Experimental Validation of Fog Models for FSO under Laboratory Controlled Conditions

Transcription

1 213 IEEE 24th International Symposium on Personal, Indoor and Mobile Radio Communications: Fundamentals and PHY Track Experimental Validation of Fog Models for FSO under Laboratory Controlled Conditions M. Ijaz 1, Z. Ghassemlooy 1, H.Le-minh 1, S. Zvanovec 2, J. Perez 1, J.Pesek 3 and O. Fiser 4 1 Optical Communications Research Group, Faculty of Engineering and Environment, Northumbria University, UK. 2 Dep. of Electromagnetic Field, Czech Technical University, Prague, Czech Republic 3 Faculty of Electronics and Informatics, University of Pardubice, Czech Republic 4 Inst. of Atmtospheric Physics, Prague 4, Czech Republic {muhammad.ijaz, z.ghassemlooy}@northumbria.ac.uk Abstract In this paper, we present the experimental results forfree space optics (FSO) communication systems operating at visible and near infrared wavelengths (from.6 m to 1.6 m) under fog conditions. Different empirical fog models are typically used to characterize the fog attenuation of optical beam in FSO systems. A number of empirical fog models are evaluated in order to verify them experimentally and to fill several unmeasured gaps within the entire spectrum from the visible near infrared (NIR) range for light to dense fog conditions. The experimental results in the controlled laboratory fog environment are compared with the selected empirical fog models in order to practically validate their performance over wide range of wavelengths. The results indicate wavelength dependency of the fog attenuation for visibilities higher than 15 m, whereas the cases with shorter visibilities contradict the Kim model. Keywords-component; Free space optics, Fog attenuation, Visibility I. INTRODUCTION FSO is an emerging technology utilizing modulated optical signals usually from visible and NIR wavelengths, capable of transmitting Gigabit Ethernet through atmospheric channel [1]. FSO communications are attracting attention in the last mile access network (LMAN) as an alternative technology for the short range communications (< 1 km) to bridge the last mile optical-access networks (OAN), e.g. building to building and campus to campus. This is due to their cost effectiveness, no need for trenches busy streets, fibre alike bandwidth, implementation in a non-licensed spectrum, relatively lower power consumption and immunity and security compared with other microwave (millimetre wave) RF technologies [2]. However, the performance of an FSO communication is under severe scrutiny on the adverse outdoor atmospheric conditions, particularly fog and turbulence [3, 4]. Fog is the composition of very fine water particles suspended in the air forming a cloud near the ground. The formation of the water particles takes place mainly due to the evaporation of liquid water or by the sublimation of ice, where the state of ice changes into vapours. Principally, fog particles reduce the visibility near the ground and the meteorological definition of fog is when the visibility drops to near 1 km [5]. The presence of dense fog reduces the visibility V due to the scattering and absorption of the optical signal, consequently, minimizes the transmission length of FSO links to.5 km [6]. Due to the rigid and very robust changes in outdoor FSO atmospheric conditions, the fog rather is very dynamic in size and heterogeneous in nature [7]. Generally, due to this fact the characterization of the fog is based on the measured V (km). There are a number of empirical models, which estimates the outdoor fog attenuation based on the measured V from the visible to NIR range of the spectrum. Some of the existing fog models are based on the experimental data [8, 9], while other are obtained using the theoretical considerations [1, 11]. However, the behaviour of the resultant fog attenuation of these models is different comparatively especially in the range of dense fog. Only few attempts have been made to verify the fog model practically using individual wavelengths. However, there is still a lack of the experimental data to verify the fog models practically for the visible to NIR range of the spectrum explicitly. The experimental setup to characterize V and the corresponding fog attenuation for outdoor FSO for the simultaneous visible to NIR range of the spectrum is very challenging. This is due to the long waiting observation time for reoccurrence of dense fog (V <.5 km) and the difficulty in controlling and characterising the fog homogeneously along the length of the FSO link. Hence an indoor atmospheric laboratory chamber is