Solar-Blind Optical Wireless Communications Over 80 Meters Using a 265-nm High-Power Single-Chip DUV-LED Over 500 mW in Sunlight

Direct eye-pattern measurements based on a deep-ultraviolet (DUV) light emitting diode (LED) optical wireless communication system were performed in sunlight. A state-of-the-art, high-power DUV-LED was developed and used as a transmitter consisting of only one LED chip with a light output power greater than 500 mW at its peak emission wavelength of 265 nm. A DUV light receiver was also developed that excluded sunlight with high performance. We demonstrated clear eye-patterns at 1 Mbps in both line-of-sight (LOS) and non-line-of-sight (NLOS) configurations over a communication distance of 80 meters in sunlight.


I. INTRODUCTION
O PTICAL wireless communications is receiving growing attention owing to a potentially huge range of unlicensed spectral bands with large bandwidth, which may be used to send data at high speeds without being affected by electromagnetic interference [1], [2]. Visible and infrared optical communications have been widely applied in indoor line-of-sight (LOS) communications. In these communications applications, a signal transmitter and receiver must be linked in a straight line within a free space. These optical signals are subject to interference from sunlight when used in outdoor environments, which markedly affects the quality of digital signals and the speed of data transmission. Deep-ultraviolet (DUV) light, which covers an emission wavelength range of 200-280 nm, does not naturally exist at the Earth's surface because of strong absorption by the ozonosphere before arriving at the ground level. Thus, DUV light signals could avoid the influence of background noise arising from sunlight. In addition, DUV-light can be strongly scattered as it interacts with aerosols or molecules in the atmosphere, which suggests potential for use in non-line-of-sight (NLOS) wireless communications. These unique characteristics of DUV light make solar-blind DUV communications attractive for applications in NLOS optical communications [3], [4].
The light source is one of the most important components in optical communications. The study of DUV communication systems in outdoor environments dates back to at least the 1960s [5], and a high data rate of 1.2 Mbps for NLOS DUV communications was realized experimentally in 1990 with the use of mercury lamps as the light source [6]. The aluminum gallium nitride (AlGaN)-based DUV light-emitting diode (LED) is a new light source that can replace bulky, fragile, and toxic mercury lamps. The LED has the advantages of being small in size with a compact design, modulable spectrum, and broad bandwidth. Rapid progress has also been made in development of LOS optical communications through the use of DUV-LEDs as signal transmitters over the past several years. Omar et al. [7], [8] used DUV-LEDs operating at 279 nm to achieve highspeed communications through diffuse-LOS within a certain angle range up to 5 meters by improving the communication modulation method. He et al. [9] achieved a large bandwidth DUV-LEDs with the use of micro-LED technology and realized a 1 Gbps communication speed at face-to-face distances. Zhu et al. [10] improved the bandwidth up to a record value of 452.53 MHz for sub-280 nm DUV-LEDs and reported a data rate of 2 Gbps over 0.5 meters. Daniel et al. [11] also used micro-LED technology and designed a wavelength division multiplexing UV communication system based on three-band light sources, and finally achieved high-speed communications of 10 Gbps at a distance of 0.5 meters. NLOS communications capabilities have also greatly developed with data rates close to Mbps and ranges of hundreds of meters [12], [13], [14], [15], [16], [17], [18]. Theoretical models for estimating the performance of communication links based on Rayleigh and Mie scattering have also been extensively studied [19], [20], [21], [22], [23], [24]. This research has confirmed the feasibility of NLOS optical communications by DUV-LEDs; however, transmission distances and data transfer rates remain low. To further improve NLOS communication rates and distances while maintaining a high signal-to-noise ratio, a DUV light emitter with a much higher light output power would be desirable [15].  including to the skin and eyes of humans [4], the DUV light intensity that remains after Rayleigh and Mie scattering occur in the atmosphere is extremely weak. Therefore, NLOS DUV communication is basically safe for both human eyes and skin at the receiver side. In addition, a well-designed receiver system should eliminate background noise signals from sunlight owing to a lack of high-quality solar-blind ultraviolet photodetectors [25].
In this paper, we performed direct measurements of eye patterns of a DUV-LED communication system in LOS and NLOS configurations in sunlight. To extend the communication distance, we developed a high-power single-chip 265-nm DUV-LED with a maximum output power exceeding 500 mW. Moreover, to enhance the signal to noise ratios, we developed a light receiving system, which excluded sunlight with high efficiency. Using the developed system, we successfully measured clear eye patterns at 1 Mbps for both long-distance LOS and NLOS configurations. We then focused on jitter analysis of the signals measured using these two configurations. Fig. 1(a) shows a photograph of the light transmitting system. We generated pseudo-random bit sequence (PRBS) patterns with a digital waveform generator (PXIe 6547, National Instruments, USA). The PRBS patterns were transferred to the power supply of the LED (ALP-7233ALC10-OP-a, Asahidata Systems Co., Ltd.), resulting in direct current modulations. We set the modulation frequency of the LED at 1 MHz, which is almost the frequency limit of our power supply. The whole transmission system was controlled by LabView (National Instruments, USA). The modulated DUV light was collimated by a UV Fused Silica Plano-Convex Lens (LA4384-UV, Thorlabs, f = 90.0 mm, Ø75 mm, AR Coating: 245-400 nm) and transmitted to air. Fig. 1(b) shows a photograph of the light receiver system. This system comprised an optical system, preamplifiers, and digital oscilloscope. The optical subsystem [ Fig. 2(a)] collects the signal transmitted by a UV Fused Silica Plano-Convex Lens (LA4795-UV, Thorlabs, f = 200.0 mm, Ø75 mm, AR Coating: 245-400 nm). To exclude the sunlight from the collected light, two cold mirrors [ Fig. 2(b)], which reflect any light at wavelengths shorter than 270 nm, were inserted in the optical path after the lens.  The performance of these two cold mirrors is important for excluding sunlight. To show the optical characteristics of the mirror, we measured angle-resolved reflectance spectra of the mirror with an ultra-violet (UV)/visible (VIS)/near-infrared (NIR) spectrophotometer (V-770ST with ARMN-920, Jasco, Japan). We varied the incident angle from 5°to 75°in 10°steps. This mirror was designed so that the reflectance was maximized at a wavelength of 265 nm and at an incident angle of 45°. The reflectance was greater than 80 % from 5°to 55°. Conversely, the reflectance was below 10 % from the UV to the IR region. Therefore, the cold mirrors operated as a filter excluding light longer than 270 nm from the environment.

