Open Access
Add to BibSonomy
Abstract
Extending the field-of-view (FoV) of underwater wireless optical communication (UWOC) receivers can significantly ease the need for active positioning and tracking mechanisms. Two bundle of scintillating fibers emitting at 430- and 488-nm were used to detect two independent signals from ultraviolet and visible laser sources. A zero-forcing approach to minimize inter-channel crosstalk was further implemented. A net aggregated UWOC data rate of 1 Gb/s was achieved using two wavelengths and a non-return-to-zero on-off keying scheme.
© 2021 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Underwater wireless optical communication (UWOC) is seen as a potential technique to complement acoustic wave communication and to provide connectivity between future generation Internet of Underwater Things (IoUT) devices [1,2]. One of the major challenges for UWOC systems deployment is fulfilling the pointing, acquisition, and tracking (PAT) requirements. The detection areas of commercially available high-bandwidth photodetectors (PDs) are limited to only a few tens of mm$^{2}$, due to the limit imposed by the resistor-capacitor (RC) time constant [3,4], which results in need a to use optical focusing elements to increase the received power. The focusing elements have limited angle of view that requires maintaining a perfect system alignment. Such a condition could not be continuously ensured in a harsh underwater environment that could be subject to various propagation effects and mobility. It is possible to concentrate light into high-bandwidth photodetectors and increase the signal-to-noise ratio (SNR) at the receiver using optical elements, including lenses and compound parabolic concentrators (CPCs) [5]. However, lenses and CPCs are based on reflection and refraction. They, therefore, conserve étendue, which limits the field-of-view (FoV) of the detector [6]. One solution to extend the FoV of optical detectors beyond the étendue limit is through the use of fluorescent materials [7–9]. The authors of [7] reported the design of a slab luminescent solar concentrator (LSC) with a $\pm 60^\circ$ FoV and a collection gain of 12. The demonstrated fluorescent antenna was used to conduct a 190-Mb/s on-off keying (OOK) VLC transmission. Another demonstration using an LSC with a 100-MHz bandwidth and an optical gain of 3.2 reported the transmission of 400-Mb/s signals using an orthogonal frequency division multiplexing (OFDM) modulation scheme [8]. The fact that two LSCs with different materials can be used to decode independent signals carried by different wavelengths enables wavelength-division multiplexing (WDM) transmission as demonstrated in [9]. Scintillating fibers have also been used in indoor optical wireless [10], and underwater wireless optical communication (UWOC) [11] scenarios to extend the FoV of optical receivers. Although many reports in the literature have shown the effectiveness of having multiple receivers to extend the FoV (i.e., receiver diversity) [12], the increase of power consumption and computation complexity scale up with the increase of the number of receivers . Additionally, the limited spatial dimensions for underwater nodes pose more challenges in increasing the number of receivers to achieve a wide FoV. Therefore, a single receiver with a passive optical antenna to extend the FoV is highly attractive.
The concept of luminescent scintillating fibers relies on the optical absorption of incoming light beam photons by dye molecules doped in the fiber core, which emits secondary photons at longer wavelength with a similar process as LSCs. Authors of [10] demonstrated a multi-Gbps OFDM transmission using a planar detector with an active area of 126 cm$^{2}$. The demonstration was based on luminescent scintillating fibers that absorb blue light and re-emit green light. The concept of the use of scintillating fibers for the design of a large area detector has been expanded in [11]. The authors reported the design of a 36 cm$^{2}$ detector for UWOC applications. Restricted by the fiber type used for the detector design, the detector was used to detect UV light emitted by a 377-nm laser. The received signal is down-converted to blue light and decoded by an avalanche photodetector (APD). A data rate of 250-Mb/s over an underwater link was reported using an OOK modulation with a bit error ratio (BER) below the limit of forward error correction (FEC) of $3.8\times 10^{-3}$. The flexible nature of plastic optical fibers makes it very straightforward to fabricate practical shapes optimized for different applications [13,14] along with optical wireless communication [10,11]. The differences between the various high FoV solutions proposed in the literature are summarized in Table 1. Among all the demonstrated techniques, scintillating fibers offer flexibility and can be used to obtain near 360$^\circ$ FoV receivers.
Table 1. Comparison between proposed large FoV OWC detectors in the literature.