designed so that the atmospheric fog channel can be controlled; similar to previous attempt made for controlled atmospheric turbulence studies [12]. The visibility and fog attenuation can be therefore carried out in homogeneous fog conditions for visible to NIR band of optical spectrum and compared theoretically with the fog models. Here we report the experimental results for the fog attenuation at the visible and NIR spectrum against the measured V. The selected empirical models are used to compare with the experimental data to verify the wavelength dependent fog attenuation. This paper is organised as follows: modelling of the fog attenuation is outlined in Section II, calibration of lab based fog to real outdoor fog (ROF) is presented in Section III whereas the experimental setup is explained in Section IV. In Section VI results and discussion are presented. The conclusions are drawn in Section V. II. MODELLING OF THE FOG ATTENUATION In practice, the link V (i.e. the meteorological visual range) is used to characterize the fog attenuation. Using the Koschmieder law, meteorological V (km) can be expressed in terms of the atmospheric attenuation coefficient βλ and visual threshold T th at.55 μm wavelength and is given as [13]: /13/$ IEEE 19

2 1 log1 ( T ) V = th, (1) β λ where β λ is normally expressed in db/km, and is mathematically defined in [14]. Due to the complexity in quantifying the particle size and distribution, the fog induced attenuation is predicted using empirical fog models [15]. Kruse devised an empirical relationship between V, β λ and wavelength λ as [15]: 1 log1 Tth λ β λ =, (2) V ( km ) λo where T th is the transmission threshold and has been mostly taken as 2% of original optical power (sometimes this threshold is set on 5%), o is the maximum spectrum of the solar band (55 nm) and q is the coefficient related to the particle size distribution in the atmosphere defined by Kruse as: q 1.6 for V > 5 km q = 1.3 for 6 < V < 5 km. (3) 1/ 3.585V for < V < 6 km Since the value of q in (3) was defined from haze particles present in the atmosphere, the estimation of the fog attenuation using the Kruse model is considered to be not accurate enough smaller visibilities. Therefore, in order to accurately predict the fog attenuations, Kim modified the Kruse model with following q values [1]: 1.6 for V > 5 km 1.3 for 6 < < 5 km V q =.16V +.34 for 1 < V < 6 km. (4) V.5 for.5 < V < 1km for V <.5 km Equation 4 incorporating with (2) indicates that the atmospheric attenuation coefficient βλ is wavelength independent for V <.5 km. However, recent experimental data at wavelengths of.83 and 1.55 m [16] shows that there is the wavelength dependency in the dense fog conditions with visibility V <.5 km. In [16] Naboulsi proposed expressions to predict the wavelength dependent fog attenuation coefficient for the convection and advection fogs for wavelengths from.69 to 1.55 μm and visibilities ranging from 5 m to 1 km. The attenuation coefficient for radiation fog is given by: (.11478λ ) 1 ln, ( ) β λ con = (5) V km The attenuation coefficient for advection fog is given by [16]: ( ) 1 ln λ + λ + β λ adv =, (6) ( ) V km Recently, in [17] a wavelength dependent model for fog and haze based on Mie scattering theory has been proposed. The model is applicable for the wavelength range of.2 m < < 2 μm for visibility range of 1 km and is defined as: s( r e ) 1log1 Tth λ β λ =, (7) V( km) λo where, s 2( tanh( p ( w + p )) 1) + ( p ( w + p )) = 1 4 exp 3 5 is a function of effective radius (r e ) of the fog particle distribution, 1/ 2 w= log (r e ) and re = re (.5/ V ). p i are parameters dependent on the selected wavelength interval and are given in [17]. This model is more complex to use but only in cases when particle size distribution and effective radius data of fog particles at given visibility and wavelength are anticipated. However, the summary of selected empirical fog models clearly demonstrates that in spite of a significant number of investigations, these models needs to be explicitly verified in practice, not for a selective or specific wavelength but for the entire spectrum from the visible to NIR range. III. CALIBRATIONS OF LAB BASED FOG TO REAL OUTDOOR FOG In [18], measurements showed that the occurrence of fog starts when the relative humidity H of the real outdoor atmosphere (ROA) approaches 8 %. The density of the resulting fog reaches.5 mg/cm 3 for H > 95%. Thus, under conditions of high water vapour concentration, the water condenses into tiny water droplets of radius 1 2 m in the atmosphere. It is possible to simulate fog in the lab by achieving H close to 95%. Hence, artificial steam fog is produced by generating water based steam to mimic the ROF as shown in Fig. 1. Fog Steamer Artificial fog Fig. 1. The mechanism of artificial fog formation. In addition, to demonstrate the physical similarity of the lab based fog to ROF, the ROF attenuation data from field experiments at Prague published in [19] and Metrological Institute, Czech Republic [2] are compared with Lab based fog attenuation data. The mean attenuation data from Prague for < V < 1 km and Nadeem et al, for V <.6 Km shows a very good agreement with the measured lab data at.83 μm as 2

