Over the years, free-space optical (FSO) communication has attracted considerable research interest owing to its high transmission rates via the unbounded and unlicensed bandwidths. Nevertheless, various weather conditions lead to significant deterioration of the FSO link capabilities. In this context, we report on the modelling of the channel attenuation coefficient (β) for a coastal environment and related ambient, considering the effect of coastal air temperature (T), relative humidity (RH) and dew point (TD) by employing a mobile FSO communication system capable of achieving a transmission rate of 1 Gbps at an outdoor distance of 70 m for optical beam wavelengths of 1310 nm and 1550 nm. For further validation of the proposed models, an indoor measurement over a 1.5 m distance utilizing 1310 nm, 1550 nm, and 1064 nm lasers was also performed. The first model provides a general link between T and β, while the second model provides a relation between β, RH as well as TD. By validating our attenuation coefficient model with actual outdoor and indoor experiments, we obtained a scaling parameter x and decaying parameter c values of 19.94, 40.02, 45.82 and 0.03015, 0.04096, 0.0428 for wavelengths of 1550, 1310, 1064 nm, respectively. The proposed models are well validated over the large variation of temperature and humidity over the FSO link in a coastal region and emulated indoor environment.
© 2018 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
Free-space optical (FSO) communication has emerged as a transformative technology that gained a great deal of attention over the past few years [1,2]. By using optical beams from visible and near infrared wavelength regimes, FSO can propagate through them to transmit and receive information through atmospheric channels and transparent media. FSO is capable of providing transmission rates up to 1.2 Tbps utilizing new optical modules . This is due to its license-free bandwidth, cost-effective technology, relatively lower power consumption, and high information security and immunity levels compared to other radio frequency (RF) technologies . Also, FSO was proposed to be integrated with next generation network as an appealing solution that satisfies the quality of service (QoS) demands of new traffic volumes. Examples include cases of bottleneck problems (last mile access) and data congestion in the backhaul network [5,6].
1.2. Related Work
Despite all the advantages of FSO communications, FSO systems can experience severe performance degradation under various atmospheric conditions that are attributed to temperature and humidity variations across the FSO links . In fact, many researchers have focused on the effect of weather conditions on communication performance between FSO transceivers. For instance, the effects of fog on link attenuation were characterized in [8–12] by utilizing the outdoor recorded measurements in different countries at various wavelengthes. Previous works have addressed the effects of atmospheric phenomena over the FSO links using experimental and numerical simulation evaluation based on the visibility distance and the scattering of different particles sizes in the air. Specifically, based on the Beer-Lamber law and piecewise function to identify fog, rain and snow models, they investigate the atmospheric attenuation effects over different wavelengthes; i.e., 670 nm, 785 nm and 1550, as seen in [13–16].
In particular, the impact of fog was investigated via simulation methods in a controlled environment using a theater smoke machine . A controlled fog environment with different modulation schemes, relaying-assisted system, various atmospheric turbulence regimes, and link budget based on quantify the geometrical loss were utilized and evaluated in [18–21]. In a similar way, the work in  focused on the effect of sandstorms in desert areas using a chamber that pumps different kinds of dust across the FSO beam. Recently, Libich et al. combined the effect of atmospheric turbulence, dust particles and ash to assess the performance of Q factor over FSO link at wavelengthes of 670 nm and 830 nm . In addition,  studied the availability and link budget over FSO links under different weather conditions and variable distance between the FSO transceivers. Also, some works have experimentally investigated the optical attenuation coefficient in Graz and in La Turbie near Nice, France. They concluded that the FSO link attenuation can reach 120 dB/km in a mild foggy situation, and it can exceed 480 dB/km under severe foggy conditions during the winter season [25,26]. Therefore, as a technologically beneficial way to increase the FSO link availability and reliability when adverse weather conditions occur, hybrid systems have been proposed with a secondary link to enhance the main FSO channel. As an example, researchers in  estimated the availability of a hybrid system consisting of an FSO as a primary channel and 40 GHz as an alternative channel. The hybrid system reached 99.92%, while the availability of only FSO link reached 96.8%.