In this context, we demonstrate a dual-wavelength scintillating fibers-based detector suitable for UWOC systems. Two types of scintillating fibers were used to detect signals emitted by two different sources among 377-, 405-, and 450-nm lasers. A passive optical filter is used to separate the signals converted to higher wavelengths. A zero-forcing equalization approach is further applied to minimize crosstalk between both signals. 1-Gb/s data rate using non-return-to-zero (NRZ) OOK was achieved.
1.1 Optical characterization
To select the optimal wavelengths of the light sources for WDM, it is essential to characterize both fibers optically. The emission spectra, alongside absorption spectra, provide crucial information about the level of inter-channel crosstalk. To verify the down-conversion process of the scintillating fibers, we used an Agilent Cary Eclipse fluorescence spectrometer to firstly measure the light intensity as a function of excitation wavelength for each fiber. This absorption measurement leads us to identify each fiber’s peak excitation wavelength to obtain the photoluminescence (PL) measurement at the peak excitation wavelength. Figure 1(a) shows the absorption measurement for both fibers. The figure is also color-coded for better visualization. We limit the wavelength range between 300- and 520-nm, which covers the range of wavelengths of interest for our case. It can be seen that both fibers exhibit a wide range of excitation wavelengths. Furthermore, the optimal excitation wavelength for the blue fiber is at 377-nm, while the green fiber has a peak excitation wavelength at 409-nm. Therefore, we choose 377-nm and 405-nm lasers for our experiment. There is a clear overlap between the two absorption spectra. For example, using a 377-nm laser can maximize the excitation of the blue fiber but also causes about 78% of crosstalk, with respect to the used wavelength, at the green fiber. However, due to the presence of the long-pass filter, with its transmittance shown in the black curve, the crosstalk can be diminished. In addition, using a 405-nm laser can fully excite the green fiber while also causing about 32% of crosstalk. Due to the broad absorption spectra, we can eliminate the induced interference by selecting a longer wavelength that does not cause crosstalk to the blue fiber. From Fig. 1(a), we chose a high-power 450-nm laser due to its availability. However, an adequate wavelength can be selected, maximizing the excitation without causing crosstalk such as 420-430 nm. The penalty of using 450-nm compared to 420-430-nm to excite the green fiber is that double the laser power is needed to reach the same excitation level of the blue fiber when excited by a 377-nm laser. Nevertheless, by fixing the excitation wavelengths at 377-nm and 405-nm for the blue and green fibers, respectively, the PL measurement in Fig. 1(b) was obtained. The down-conversion process can be confirmed as the blue fiber has a peak emission at 430-nm while the green fiber has a peak emission at 488-nm. It can be argued that emitted photons of blue fiber (i.e., 430-nm) can also cause inter-channel interference. This is because the trapping efficiency of the fiber is dependent on the refraction indices of the core $n_{\textrm {co}}$ and cladding $n_{\textrm {cl}}$ layers (1.6 and 1.49, respectively). The fraction of power of the re-emitted light $P_{\textrm {t}}$ that is guided towards the ends of the fiber is [16]
where $\theta$ is the maximum angle that ensures guidance. For the above-mentioned refractive indices values, the trapped light power is about 6.88% of the total generated power with a maximum angle of re-emission of 21.4 degrees, ensuring total-internal-reflection [16]. Therefore, the rest of the emitted energy can be lost through secondary absorption of the fluorescent molecules or escapes outside the fiber. This indicates that inter-channel crosstalk is possible when the two fibers are stacked on top of each other. Therefore, the emission of 430-nm, which is generated from the blue fiber, can affect the value of $h_{GB}$.Another crucial optical characterization is the Time-Resolved Photo-Luminescence (TRPL) measurement which provides an insight into fluorescence lifetime. As shown in Fig. 1(c), the fluorescence lifetime of the blue and green fibers are almost identical, with corresponding $-$3-dB bandwidth of 83.32-MHz and 88.92-MHz, respectively. Because the antenna’s bandwidth is determined by Four processes [10], we confirmed that the fluorescence lifetime is the dominant factor by measuring the system’s frequency response. As shown in Fig. 1(d), the $-$3-dB of the overall system for the blue and green channels are identical to those obtained from fluorescence lifetime. Although the $-$3-dB bandwidth seems limited, the frequency response of the channels decays slowly, followed by a sharp decrease at 400-MHz due to the limited bandwidth of the photodetector. Therefore, a $-$10-dB of about 150-MHz can still be utilized for high-speed communication.