3 illustrated in Fig. 2. This confirms the physical characteristics of lab based fog resembles to the ROF Mean ROF data, Prague Mean ROF data, Nadeem et al, 21. Lab Data Visibility (Km) Fig. 2. Comparison of measured mean ROF attenuation for two field experiments and the mean lab based fog attenuation data at.83 m. IV. EXPERIMENT SETUP The laboratory based FSO link consists of an optical transmitter end T x and an optical receiver end R x separated by the atmospheric chamber, see Fig 3(a). The controlled channel, represented by the atmospheric chamber has a dimension of Transmitter (Tx) Lens LS1 Halogen Source Laser diode =.55 μm Laser End (T x ) Aerosol control Controlled Atmosphere P T L= 5.5 m, Vol =.495 m 3 Fans (b) (a) Lens OSA Optical Power meter Receiver (Rx) GPIB Data Acquisition System Fig. 3. (a) The experimental set up to measure the fog attenuation and visibility and (b) the laboratory controlled atmospheric chamber and FSO link setup, and (c) the inset, shows the scattering of light due to the fog in the atmospheric chamber cm 3 as shown in Fig. 3(b) and FSO link length of 6 m. The fog was generated using the fog machine (water steam) with 1% humidity to replicate the real outdoor fog in P R Air Outlet Fan Air Outlet Receiver End (R x ) OSA Fog/smoke inlet Scattering of optical signal The Chamber with the controlled amount of fog (c) (c) this study. The experiment was carried out using the continuous LS-1 tungsten halogen light source with a broad spectrum from.36 to 2.5 m and an optical receiver R x using an Anritsu MS91B1 optical spectrum analyzer (OSA) with a spectral response from.6 to 1.75 m. The amount of fog in the atmospheric chamber is controlled to settle down homogeneously along the length of chamber before data acquisition process. An automatic data acquisition (DAQ) system is developed by connecting the OSA to a computer using GPIB bus and LABVIEW control environment. In order to measure the fog effect on wavelength spectrum of μm, the received optical spectrum power P R is measured at R x before and after the injection of the fog into the atmospheric chamber. The normalized transmittance T was calculated from P R with fog to without fog. We measured β λ corresponding to the measured T from light to dense fog condition for the spectrum of μm. The link V was measured simultaneously with the β λ using (1) along the length of the chamber using standard wavelength of.55 μm, see Fig.3 (c). The geometric and the other losses were not taken into account for T x, as P R was measured both before and after the fog at R x to attain the wavelength dependent fog loss. V. EXPERIMENTAL RESULTS The measured fog attenuation (db/km) for the wavelength spectrum of μm for the dense to the light fog conditions (.15 km < V <.9 km) is shown in Fig. 4. The experimental data shows that as the V (i.e. the very dense fog conditions with V <.15 km), the received optical signal was significantly lower than the OSA minimum sensitivity, hence no conclusive measurement could be taken to explore the wavelength dependence. However, the measured optical power clearly demonstrates the fog attenuation dependency on operating wavelength for V >.15 km. For example, fog attenuation decrease from 586 db/km to 512 db/km from visible to NIR spectrum at V =.65 km and from 14 db/km to 97 db/km at V =.24 km Visibility (km) V =.65 km Fig. 4. The measured fog attenuation for the visible to NIR spectrum against the visibility (km). The measured fog attenuation for a very dense fog (V =.48 km) and thick fog (V ~.3 km) against the wavelength is shown in Fig. 5. The measured fog attenuation is notably.8 V =.1 km V =. 24 km V =.496 km V =.78 km