There is a variety of factors that can degrade FSO channels depending on the local environment. As a result, there is a substantial need to characterize the reliability of FSO channels under several weather conditions. In this work, we examine the effect of weather conditions over FSO channels in a coastal environment where the temperature and humidity are widely varying. The proposed analysis illustrates the suitability of FSO systems within the variation of temperature and humidity conditions. For this purpose, we derived two mathematical attenuation coefficient models based on our measurements for different wavelengths at 1550 nm, 1310 nm and 1064 nm that explain the relation between temperature, humidity and attenuation coefficient of optical links. Moreover, we experimentally validated the derived analysis through outdoor measurements as well as indoor chamber that emulated the coastal weather conditions.
The structure of this paper is arranged as follows. A theoretical analysis of meteorological definitions and attenuation coefficient is derived in Section 2. In section 3, we detail the setup of the experiment. The outdoor and indoor experimental measurements are illustrated and compared to the derived analysis and indoor setup in Section 4, while we elaborate on the impact and significance of our findings and future work in Section 5.
2. Dependence of link attenuation with respect to temperature and humidity
2.1. Meteorological analysis
Humidity is a quantitative measure of the amount of water vapor in air. Water vapor refers to the gas form of water, which is not detectable by the human eye, as illustrated in Fig. 1. At dew point temperature (TD), the water vapor will be saturated in cooled-down air. In this state, a group of water vapors will eventually condense into liquid dew. One measure of humidity is the relative humidity (RH), which is defined as the ratio of actual vapor density to the maximum saturated density, subject to variation in current air temperature (T). As the RH ratio approaches unity, the air will be completely mixed and saturated with water vapor and consequently rainy conditions will take place. RH can be defined as30], we approximate K = 0.97. More accurate representation from  that expresses the RH with respect to TD and T is given by
All three proposed models show an inverse relation between RH and T, as we will explain in Section 3.
2.2. FSO attenuation coefficient dependence with respect to temperature
A thorough search of the relevant literature yielded no previous work characterizing the effects of temperature variations in a coastal environment. Based on [24, 30], let us assume that β is the channel attenuation coefficient in 1/km. The attenuation coefficient is always inversely proportional to the temperature at that time. Therefore, the relation between T and β can be modeled as(5) within the temperature range (0 to T) leads to
2.3. FSO attenuation coefficient dependence with respect to relative humidity
Atmospheric moisture absorbs considerable portions of the transmitted optical power, depending on the density of water vapor in the atmosphere. To derive a model that clarifies the relation between the FSO link attenuation coefficient and the relative humidity, we substitute (2), (4), and (3) into (7), giving the following expressions
In Section 3, we evaluate the most accurate model for a coastal environment by utilizing the root mean square error (RMSE) and the goodness of fit R2 based on our measured T, RH and β.
3. Experimental setup
In this work, a complete mobile FSO system has been designed and integrated to measure FSO link attenuation coefficient in outdoor and indoor experiments. This system consists of the followings units:
- Two portable carts integrated with breaking system to make sure that the FSO transceivers are stable.
- Four Koruza transceivers  operating at two wavelengthes (i.e. 1310 nm and 1550 nm) to transmit and receive simultaneously. The main system parameters are summarized in Table 1. We note that the attenuation coefficient is the inherent property of FSO channel itself and it remains identical with different system configurations.
- Two wireless modems with data subscriber identification module (SIM) cards, to upload the measured data automatically at each cart via internet to any server or a storage cloud.
- An accurate weather station (WS-2095) to measure the local weather measurements (T range between −40°C to 65°C with accuracy ±0.03°C, RH range between 1% to 99% with accuracy ±5%).
- A deep cycle battery or solar panel system to provide the power required for system operation.
- Breadboards and mounts to fix the FSO units and to align optical beam.
A detailed description of the system and its integration is shown in Fig. 2(a). The mobile FSO system was tested in outdoor and indoor environments.
3.1. Outdoor experimental setup
The designed mobile FSO system has been deployed at a transmission distance of 70 m in a coastal environment. The typical weather conditions at the King Abdullah University of Science and Technology (KAUST) are warm and wet with minimum variations in temperature levels as it is located near the Red Sea (Thuwal, Saudi Arabia) in Figs. 2(b) and 2(c). To evaluate the proposed mobile FSO system in outdoor environment, we first deployed it for a short period of August 2016 [32,33]. Then, we observed the link quality and weather information during a substantial period of the year 2017 (i.e. January to September), which is inclusive of the highly varying humidity and temperature conditions from 21th of June until 22th of September, 2017.