Fig. 1. Photophysics of scintillating fibers: (a) Absorption spectra of both fibers are shown in blue and green curves, and the black curve shows the filter transmittance. (b) The corresponding emission spectra for the blue and green fibers show an emission peak of 430- and 488-nm, respectively. (c) Time-resolved fluorescence measurement of the two fibers, excited at the peak absorption wavelengths. (d) Normalized frequency responses of the UV and blue fibers, showing $-$3-dB bandwidth in the range of 83- to 89-MHz, respectively.
To verify the omnidirectional capability of the scintillating fibers, we measured the FoV as shown in Fig. 2(a). The FoV measurement was conducted by rotating the 377-nm laser around the scintillating fiber that is coupled to a power meter. The 377-nm laser was placed at different angle locations indicated by the red dots in Fig. 2(a). Due to the presence of the power meter, we measured the FoV at angles between 0-150$^\circ$ and 210-360$^\circ$. Figure 2(a) shows that scintillating fibers can exhibit near 360$^\circ$ FoV, depending on the design, which is highly convenient for underwater wireless optical communication. Figure 2(b) (left-hand-side) shows a compact WDM design where the fibers are intertwined inside each other to be fit inside a water capsule for future deployment in the sea. The fibers under illumination can also be seen in Fig. 2(b) (right-hand-side).
Fig. 2. (a) FoV measurement of scintillating fiber showing omnidirectional signal reception. The blue circular curve indicate a constant signal receivption along all angles (b) Example of compact underwater capsule design for WDM with wide FoV before illumination (left) and after illumination (right). Fiber B and Fiber stand for the blue and green fibers, respectively. (c) Total efficiencies measurement of both fibers.
In addition, we measured the overall efficiency of both fibers which takes into account the trapping, conversion and collection efficiencies. Figure 2(c) shows the relationship between the output power with and without the fibers. The slope of the linear fit indicates total efficiencies of 1.58% and 0.873% for the the green and blue fibers, respectively. These measurement also align well with previously measured efficiency value in the literature [10]. With further improvement of the fluorescent dyes and fabrication enhancement of core and cladding, the conversion and trapping efficiencies can be improved. In addition, by engineering a specific phase plate for each fiber string, the light output from each fiber string can be focused to a single point, thereby improving the collection efficiency. Nevertheless, even with these low efficiencies, we can still obtain high speed communication as demonstrated in the next sections.
2. Experimental setup and data-stream generation
Figure 3(a) shows the experimental setup used in the work. A 377-nm laser diode (Nichia, NDU4116) was used as a transmitter alongside a 405-nm laser diode (Nichia, NDV4316). The two beams were collimated then combined using a beam splitter (Thorlabs, BSW20). For cooling, the 377-nm laser was mounted on a thermoelectric-cooled laser mount (Thorlabs, TCLDM9), which was controlled by an electro-thermal controller (Thorlabs, L375P70MLD). The 405-nm laser was connected to an external fan powered by a power supply (Rigol, DP832A). An arbitrary waveform generator (AWG) (Siglent, SDG6052X) was used to output NRZ-OOK signals. The output signal from the AWG was then combined with the DC source via a bias-tee then fed into the 405-nm laser. A BER tester (Agilent J-BERT, N4903B) was used to output the NRZ-OOK signal provided to the 377-nm laser. The two beams enter a water tank that is 1-m long, 12-cm wide, and 12-cm high, filled with pure water with attenuation coefficient of $c=0.07$ $m^{-1}$. At the end of the tank, two scintillating fibers (Saint-Gobain Crystals, BCF-10, BCF-92) were attached to the inner sidewall of the tank. Each fiber type was first tightly packed to form an array of scintillating fibers with a large detection area of 1 inch in width. The longitudinal extent of the fibers was elongated to reach a silicon avalanche photodetector (Thorlabs, APD430A2/M) which was placed outside of the tank as shown in Fig. 3(a). The facets of the fibers at the APD end were polished using sanding paper to maximize the output light power and SNR. The cleaved fiber bundle was squeezed to form a circular array and directly coupled to the APD. The second type of scintillating fiber (blue-green shifter, and denoted as the green fiber throughout this paper) was stacked underneath the first fiber array (UV-blue shifter and denoted as the UV fiber throughout this paper) then coupled to its corresponding APD in the same manner. An optical long-pass filter with a cut-off wavelength of 400-nm (Thorlabs, FELH0400) was sandwiched between the fiber arrays so that an incoming 377-nm laser light can excite the first layer of the fibers but blocked at the second layer. However, the 405-nm laser beam can penetrate through the filter and excite the second layer of the fibers. Therefore, the multiplexed wavelength can be separated using the scintillating fibers with the advantage of wide FoV and omnidirectional capabilities. The optoelectrical converted signals from the APDs were then fed into a high-speed oscilloscope (Tektronix, MDO3104) and the BER tester. The connection was alternated depending on which channel is the victim and the aggressor.