A. Characteristics of Transmitting and Receiving Systems
Moreover, light absorbent foils (Metal Velvet, Akctar, Israel) were affixed to the enclosure to absorb stray light scattered in the system. In addition, we inserted a solar-blind filter that passed light at wavelengths shorter than 270 nm (FF01-276/SP-25, Semrock, USA) in front of the photodetector. This filter blocks unnecessary light which may be slightly reflected by the cold mirrors. After passing through the short pass filters, the transmitted signals were detected by a photomultiplier tube (PMT) (R7154, Hamamatsu, Japan). The applied voltage on the PMT was −1000 V. The detected signals were amplified by two high speed amplifiers (C1184, Hamamatsu, Japan and DHVA-101, FEMTO Messtechnik GmbH, Germany) and measured by a digital oscilloscope (MSO54, Tektronix, USA). The gain of C1184 was 28 dB ± 2 dB, and we varied the gain of DHVA-101 depending on the observed signal strength. Analysis of the measured signals was conducted on the same oscilloscope. To evaluate the performance of sunlight exclusion, we measured the residual signal from sunlight with the system. In the evaluation, Authorized licensed use limited to the terms of the applicable license agreement with IEEE. Restrictions apply.  we changed the number of cold mirrors to verify the stray light exclusion performance. Fig. 4(a) shows the waveform of the detected signals. The blue and orange curves indicate the waveforms in the presence of a single mirror and double mirrors, respectively. The sunlight signals were observed in both cases, but the signals were markedly lower in the double cold mirror system case.
To discuss the performance quantitatively, a histogram of the detected signals is shown in Fig. 4(b). The horizontal and vertical axes indicate the detected signal in mV and probability normalized to unity, respectively. The blue and red bins correspond to the cases of single and double mirrors, respectively. The mean value and the standard deviation (SD) of the two bins are shown in Table I.
Using the mean value and SD obtained, we added solid curves representing the normal distribution for the single and double mirrors as the guide to the eye. The blue curve did not fit the histogram for the single mirror well, thus indicating that the detected signal did not originate from electronic noise, which is expected to have a normal distribution. However, the red curve fitted well to the histogram for the double mirrors. The mean value of the signal for the double mirror system was close to zero, indicating that the detected signal mainly originates not from sunlight but rather from noise of the system. Moreover, the SD for the double mirror system was approximately one third that of a single mirror system. These results indicate that the sunlight was sufficiently excluded in the light receiving system when the cold mirror was doubled. Hence, we applied the double cold mirror in the following communication experiments.