The fluctuation of links quality (1310 nm and 1550 nm) was monitored in the outdoor cases using the Koruza system. The weather information (T, RH, and TD) were recorded using a WS-2095 weather station, which we installed next to the FSO transceivers. As shown in Fig. 2(a), the received power was uploaded to a predefined server via the wireless modem as T, RH, and TD values were stored in the weather station data logger. The values in the FSO system and weather station data logs were updated every ten minutes. Finally, based on the outdoor measured data, we estimated the parameters x and c in (7) at unit-operating wavelengths, 1310 nm and 1550 nm.
3.2. Indoor experimental setup
As a validation of the outdoor measurements behavior, we designed a controllable climate chamber with the size of 1.5 × 0.35 × 0.45 m3 and 5 mm glass thickness in Fig. 3 to emulate the temperature and humidity of a coastal environment. We planed to make the chamber portable for the following reasons:
- To facilitate the transportation and storage processes.
- To make it compatible with our mobile FSO system and upgradable for use in different locations.
- To make it applicable for utilization for various atmospheric effects (dust, rain, smoke, etc.) and various wavelengthes (the glass thickness is 5 mm which was calibrated when we calculated the measured attenuation coefficients).
The proposed chamber dimensions have been chosen and emphasized to be portable over movable cart. We validated the accuracy of indoor chamber experiment in comparison with proposed outdoor model by emulating the real outdoor weather measurements in indoor chamber. Also, the glass thickness is 5 mm which was calibrated when we calculate the measured attenuation coefficients. It was equipped with a digital control unit that kept the desired temperature and humidity uniform inside the chamber using moisture and heat generators utilizing precise humidity and temperature sensors. The digital control unit adjusted three fans for ventilation and one fan inside the chamber to keep the temperature and humidity stable within the setup. The utilized control unit has a spare input and output ports for extra sensors and peripherals to characterize more atmospheric effects. In addition, by utilizing this chamber, we were able to obtain the parameters x and c of (7) as we discussed in section 2.2. Moreover, this chamber allowed us to evaluate the optical transmission capabilities of additional wavelengths, e.g., 1064 nm (Changchun Optoelectronics Industries, model MIL-III-1064 diode-pumped solid-state laser, and measured using a photodetector from [Newport, model 818–IR] equipped with an optical density 3 [OD3] attenuator [Newport, model 883–IR], and a power meter [Newport, model 2936–C]), which cannot be measured by Koruza units.
4. Experimental results and analysis
To carry out our investigation and validate our analysis, we evaluated the attenuation models using outdoor measurements and compared them to the results obtained in the indoor climate chamber. In this context, we show the relation between the outdoor atmospheric attenuation coefficient (1/km) and the weather station data. The attenuation coefficient is computed with fitted parameters x and c by measuring the variation between the transmitted power and the instantaneous received power in decibels (μdB) over the transmission distance (km). To obtain an accurate fitted attenuation coefficient model, we used MATLAB curve fitting tool box because it extracted RMSE and R2 values that can assist in the assessment of how well the regression works between the measured data and the proposed empirical model. For clarity and to avoid crowded data throughout the Figs., we show the relation between T, RH, and β over a part of the sampled time.
Before we validated the attenuation coefficient models from the measurements in FSO links, we first showed the meteorological relation between RH and T from the measured data in a weather station. As illustrated in , the absorption by water, CO2 and ozone molecules affect the FSO beam propagation at wavelength between 400 nm and 2400 nm. In this work, we experimentally quantify the absorption effects at wavelengths within the atmospheric transmittance window addressed in  by focusing on the variations in changes in T and RH in the air. Fig. 4 demonstrates an inverse behavior between T and RH explained in section 2.1.
In this work, we focus on the atmospheric effects of RH, TD and T on homogeneous link. In specific, during the measurements some accidental impairment occurred under the several circumstances such as birds standing over the system, a strong sand storm, construction and digging operation nearby the deployed system, cleaning the dust and sands over the buildings, sudden misalignment during the day hours due to the thermal expansions between the building, etc. So, we excluded some outlier data due to the aforementioned occasional events by observing the link on a daily basis. Moreover, we were averaged over successive samples over 5 minutes in order to mitigate the potential effect of pointing error or atmospheric turbulence as well.