To test the communication performance, a pseudo-random binary sequence (PRBS) was generated using a linear feedback shift register, which was implemented in a MATALB platform. The PRBS sequence has a polynomial order of 17. However, the PRBS signals for the two data channels have different seeds. The data signals were generated according to the following polynomials
Fig. 3. (a) Schematic illustration of the experimental setup for the dual-wavelength UWOC transmission using scintillating fibers receiver over a 1-m long channel. (b) Block diagram of the dual-channel communication scenario.
3. Results and discussion
3.1 Separating channels using passive optical elements
In this section, we begin to characterize the effect of crosstalk with different combinations of source wavelengths (i.e., 377, 405, and 450 nm). First, we only switched on the UV laser and measured the BER at different data rates, which serves as a reference curve. As shown in Fig. 4(a), an aggregated data rate of 1 Gb/s (i.e., 500 Mb/s from each fiber) with a BER value of $3.5\times 10^{-3}$ was achieved, which is below the forward error correction (FEC) limit of $3.8\times 10^{-3}$. Afterward, the blue channel was aggressed by the 405-nm laser. As seen from Fig. 4(a), significant deterioration in the BER is observed, causing a reduction in the data rate by 20%. To solve this issue, we replaced the 405-nm laser with the 450-nm laser. Clearly, the results show that the blue channel was not compromised when aggressed by the 450-nm laser.
Fig. 4. BER versus data rate for various channel conditions. (a) when blue channel is aggressed by 405-nm laser, and when the blue channel is aggressed by 450-nm laser, (b) the effect of removing the optical filter when the green channel is aggressed by 377-nm laser, and the effect of green channel when aggressed by 377-nm laser with filter installed.
Next, we investigate the effect of removing the filter between the two stacked fibers, so now the 377-nm laser is considered an aggressor while the green channel is treated as a victim. Figure 4(b) shows the effect of removing the long-pass filter which separates the two fiber arrays. It can be seen that communication is challenging in such a scenario as the crosstalk causes incorrect detection of the bit sequence. Finally, when the filter is sandwiched between the two fibers, the green channel can be demodulated correctly. This demonstrates one potential technique of achieving UWOC WDM using entirely passive optical elements with the correct selection of source wavelengths.
Figure 5 shows the corresponding eye diagrams for the abovementioned conditions, measured at 200-Mb/s. Figures 5(a) and (b) are presented as the reference eye diagrams when both channels are crosstalk-free. When a 405-nm laser aggresses the blue channel, the eye height collapses significantly, as seen in Fig. 5(c). This is because the crosstalk level from the 405-nm wavelength to the blue channel is around one-third, the eye diagram still preserves its features and could be corrected using error correction techniques. By replacing the 405-nm with the 450-nm laser with the resultant eye diagram in Fig. 5(d), the eye-opening is restored as in Fig. 5(a). Finally, the corresponding eye diagrams showing the effect between removing and installing the long-pass filter are illustrated in Fig. 5(e),(f). In this case, because the crosstalk level from the UV laser to the green fiber is comparable to the victim’s signal, the eye diagram completely closes.
Fig. 5. The measured eye diagram at 200-Mb/s for (a) 377-nm laser at the blue channel, (b) 450-nm laser at the green channel, (c) when the blue channel is aggressed by 405-nm laser, (d) when the blue channel is aggressed by 450-nm laser, (e) when the filter is removed from the green channel and aggressed by 377-nm laser, and (f) when the green channel is aggressed by 377-nm laser in the presence of the optical filter.