4 higher at the visible range than at the NIR range for V <.5 km as shown in Figs. 5 (a) and 5 (b). The comparison of the selected empirical fog models with the measured attenuation spectrum shows as expected from previous published works that Kruse model underestimates the fog attenuation. The Kim model contradicts the measured data since it suggests the wavelength independency at V <.5 km. The Naboulsi advection and convection models overestimate the measured fog attenuation for NIR range. The Grabner model shows that the attenuation decreases linearly from the visible to NIR range of the spectrum at a much slower sloop since it underestimates in the visible range. This model expresses the best fit over almost whole investigated band, however overestimates the fog attenuation at the NIR spectrum as depicted in Figs. 5 (a) and 5 (b). that, a small discrepancy between indoor measured data from visible to NIR range and outdoor fog model is expected. Moreover, notice that the peak attenuation as depicted in Fig. 5 (b) at measured values for wavelength range of μm is due to strong water absorption, which is consistent to the published results in [21], and has to be underestimated to fair comparison with previous mentioned models. RMSE Data (db) kim Kruse Nab-advNab-convGrabner (a) (a) Kim model Kruse model 21 3 Naboulsi Adv 24 Naboulsi Con 24 Grabner (a) 45 Kim model Kruse model 21 4 Naboulsi Adv 24 Naboulsi Con 24 Grabner (b) Fig. 5. Comparison of the selected empirical models with measured fog attenuation (db/km) from the visible to NIR spectrum, (a) V =.48 km and, (b) V =.3 km. The root mean square error (RMSE) for Kim, Kruse, Naboulsi advection, Naboulsi convection and Grabner models from the measured spectrum of FSO attenuation by fog with V =.48 km and V =.3 km are illustrated in Figs. 6 (a) and 6 (b), consecutively. The RMSE values show that the performance of the Kim and Grabner model is better than other models for water based fog attenuation measurements. Note RMSE Data (db) 1 5 kim Kruse Nab-advNab-convGrabner (b) (b) Fig. 6. Root Mean Square Error of selected empirical fog models from measured spectrum attenuation data, (a) V =.48 km and, (b) V =.3 km. The measured fog attenuation for light fog (V =.783 km) for the visible to NIR range of spectrum is shown in Fig. 7. The comparison of the selected empirical fog models with the measured spectrum attenuation shows that Kruse model underestimates the fog attenuation at V ~.783 km. Naboulsi advection, convection and Grabner models overestimate the measured fog attenuation for the NIR range of the spectrum. However, the measured attenuation spectrum is very close to Kim model at V =.783 km Kim model Kruse model 21 Naboulsi Adv Naboulsi Con 24 Grabner Fig. 7. Comparison of the selected empirical models with measured fog attenuation (db/km) from the visible to NIR spectrum at V =.783 km. 22

5 RMSE Data (db) kim Kruse Nab-adv Nab-convGrabner Fig. 8. Root Mean Square Error of selected empirical fog models from measured attenuation data at V =.783 km. RMSE for Kim, Kruse, Naboulsi advection, Naboulsi convection and Grabner models from the measured spectrum of FSO attenuation by fog with V =.783 km is presented in Fig 8. The RMSE value is around 2 db for Kim model which is better than Grabner model (3.5 db) and than other models ( > 4 db). This validates that Kim model is more realistic to use when V >.5 km. This model does not take into account the wavelength dependency for V <.5 km. However, the experimental data and selected empirical models show that the optical signal attenuation due to fog is wavelength dependent for V <.5 km. This validates that the Kim model need to be revised for V <.5 km to predict the wavelength dependent fog attenuation in particularly in the dense fog conditions. VI. CONCLUSION In this paper, we have experimentally demonstrated the impact of fog on signal attenuation for a wide range of optical spectrum of FSO communications link. We have verified that the fog attenuation is wavelength dependent for the spectrum range of.6 m < < 1.6 μm for V <.5 km. The resultant fog attenuation decreases linearly from the visible towards NIR wavelengths as predicted by Kim model for V >.5 km, however, contradicts the Naboulsi model except for selected wavelengths. The Grabner model produces good estimates over wide visibility ranges but starts to slightly overestimate the experimental data with the visibility increasing over.5 km. This clearly demonstrates that the selected empirical fog models have to be applied with careful consideration and there are not as precise for particular visibility ranges and fog types. Therefore, the composition and physical evolution of real outdoor aerosols such as fog and smoke in terms of particle size, liquid water content, particle surface area and visibility range are important factors to understand its effect on the propagating optical beam. Joint European teams