4.1. Outdoor experimental results and analysis
- where, Pr is the received power (μW),
- Pt is the transmitted power (μW),
- a1 is the diameter transmitter aperture (m),
- a2 is the diameter receiver aperture (m),
- L is the link distance (km),
- θ is the beam divergence (mrad),
- β is the attenuation coefficient (1/km).
The total of the received power is calculated as the total of power transmitted and the collection aperture area. It is approximately inversely proportional to the square of the beam divergence and the square of the link range.
The received power for both transceivers Koruza units (1550 nm – 1310 nm) was plotted in Fig. 5. As shown in this Fig. 1550 nm wavelength has less sensitivity to the weather conditions, where the transmitted power for both 1550 nm and 1310 nm wavelengthes are similar. Based on Beer-Lambert law:22](1). We can calculate β by subtracting the received power from the transmitted power in dB and divide the result by the distance in km multiplied by the constant 4.3429.
The relationship between the recorded T and the β is represented in Fig. 6; in this case, β changes inversely with T as shown in Fig. 7. Based on the measurements of β and T, we applied a curve-fitting method on the 1550 nm and 1310 nm measurements and found the exponential attenuation models that are matching the derived Eq. (7). Table 2 shows the values of x and c, as well as the fitting accuracy parameters (R2 and RMSE).
Fig. 7 displays β values obtained from the curve-fitting models alongside with the measured values. We can see that the exponentially regressed attenuation model can fit well with the measured β values at both wavelengths. Moreover, it implies that, at the lower wavelength (1310 nm), the FSO link will attenuate more rapidly with increasing transmission distance under the same temperature condition.
Regarding the effect of RH, we discern from Fig. 8 that the measured β values are proportionally varying with the changes in RH. Hence, β increases as RH increases.
To derive an attenuation model in relation to RH, we compared β values from (8), (9), and (10) with the measured β values of 1310 nm and 1550 nm as shown in Fig. 9. β values are calculated by substituting the measured TD and RH into three attenuation models. As we can see in Fig. 9 and Table 3, model 3 in (10) is the most accurate model corresponding to the relationship between RH and β at both wavelengthes.
4.2. Indoor experimental results and analysis
For the indoor case, we kept the inverse relation T and RH in the climate chamber as in Fig. 4 observed by outdoor weather station measurements. The measurements for each point has been averaged for 30 seconds. Fig. 10 shows the behavior of measured and fitted β values at the wavelengths of 1064 nm, 1310 nm and 1550 nm as T changes. In addition, we plotted the outdoor derived models for 1310 nm and 1550 nm to confirm that both experiments are cross-checking. We can observe that the values extracted from the outdoor models are reasonably close to the indoor measurements. Therefore, we conjecture that the derived outdoor models are accurate. Moreover, we highlight that we could conveniently derive the attenuation model at 1064 nm wavelength in laboratory level which is not necessarily given that we are operating a mobile FSO unit. Finally, as can be seen in Table 4, the estimated values of x and c are adequately close to the values in Table 2 with good fitting parameters.
In a similar manner, to derive an attenuation model corresponding to RH values for indoor measurements, we plotted data from (8), (9), and (10) and compared them with the measured β values of 1550 nm, 1310 nm and 1064 nm beam wavelengths, respectively. Fig. 11 and Table 5 show that model 3 is the most accurate model for the three wavelengthes. We can conclude an excellent matching with the measured data.
In summary, we proposed and validated two mathematical attenuation coefficient models that pinpoint the relationship between T, RH, and β over FSO channels in coastal regions. We designed and implemented a mobile FSO system based on low-cost and open-source transceivers. This system can be deployed and utilized across various environments, and exhibited sustainable mobile FSO channels at a transmission distance of 70 m. Both outdoor and indoor measurements are utilized to determine the scaling and decaying parameters in attenuation coefficient models and validated their goodness of fit. As a result, this study elucidates how humidity and temperature adversely affect the FSO links. The proposed attenuation coefficient models of temperature and relative humidity have been validated with scaling parameter x values of 19.94, 40.02 and 45.82, and decaying parameter c values of 0.03015, 0.04096, and 0.0428, corresponding to different beam wavelengths, i.e., 1550, 1310, and 1064 nm. We found that 1550 nm is the least sensitive beam wavelength affected by humidity and temperature. Our work paves the way for improving indoor and outdoor systems to make them sustainable and reliable in extreme environments through the investigation of diverse models to present the effects of weather conditions, such as pressure, rain, and wind over optical links with different wavelengths.