To further quantify the effect of the crosstalk, we measured the SNR of the victim’s signal (i.e., blue channel) versus varying levels of transmit power of aggressing lasers (i.e., 450 and 405 nm). We first placed a power meter (Newport, 2936-C) in front of the fiber arrays to independently measure the incident light power from 450-nm and 405-nm lasers. We then recorded the drive currents for both lasers that lead to the same incident power at the plane of the fibers. The results in Fig. 6 show that using the 405-nm laser yields an linear decay of the victim’s SNR with a decay rate of 0.53 dB.mW$^{-1}$. This decay corresponds to the increased level of crosstalk, causing the eye-opening to collapse linearly, relative to the power increase from the aggressor. However, using the 450-nm laser shows a constant level of SNR, confirming a crosstalk-free channel.
3.2 Software-based equalization
In the next section, we demonstrate the use of software-based equalization to correct for crosstalk. This method is beneficial when the receiving antennas are designed for practical use, especially when installing a flexible filter that separates the two fibers is challenging. Such a design will be described in Section 3.3. Having a digital signal processing unit allows the two fibers (or more) to be mixed for compact antenna design while still achieving the WDM capability. Among all equalization techniques, we opted for a linear zero-forcing (ZF) equalizer for its simplicity and efficiency. The standard $2 \times 2$ multi-input multi-output (MIMO) model is applicable, where the received symbols of the first and second channels are modeled as
For real and squared matrices, this expression reduces to simply the inverse of the $H$ matrix. Therefore, while ignoring the noise term, the estimated symbols can be written as
To this point, the task is to acquire the $H$ matrix, which can be obtained by two methods. The first is by referring to Fig. 1(a) and denote the elements in the $H$ matrix as the level of absorption of the fiber to incident wavelength. However, This method does not account for the effect channel fading. The second is to send each data stream with a preamble containing known bits to both receiver and transmitter (i.e., training bits). The training bits of the dual-channel are orthogonal, allowing us to extract all elements in the channel matrix. This means that when the UV laser is sending a symbol ‘1’, the 405-nm laser sends a ‘0’ and vice versa. For example, when the UV laser is on state, we can measure both $h_{BB}$ and $h_{GB}$ by computing the ratio of the transmitted voltage to the received voltage from the blue and green channels, respectively. The rest of the channel elements can be estimated when the UV transmits a ‘0’ while the Violet laser transmits a ‘1’. This method is more accurate because it accounts for the effect of channels fading. The calculated channel matrix for the 377- and 405-nm lasers is
We can notice that the value of $h_{GB}$ and $h_{BG}$ are quite similar from the one in Fig. 1 with slight offset due to various attenuation factors such as the propagation in the water. After the equalization using Eq. (6), a bathtub-like curve that resembles the BER versus decision threshold values at 200 Mb/s is shown in Fig. 7. Since we transmit a total signal length of $1\times 10^{5}$, a BER of zero was plotted as $1\times 10^{-5}$ to fit in the log scale. As illustrated in Fig. 7, the BER curves of blue and green channels without equalization dip below the FEC limit. However, the green curve dips further, which can be due to the higher responsivity of the APD at longer wavelengths. Figure 7(a) shows the corresponding eye diagram of the blue channel. It can be noticed that there exist four voltage levels as expected. The corresponding equalized eye diagram is shown in Fig. 7(b). Clearly, the eye-opening is restored, and the BER curve falls to zero. Similarly, the un-equalized eye diagram of the green channel shown in Fig. 7(c) was corrected to Fig. 7(d). Due to the high scattering property of UV light underwater, UV communication is preferred for non-line-of-sight [17,18]. Therefore, visible wavelengths from violet to green are best suited for LOS applications. Hence, we tested the feasibility of using the 405-nm laser as a transmitter for the blue channel while the 450-nm laser is used for the green channel. According to Fig. 1(a), for these wavelengths, the green channel experiences a crosstalk level of approximately 95%, while the blue channel experiences negligible crosstalk.
Fig. 7. Bathtub curve for blue and green channels using 377- and 405-nm transmitters. A BER of $10^{-5}$ represents a BER of zero since the transmitted signal is 100 kbit in length.
