data from a robot located underwater to a base station located somewhere above the water channel. Forward error correction techniques can be used to improve the BER values recorded in this study for the extension of the transmission distances when necessary.
5 Conclusions
This paper investigated and demonstrated a simple and low-cost underwater optical wireless link for transmission of clock and data signals. A laser emitting at 650 nm was used as the optical source. A water tank measuring $0.7 \times 0.1 \times 0.1$ meters (length, width, and height, respectively) was used to emulate an aquatic channel. The underwater transmission performance was evaluated against the free space optical channel. When the 2-kHz clock was transmitted in free space, the signal spectrum was not distorted. That is, the recovered clock signal had the same spectral shape as the modulating electrical clock signal. However, amplitude noise was evident in the detected clock signal after free space transmission. The phase noise values of the recovered clock, especially at high offset frequencies, was also high compared to that of the electrical signal.
Underwater transmission decreased the detected clock’s amplitude due to water attenuation. The phase noise of the detected clock after underwater transmission was high compared to the clock after free space transmission. However, the spectral profile of the received clock signal after underwater transmission matched that of the clock signal after free space transmission. Therefore, the out of band amplitude noise present in the recovered clock signal and the reduction in the clock signal power can be reduced by using narrow band filters and amplification, respectively. Nevertheless, the attained phase noise values after underwater transmission could allow for the clock to be used for synchronization between the base station and the deployed underwater robotic measuring system(s).
Qualitative analysis of the transmitted PRBS data after free space and underwater transmission at a peak voltage of about 0.8 V showed error-free data for both channels. However, quantitative results in term of BER showed that free space transmission had a better performance compared to underwater transmission. For example, at a biasing voltage of about 4 V, underwater PRBS data transmission attained a BER value of 10−5 compare to a BER value of 10−12 after free space transmission. This implies that forward error correction techniques will be needed to attain error-free underwater data transmission over 1 meter when a 650 nm laser source is used as an optical source at 2 Mbps.
Funding
Telkom; Dartcom; ATC South Africa; Lambda Test Equipment; National Research Foundation; South African International Maritime Institute; Marine Research Unit; South Africa Radio Astronomy Observatory.
Acknowledgements
This research is funded by Telkom, Dartcom, ATC South Africa, Lambda Test Equipment, the National Research Foundation, South African International Maritime Institute, Marine Research Unit, and South Africa Radio Astronomy Observatory, to whom the authors would like to express their gratitude.
Disclosures
The authors declare no conflicts of interest.
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.
References
1. Z. Jiang, “Underwater acoustic networks–issues and solutions,” International journal of intelligent control and systems 13(3), 152–161 (2008).
2. P. Saini, P. S. Rishi, and S. Adwitiya, “Path loss analysis of RF waves for underwater wireless sensor networks,” in 2017 International Conference on Computing and Communication Technologies for Smart Nation (IC3TSN), pp. 104–108 (2017).
3. X. Zhai, Z. Meng, H. Zhang, X. Xu, Z. Qian, B. Xue, and H. Wu, “Underwater distance measurement using frequency comb laser,” Opt. Express 27(5), 6757–6769 (2019). [CrossRef]
4. Z. Zeng, F. Shu, Z. Huihui, D. Yuhan, and C. Julian, “A survey of underwater optical wireless communications,” IEEE Commun. Surv. Tutorials 19(1), 204–238 (2017). [CrossRef]
5. J. Shi, Z. Shengye, and Y. Can-Jun, “High frequency RF based non-contact underwater communication,” in 2012 Oceans-Yeosu, pp. 1–6 (2012).
6. S. Q. Duntley, “Light in the sea,” J. Opt. Soc. Am. 53(2), 214–233 (1963). [CrossRef]
7. M. Atef, R. Swoboda, and H. Zimmermann, “Real-Time 1.25-Gb/s Transmission Over 50-m SI-POF Using a Green Laser Diode,” IEEE Photon. Technol. Lett. 24(15), 1331–1333 (2012). [CrossRef]
8. B. Cochenour, L. Mullen, and J. Muth, “Temporal response of the underwater optical channel for high-bandwidth wireless laser communications,” IEEE J. Ocean. Eng. 38(4), 730–742 (2013). [CrossRef]
9. D. Tsonev, S. Videv, and H. Haas, “Towards a 100 Gb/s visible light wireless access network,” Opt. Express 23(2), 1627–1637 (2015). [CrossRef]
10. N. G. Jerlov, Marine Optics; Elsevier Oceanography Series: (Elsevier, 1976)
11. S. K. Sahu and P. Shanmugam, “Atheoretical study on the impact of particle scattering on the channel characteristics of underwater optical communication system,” Opt. Commun. 408, 3–14 (2018). [CrossRef]
12. L. Prieur and S. Sathyendranath, “An optical classification of coastal and oceanic waters based on the specific spectral absorption curves of phytoplankton pigments, dissolved organic matter, and other particulate materials,” Limnol. Oceanogr 26(4), 671–689 (1981). [CrossRef]
13. J. Xu, Y. Song, X. Yu, A. Lin, M. Kong, J. Huan, and N. Deng, “Underwater wireless transmission of high-speed QAM-OFDM signals using a compact red-light laser,” Opt. Express 24(8), 8097–8109 (2016). [CrossRef]
14. Hawkeye Detector Datasheet, Available online: https://docs.rs-online.com/a2b4/0900766b813e4013.pdf