King Abdullah University of Science and Technology (KAUST) baseline funding (BAS/1/1614-01-01); King Fahd University of Petroleum and Minerals (KFUPM) Initiative (REP/1/2878).
References and links
1. C. Shen, Y. Guo, H. Oubei, T.-K. Ng, G. Liu, K.-H. Park, K.-T. Ho, M.-S. Alouini, and B. S. Ooi, “20-meter underwater wireless optical communication link with 1.5 Gbps data rate,” Opt. Express 22(24), 25502–25509 (2016). [CrossRef]
2. M. Khalighi and M. Uysal, “Survey on free space optical communication: a communication theory perspective,” IEEE Commun. Surveys and Tutorials 16(4), 2231–2258 (2014). [CrossRef]
3. H. Hogan, “Data demands drive free-space optics,” Photonics Spectra 2, 38–42 (2013).
4. L. Andrews, R. Phillips, and C. Hopen, Laser Beam Scintillation with Applications (Academic, 2001). [CrossRef]
5. S. Al-Gailani, A. Mohammad, and R. Shaddad, “Enhancement of free space optical link in heavy rain attenuation using multiple beam concept,” Optik-International Journal for Light and Electron Optics 124(21), 4798–4801 (2013). [CrossRef]
6. C. C. Davis, I. I. Smolyaninov, and S. D. Milner, “Flexible optical wireless links and networks,” IEEE Commun. Mag. 41(3), 51–57 (2003). [CrossRef]
7. H. Henniger and O. Wilfert, “An introduction to free-space optical communications,” Radio Eng. Mag. 19(2), 203–212 (2010).
8. M. Esmail, H. Fathallah, and M.-S. Alouini, “Outdoor FSO communications under fog: attenuation modeling and performance evaluation,” IEEE Photon. J. 8(4), 2894–2903 (2016). [CrossRef]
9. M. S. Khan, M. Grabner, S. S. Muhammad, M. S. Awan, E. Leitgeb, V. Kvicera, and R. Nebuloni, “Empirical relations for optical attenuation prediction from liquid water content of fog,” Radio engineering 21, 911–916 (2012).
10. M. Al Naboulsi, F. De-Fornel, H. Sizun, M. Gebhart, E. Leitgeb, S. Sheikh Muhammad, B. Flecker, and C. Chlestil, “Measured and predicted light attenuation in dense coastal upslope fog at 650, 850, and 950 nm for free-space optics applications,” Opt. Eng. 47(3), 1–14 (2008). [CrossRef]
11. A. Mostafa and S. Hranilovic, “Channel measurement and Markov modeling of an urban free-space optical link,” IEEE/OSA J. of Optical Commun. Net. 4(10), 836–846 (2012). [CrossRef]
12. A. El-Wakeel, N. Mohammed, and M. Aly, “Free space optical communications system performance under atmospheric scattering and turbulence for 850 and 1550nm operation,” Appl. Opt. 55(26), 7276–7286 (2016). [CrossRef] [PubMed]
13. I. I. Kim, B. McArthur, and E. J. Korevaar, “Comparison of laser beam propagation at 785 nm and 1550 nm in fog and haze for optical wireless communications,” Proc. SPIE 4214, 26–37 (2001). [CrossRef]
14. M. Grabner and V. Kvicera, “Multiple scattering in rain and fog on free-space optical links,” J. Lightwave Tech. 32(3), 513–520 (2014). [CrossRef]
15. M. A. Naboulsi, H. Sizun, and F. de Fornel, “Fog attenuation prediction for optical and infrared waves,” Opt. Eng. 42, 319–329 (2004). [CrossRef]
16. M. Grabner and V. Kvicera, “Multiple Scattering in Rain and Fog on Free-Space Optical Links,” J. Lightwave Techn. 32(3), 513–520 (2014). [CrossRef]
17. L. Mustafa, E. Cerncic, and B. Thomsen, “FSO artificial low-cost fog attenuation experiment design,” in Proc. Int. Symp. Commun. Systems, Netw. and Digital Sig. Process. (CSNDSP) (Prague, 2016), pp. 978–983.