Figure 8 shows BER curves for the equalized and un-equalized received signal. The corresponding eye diagram is represented in Fig. 8(a). However, it also shows that the un-equalized green channel cannot sustain reliable communication due to the significant crosstalk. Because the crosstalk is about 95%, we can observe two clear eyes stacked on top of each other in Fig. 8(b). This is because when both the victim and the aggressor are transmitting a ‘1’, the superposition of both signals generates a signal with double the voltage. While the victim transmits a ‘0’ and the aggressor transmits a ‘1’, we cannot distinguish between the genuine ‘1’ from the crosstalk, which means that the communication is completely perturbed. Therefore, obtaining the $H$ matrix using the training symbols is not possible because the training symbols are corrupted. In this case, the system can adopt a calibration step prior to transmission. The calibration procedure is carried out by switching on one laser and while the second laser is off and measure the DC level for a long interval at the victim’s channel. The matrix coefficient is the expected value of the observations. The same procedure is applied to the second laser. Thus the matrix elements can be obtained. With this technique, the equalized eye diagram for the green channel is drawn in Fig. 8(c), with a measured channel matrix of
3.3 Potential design
From these results, we propose a water-air optical communication relay using the scintillating fibers-based transceiver. The proposed device is demonstrated in Fig. 9. The structure consists of a spherical transparent enclosure where two types of scintillating fibers are wrapped across the inner perimeter of the spherical enclosure, forming a large-detection-area receiver. The two ends of the fibers are directly coupled to two photodetectors. Furthermore, due to the compactness of laser diodes, two lasers can be installed, which relay signals from air to water. The structure can also be self-powered by utilizing solar energy [19] or ocean wave energy [20].
Fig. 9. Illustration of the proposed optical transceiver in potential transmitting and receiving modes.
4. Conclusion
In conclusion, we demonstrated a scintillating fibers’ optical antenna that utilizes wavelength-division multiplexing to increase the communication throughput. We demonstrated the optimal selection of wavelengths based on the fibers’ characteristics. In addition, we reported on two ways of mitigating the crosstalk between the two signals using a passive optical element and a software-based zero-forcing equalizer. A net data rate of 1-Gb/s was achieved using an NRZ-OOK modulation. Our proof-of-concept work paves the way toward large-area detection and near-omnidirectional optical transceivers for underwater and water-to-air light-based communication applications.
Funding
King Abdullah University of Science and Technology (BAS/1/1612-01-01, BAS/1/1614-01-01, GEN/1/6607-01-01, KCR/1/2081-01-01, KCR/1/4114-01-01); King Abdulaziz City for Science and Technology (KACST TIC R2-FP-008).
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. Zeng, S. Fu, H. Zhang, Y. Dong, and J. Cheng, “A survey of underwater optical wireless communications,” IEEE Commun. Surv. Tutorials 19(1), 204–238 (2017). [CrossRef]
2. C.-C. Kao, Y.-S. Lin, G.-D. Wu, and C.-J. Huang, “A comprehensive study on the internet of underwater things: applications, challenges, and channel models,” Sensors 17(7), 1477 (2017). [CrossRef]
3. Z. P. Zhao, J. G. Liu, Y. Liu, and N. H. Zhu, “High-speed photodetectors in optical communication system,” J. Semicond. 38(12), 121001 (2017). [CrossRef]
4. K.-T. Ho, R. Chen, G. Liu, C. Shen, J. Holguin-Lerma, A. A. Al-Saggaf, T. K. Ng, M.-S. Alouini, J.-H. He, and B. S. Ooi, “3.2 gigabit-per-second visible light communication link with ingan/gan mqw micro-photodetector,” Opt. Express 26(3), 3037–3045 (2018). [CrossRef]
5. J. Kahn and J. Barry, “Wireless infrared communications,” Proc. IEEE 85(2), 265–298 (1997). [CrossRef]
6. R. Mulyawan, A. Gomez, H. Chun, S. Rajbhandari, P. P. Manousiadis, D. A. Vithanage, G. Faulkner, G. A. Turnbull, I. D. W. Samuel, S. Collins, and D. O’Brien, “A comparative study of optical concentrators for visible light communications,” in Broadband Access Communication Technologies XI, vol. 10128B. B. Dingel, K. Tsukamoto, and S. Mikroulis, eds., International Society for Optics and Photonics (SPIE, 2017), pp. 142–147.