are nowadays focused within the frame of COST actions research on these aspects. ACKNOWLEDGMENT The authors would like to acknowledge the financial support received from the Faculty of Engineering and Environment, Northumbria University, UK and the EU FP7 Cost Actions of IC82 and IC111. REFERENCES [1] E. Ciaramella, Y. Arimoto, G. Contestabile, M. Presi, A. D'Errico, V. Guarino, and M. Matsumoto, "1.28 Terabit/s (32x4 Gbit/s) WDM transmission system for free space optical communications," IEEE Journal on Selected Areas in Communications, vol. 27, pp , 29. [2] A. K. Majumdar, J. C. Ricklin, E. Leitgeb, M. Gebhart, and U. Birnbacher, "Optical networks, last mile access and applications," Free- Space Laser Communications. vol. 2, ed: Springer New York, pp , 28. [3] K. Su, L. Moeller, R. B. Barat, and J. F. Federici, "Experimental comparison of performance degradation from terahertz and infrared wireless links in fog," Journal of Opt. Soc. Am. A, vol. 29, pp , 212. [4] W. Gappmair and M. Flohberger, "Error performance of coded FSO links in turbulent atmosphere modeled by gamma-gamma distributions," IEEE Transactions on Wireless Communications, vol. 8, pp , 29. [5] J. Yang, Z. Niu, C. Shi, D. Liu, and Z. A. Li, "Microphysics of atmospheric aerosols during winter haze/fog events in nanjing," Journal of Environmental Science, vol. 31, pp , 21. [6] Z.Ghaseemlooy, W. Poopola, and S. Rajbhandri, "Optical wireless communications, system and channel modelling with matlab," CRC Press, London, UK, 212. [7] M. S. Awan, E. Leitgeb, M. Loeschnig, F. Nadeem, and C. Capsoni, "Spatial and time variability of fog attenuations for optical wireless links in the troposphere," IEEE 7 th Vehicular Technology Conference Fall (VTC 29-Fall), Alaska,USA, pp. 1-5, 29. [8] K. W. Fischer, M. R. Witiw, and E. Eisenberg, "Optical attenuation in fog at a wavelength of 1.55 micrometers," Atmospheric Research, vol. 87, pp , 28. [9] E. Ferdinandov, K. dimitrov, A. Dandarov, and I. Bakalski, "A general model of the atmospheric scattering in the wavelength interval 3-11nm," Journal of Radio Engineering, vol. 18, pp , 29. [1] I. I. Kim, B. McArthur, and E. Korevaar, "Comparison of laser beam propagation at 785 nm and 155 nm in fog and haze for optical wireless communications," in Proc. SPIE 4214, Boston, MA, USA, 21. [11] M. A. Naboulsi, H. Sizun, and F. d. Fornel "Fog attenuation prediction for optical and infrared waves,," Journal of Optical Engineering, vol. 43, pp , 24. [12] Z. Ghassemlooy, H. Le Minh, S. Rajbhandari, J. Perez, and M. Ijaz, "Performance analysis of Ethernet/Fast-Ethernet free space optical communications in a controlled weak turbulence condition," Lightwave Technology, Journal of, vol. 3, pp , 212. [13] S. S. Muhammad, B. Flecker, E. Leitgeb, and M. Gebhart, "Characterization of fog attenuation in terrestrial free space links," Journal of Optical Engineering, vol. 46, pp , 27. [14] M. Ijaz, Z. Ghassemlooy, S. Rajbhandari, H. Le Minh, J. Perez, and A. Gholami, "Comparison of 83 nm and 155 nm based free space optical communications link under controlled fog conditions," in 8th International Symposium on Communication Systems, Networks & Digital Signal Processing (CSNDSP), Poznan Poland, pp. 1-5, 212. [15] P. W. Kruse, L. D. McGlauchlin, E. B. McQuistan, and. "Elements of infrared technology: Generation, transmission and detection," J. Wiley and sons, [16] M. A. Naboulsi, F. d. Forne, H. Sizun, M. Gebhart, E. Leitgeb, S. S. Muhammad, B. Flecker, and Christoph Chlestil, " and predicted light attenuation in dense coastal upslope Fog at 65, 85 and 95 nm for free space optics applications.," Journal of Optical Engineering, vol. 47, pp , 28. [17] M. Grabner and V. Kvicera, "The wavelength dependent model of extinction in fog and haze for free space optical communication," Journal of Optics Express, vol. 19, pp , 212. [18] M. S. Awan, L. C. Horwath, S. S. Muhammad, E. Leitgeb, F. Nadeem, and M. S. Khan, "Characterization of fog and snow attenuations for free space optical propagation," Journal of Communications, vol. 4, pp , 29. [19] M. Grabner and V. Kvicera, "Fog attenuation dependence on atmospheric visibility at two wavelengths for FSO link planning," in Antennas and Propagation Conference (LAPC), Loughborough, UK, pp , 21. [2] F. Nadeem, T. Javornik, E. Leitgeb, V. Kvicera, and G. Kandus, "Continental fog attenuation empirical relationship from measured visibility data," Radio Engineering, vol. 19, 21. [21] R. N. Clark, "Spectroscopy of rocks and minerals, and principles of spectroscopy," in Manual of remote sensing. vol. 3, ed New York: John Wiley & Sons, Inc., pp. 3-58,