18. J. Perez, Z. Ghassemlooy, S. Rajbhandari, M. Ijaz, and H. Le Minh, “Ethernet FSO communications link performance study under a controlled fog environment,” IEEE Commun. Lett. 16(3), 408–410 (2012). [CrossRef]
19. J. Libich, M. Komanec, S. Zvanovec, P. Pesek, W. Popoola, and Z. Ghassemlooy, “Experimental verification of an all-optical dual-hop 10 Gbit/s free-space optics link under turbulence regimes,” Opt. Express 40(3), 391–394 (2015).
20. N. Nor, Z. Ghassemlooy, J. Bohata, P. Saxena, M. Komanec, S. Zvanovec, M. Bhatnagar, and M.-A. Khalighi, “Experimental investigation of all-optical relay-assisted 10 Gb/s FSO link over the atmospheric turbulence channel,” J. Lightwave Thech. 35(1), 45–53 (2017). [CrossRef]
21. B. Kebapci, F. Miramirkhani, H. Nouri, and M. Uysal, “”A custom-design atmospheric channel emulator for the performance evaluation of free space optical communication systems, ”International Conference on Transparent Optical Networks (ICTON)(Girona, 2017), pp. 1–5.
22. M. Esmail, H. Fathallah, and M.-S. Alouini, “An experimental study of FSO link performance in desert environment,” IEEE Commun. Lett. 20(9), 1888–1891 (2016). [CrossRef]
23. J. Libich, J. Perez, S. Zvanovec, Z. Ghassemlooy, R. Nebuloni, and C. Capsoni, “Combined effect of turbulence and aerosol on free-space optical links,” Appl. Opt. 56(2), 336–341 (2017). [CrossRef] [PubMed]
24. A. Prokes, “Atmospheric effects on availability of free space optics systems,” Opt. Eng. 48(6), 066001 (2009). [CrossRef]
25. M. S. Khan, S. S. Muhammad, M. S. Awan, V. Kvicera, M. Grabner, and E. Leitgeb, “Further results on fog modeling for terrestrial free-space optical links,” Opt. Eng. 51(3), 1–9 (2012). [CrossRef]
26. M. S. Sheikh, B. Flecker, E. Leitgeb, and M. Gebhart, “Characterization of fog attenuation in terrestrial free space optical links,” Opt. Eng. 46(6), 066001 (2007). [CrossRef]
27. E. Leitgeb, M. Gebhart, U. Birnbacher, W. Kogler, and P. Schrotter, “High availability of hybrid wireless networks,” Proc. SPIE 5465, 238–249 (2004). [CrossRef]
28. M. G. Lawrence, “The relationship between relative humidity and the dewpoint temperature in moist air: a simple conversion and applications,” American Metrological Sociaty 86(2), 225–233 (2005). [CrossRef]
29. O. A. Alduchov and R. E. Eskridge, “Improved Magnus form approximation of saturation vapor pressure,” Appl. Meteor. 35601–609 (1996). [CrossRef]
30. S. A. Al-Gailani, A. B. Mohammad, M. S. Islam, U. U. Sheikh, and R. Q. Shaddad, “Tropical temperature and humidity modeling for free space optical link,” J. Opt. 45(1), 87–91 (2016). [CrossRef]
31. L. Mustafa, “Light speed networking,” http://www.koruza.net/.
32. W. G. Alheadary, Y. Guo, E. Stegenburgs, K.-H. Park, T. K. Ng, B. S. Ooi, and M.-S. Alouini, “Design and deployment of mobile FSO communication system,” in Proc. of “CLEO: 2017,” p. AW4B.7.
33. W. G. Alheadary, K.-H. Park, B. S. Ooi, and M.-S. Alouini, “Free-space optical channel characterization in a coastal environment,” Journal of Communications and Information Networks 2(4), 100–106 (2017). [CrossRef]
34. A. G. Alkholidi and K. S. Altowij, “Climate effects on performance of free space optical communication systems in Yemen,” Front. Optoelectron. 7(1), 91–101 (2014). [CrossRef]