7. P. P. Manousiadis, S. Rajbhandari, R. Mulyawan, D. A. Vithanage, H. Chun, G. Faulkner, D. C. O’Brien, G. A. Turnbull, S. Collins, and I. D. Samuel, “Wide field-of-view fluorescent antenna for visible light communications beyond the étendue limit,” Optica 3(7), 702–706 (2016). [CrossRef]
8. Y. Dong, M. Shi, X. Yang, P. Zeng, J. Gong, S. Zheng, M. Zhang, R. Liang, Q. Ou, N. Chi, and S. Zhang, “Nanopatterned luminescent concentrators for visible light communications,” Opt. Express 25(18), 21926–21934 (2017). [CrossRef]
9. P. P. Manousiadis, H. Chun, S. Rajbhandari, D. A. Vithanage, R. Mulyawan, G. Faulkner, H. Haas, D. C. O’Brien, S. Collins, G. A. Turnbull, and I. D. W. Samuel, “Optical antennas for wavelength division multiplexing in visible light communications beyond the Étendue limit,” Adv. Opt. Mater. 8(4), 1901139 (2020). [CrossRef]
10. T. Peyronel, K. J. Quirk, S. C. Wang, and T. G. Tiecke, “Luminescent detector for free-space optical communication,” Optica 3(7), 787–792 (2016). [CrossRef]
11. C. H. Kang, A. Trichili, O. Alkhazragi, H. Zhang, R. C. Subedi, Y. Guo, S. Mitra, C. Shen, I. S. Roqan, T. K. Ng, M.-S. Alouini, and B. S. Ooi, “Ultraviolet-to-blue color-converting scintillating-fibers photoreceiver for 375-nm laser-based underwater wireless optical communication,” Opt. Express 27(21), 30450–30461 (2019). [CrossRef]
12. O. Alkhazragi, F. Hu, P. Zou, Y. Ha, C. H. Kang, Y. Mao, T. K. Ng, N. Chi, and B. S. Ooi, “Gbit/s ultraviolet-c diffuse-line-of-sight communication based on probabilistically shaped dmt and diversity reception,” Opt. Express 28(7), 9111–9122 (2020). [CrossRef]
13. R. Mangeret, J. Farenc, B. Ai, P. Destruel, D. Puretolas, and J. Casanovas, “Optical detection of partial discharges using fluorescent fiber,” IEEE Trans. Electr. Insul. 26(4), 783–789 (1991). [CrossRef]
14. J. Farenc, R. Mangeret, A. Boulanger, P. Destruel, and M. Lescure, “A fluorescent plastic optical fiber sensor for the detection of corona discharges in high voltage electrical equipment,” Rev. Sci. Instrum. 65(1), 155–160 (1994). [CrossRef]
15. O. Alkhazragi, A. Trichili, I. Ashry, T. K. Ng, M.-S. Alouini, and B. S. Ooi, “Wide-field-of-view optical detectors using fused fiber-optic tapers,” Opt. Lett. 46(8), 1916–1919 (2021). [CrossRef]
16. J. D. Weiss, “Trapping efficiency of fluorescent optical fibers,” Opt. Eng. 54(2), 027101 (2015). [CrossRef]
17. S. Zhu, X. Chen, X. Liu, G. Zhang, and P. Tian, “Recent progress in and perspectives of underwater wireless optical communication,” Prog. Quantum Electron. 73, 100274 (2020). [CrossRef]
18. M. Sait, X. Sun, O. Alkhazragi, N. Alfaraj, M. Kong, T. K. Ng, and B. S. Ooi, “The effect of turbulence on NLOS underwater wireless optical communication channels,” Chin. Opt. Lett. 17(10), 100013 (2019). [CrossRef]
19. J. I. d. O. Filho, A. Trichili, B. S. Ooi, M.-S. Alouini, and K. N. Salama, “Toward self-powered internet of underwater things devices,” IEEE Commun. Mag. 58(1), 68–73 (2020). [CrossRef]
20. C. Rodrigues, D. Nunes, D. Clemente, N. Mathias, J. M. Correia, P. Rosa-Santos, F. Taveira-Pinto, T. Morais, A. Pereira, and J. Ventura, “Emerging triboelectric nanogenerators for ocean wave energy harvesting: state of the art and future perspectives,” Energy Environ. Sci. 13(9), 2657–2683 (2020). [CrossRef]