B. Direct Measurement of Eye Patterns of a DUV Communications System
Using the light receiving system, we performed direct measurements of eye patterns of a DUV-LED communication system in LOS and NLOS configurations during the period from 1 to  [26], [27]. A novel PhC structure containing subwavelength nanostructures was observed to enhance the LEE of DUV LEDs and a DUV LED with light output power (LOP) of more than 90 mW was demonstrated [28]. The required nanostructure was prepared via the nanoimprint lithography technique and a large-area light-extraction nanostructure increased the LOP of the DUV LED to 150 mW [29]. Recently, the area of this nanostructure was enlarged further, and an LOP of more than 500 mW was demonstrated by a single chip [30]. The high output power from the high-light-extraction nanophotonic structures does not affect the far field pattern of the LED, which is preferable for practical applications. Typical characteristics of the LED are shown in the Appendix.
First, we measured eye patterns in a LOS configuration [ Fig. 5(a) and (b)]. The communication distance between the transmitter and receiver was approximately 125 m. The DUV-LED and detector were mounted on a tripod and were approximately 1 m above the ground. The driving current of the LED was varied from 300 to 2000 mA and modulated at 1 MHz. The data rate in this case was 1.0 Mbps. The gain of the DHVA-101 was set to be 10 dB. The jitter characteristics were calculated based on the eye patterns and the corresponding bathtub curves. More than 47000 bits of data were recorded to measure the eye patterns. This quantity of data was sufficient to calculate the jitter characteristics. The other experimental conditions were the same as those of section A. Fig. 6(a) and (b) show the measured eye pattern at 300 and 2000 mA, respectively. We successfully observed a clear eye pattern at 1 MHz in the LOS configuration. Fig. 6(c) shows the jitter of the observed signal as a function of the driving current. From the jitter analysis, we found that total jitter (TJ) at the bit error rate (BER) of 10 −12 was approximately 150 ns and the variation of the TJ followed that of random jitter (RJ). When the driving current was small, data dependent jitter (DDJ) was dominant among the deterministic jitters. As the driving current was increased, duty cycle distortion (DCD) became dominant and the contribution of DDJ to the DJ became small. In the LOS configuration, RJ and periodic jitter (PJ) were almost constant. The maximum and minimum values of the eye width (EW) at the BER of 10 −12 were approximately 861 and 812 ns, respectively. The EW decreases gradually in tandem with the increasing current because of the increase in thermal noise associated with the corresponding temperature rise in the LED. However, the EW decreased by only 5.8% as the driving current was increased, indicating that the EW was almost constant. Through direct measurement of the eye pattern in the LOS configuration, we showed that the EW of the transmitted DUV signals does not strongly decrease. This characteristic is attractive for communication applications.
Following the demonstration of the DUV LOS communications, we moved onto the NLOS configuration [ Fig. 5(c) and (d)]. In this configuration, there is a building between the transmitter and receiver. To realize NLOS configuration and scatter the transmitted DUV light, we placed a plane object with an area of 50 cm square. Aluminum foils with rough surfaces were affixed to the surface of the object to scatter the DUV light. This differs from the case of the diffuse-LOS link with a direct path between the wide-angle transmitters and receivers [7], [8]. The NLOS approach in our experiments relies on nonspecular scattering from the rough surfaces to realize omnidirectional DUV communication links. The distance between the transmitter and the object (d 1 ) was fixed to be approximately 25 m. We moved the position of the receiver along the straight arrow perpendicular to the wall of the building and varied the distance between the Although, the quality of the pattern degraded compared with that of the LOS, there is a clear opening of the eye. As for the jitter characteristics, TJ follows the variation of the RJ. This characteristic is the same as that of the LOS configuration. Among the DJ, we noted that PJ, DDJ, and DCD made approximately the same contribution to the DJ. The maximum and minimum values of the EW at the BER of 10 −12 were approximately 624 and 375 ns, respectively. The EW decreased by 40% with as the distance was increased to d 2 . This decrease in EW was more pronounced than that of the LOS configuration and attributed to the strong scattering nature of DUV light. The reduction in the EW would also be affected by the amplification of the noise signals because the gain of the amplifier increases with increasing distance. However, our measurements showed that direct measurement of the eye pattern was possible even for a communication range (d 1 + d 2 ) of approximately 80 m. The data rate in this case was again 1.0 Mbps.

III. CONCLUSION
We developed a 265-nm high-power single-chip DUV-LED over 500 mW as a transmitter to extend the solar-blind optical wireless communication distance in sunlight. We also developed a light receiving system that effectively excluded stray light. Measurement of the signal under sunlight showed good sunlight exclusion performance. Using the high-power DUV-LED and light receiving system, we performed direct measurements of the eye pattern in LOS and NLOS configurations. In the LOS configuration, we successfully measured a clear eye pattern in sunlight. The modulation frequency and communication distance were 1 MHz and over 125 m, respectively. Even in the NLOS configuration, a clear eye pattern was measured. The data rate and the communication distance were 1 Mbps and over 80 m, respectively. Jitter analysis of the NLOS Authorized licensed use limited to the terms of the applicable license agreement with IEEE. Restrictions apply. configuration showed that NLOS communication was possible over a communication distance of up to 80 m. High-power DUV-LEDs effectively enable high-speed and solar-blind longdistance optical wireless LOS and NLOS communications in sunlight.

APPENDIX CHARACTERISTICS OF THE DUV LED
In this Appendix, we show the output power characteristics and far field pattern (FFP) of the developed DUV-LED. Fig. 8(a) and (b) show the output power characteristics as a function of the injection current and FFP of the developed DUV LED, respectively. The developed single-chip 265-nm DUV-LED had a record output power of approximately 550 mW at an injection current of 2000 mA under continuous-wave conditions at room temperature and a spatially uniform FFP. The maximum light output power of 550 mW was limited by the maximum value of the drive current of the power supply (2000 mA). The wavelength shift that followed the variation in the drive current from 100 to 2000 mA was only 2 nm. We adjusted the LED to ensure that the LED had a peak wavelength of 265 nm at 2000 mA.