Multipath Propagation Modeling and Measurement in a Clear-Air Environment for LOS Link Design Application

Transcription

1 Multipath Propagation Modeling and Measurement in a Clear-Air Environment for LOS Link Design Application Peter K. Odedina, Member, IEEE and Thomas J. Afullo, Senior Member SAIEE School of Electrical Electronic and Computer Engineering, King George V Avenue, Howard College Campus. University of KwaZulu-Natal, P.O. Box 404, Durban South Africa odedina@ukzn.ac.za, afullot@ukzn.ac.za Abstract - The impairment to radio signal propagation in clearair environment requires accurate prediction method and modeling for terrestrial line of sight links. This is necessary because of the unstable nature of the environment the signal is traversing. Prediction methods based on global radioclimatic models of the ITU-R can currently be made for three significant clear-air propagation effects on terrestrial line of sight links: multipath fading, distortion and depolarization. In addition, such predictions can also be made for multipath fading on very low angle satellite links, and interference between terrestrial and satellite communication systems resulting from duct propagation beyond the horizon. All these predictions explicitly or implicitly use world wide contour maps of refractive index gradient statistics for the lower 00m of the atmosphere. This paper focus on multipath propagation modeling in clear-air environment which can be used for line of sight link design application. The investigation was carried out using clear-air signal level measurement on a terrestrial line of sight link set up between the Howard College and Westville Campuses of the University of KwaZulu Natal, Durban, South Africa for a period of one year in 004. Index Terms Multipath Propagation, Digital Elevation Model (DEM), k-factor T I. INTRODUCTION ECHNIQUES for predicting the deep-fading range of the multipath fading distribution for average worst month have been available for several years []. Most of these techniques were based on empirical fits of Rayleigh-type distributions (i.e. with slopes of 0 db/ decade) to fading data for individual countries. The best known techniques in this regard are those of Moritas [] for Japan, Barnette [3] and Vigants [4] for USA, Pearson [5] and Doble [6] for the United Kingdom, Nadenenko [7] for the former Soviet Union and that of International Radio Consultative Committee (CCIR) [8] for the North-West Europe. The single-frequency or narrow-band prediction equations [] were based on the power-law form originally introduced by Morita and Kakita [9] in 958. They showed the influence of path length on the number of hours containing deep fading or so-called Rayleigh fading. Morita and Kakita fitted the number of measured hours with deep fading in the worst season to path length for 4 GHz links in Japan, but they did not make it clear how to relate the measured time with fade depth. Seven years later, Pearson [5] presented a set of curves for predicting the fade depth exceeded for 0.% of the worst month at 4 GHz in the UK, taking the path length and terrain fade roughness s as predictor variables. Like the model of Morita and Kakita, his model gave a linear relationship between the fade depth expressed in decibels and the logarithmic path length (i.e. the power-law form in probability), but it did not give a linear dependence on terrain roughness. Pearson also assumed a distribution slope of 0 db/decade for fading exceeding 0 db. Morita [] added a dependence on frequency f later by analyzing new data for different frequencies. He used a partial regression technique, fitting the path length d dependence first and the frequency f dependence afterwards. Also he introduced discrete geoclimatic variability by giving geoclimatic factors for three regions: plains, mountains and coast. In Southern Africa, Baker and Palmer proposed a model for the cumulative probability distribution of the k-factor []. While using available data for South Africa and Namibia, they concluded from regression analysis that there are climatic factors that need to be incorporated into the basic model. They concluded that the model would assist in predicting large values of the k-factor that may only be exceeded relatively rarely in the inland summer rainfall areas. Afullo et al also reported on radio refractivity and k-factor studies for Botswana [3]. Using measurements taken over three years ( ), the median value of k was determined to be., while the effective value, k e being 0.7. On the other hand when ducting data were included, they found the median k to be.03, while k e was 0.6. In [] a framework for the modeling of the pdf of k, f(k), was developed and the model determined, based on radiosonde data collected in Botswana for the period It was observed that at height spans 0-500m and 0-00m above ground level (a.g.l), the all-year median value of k, µ k is. and the standard deviation is found to vary between in all months, except in August when the deviation becomes lower at The effective value of k, k e, is found from the analytical expression in [3] to be 0.7 for height span 0-500m a.g.l., while it is 0.6 for the lower height span 0-00m a.g.l.

2 In this paper, we model multipath propagation in a clear-air environment using a line of sight link set up in Durban, South Africa. We use a signal level measurement result of ten months experiment in two of the campuses of University of KwaZulu Natal. Various parameters on the link path were investigated and how this affect the transmitted signal is explained. We start by describing the various radioclimatic parameters that is expected to be on the link path. Next we do an in depth topographical and detail contour mapping of the study area. Finally we do a comprehensive analysis and explanation of some of the plot of the signal level measurement between the transmitter and receiver for the link. II. INVESTIGATION STUDY AREA The investigation study area where the line-of-sight link was set up is in Durban, KwaZulu Natal province of South Africa. Durban is located on the coaster shore of Indian ocean on the geographical coordinate (Latitude o o 9 97' S and Longitude 30 95' E ) the climatic region is coastal savanna [6]. The line-of-sight link was established between the Howard College and the Westville campuses of the University of KwaZulu-Natal, Durban. The transmitter station was setup on the roof of the Science building at the Westville campus on the azimuth angle of o and about 78 m above sea level and the receiver station on the roof of the Electrical Engineering building at Howard College campus on the azimuth angle of o and about 45 m above sea level [7]. The terrestrial link parameters are shown in Table. The expected noise power in the receiver when no signal is transmitted lies between 80.5 to -80. dbm [8]. The power received at the receiver end can be calculated as follows [8]: P r P t FSL + Gr ant Gt ant Losses () 0 dbm35 db dbi +38.6dBi.dB db 4 dbm where P t = Power transmitted (taken as 00mW = 0 dbm); FSL = Free space loss; G r ant Receive antenna gain; G t ant = Transmit antenna gain. III. CLEAR-AIR MULTIPATH PROPAGATION MODELING In order to appropriately model the multipath propagation for the clear-air condition of our study area, there is need to get the correct digital terrain model and the contour mapping for the line-of-sight link of the study area. To this end, we have used the Arcgis software tool to develop the digital elevation model (DEM) map and the contour map of our study area; these are shown in Fig. and Fig.. Information required for detailed link design calculation, for TABLE I TERRESTRIAL LINK PARAMETERS FOR THE LOS SHF SYTEM [7] Parameter Path Length Height of transmitting antenna above the ground Altitude of transmitter station Height of receiving antenna above the ground Altitude of receiver station Carrier frequency Bandwith under investigation Transmitting power Transmitting/receiver antenna gain Transmitting/receiver beam width Free space loss Total cabling and connection losses Clear-air attenuation Receiver bandwidth Description 6.73 km 4 m 78 m 0 m 45 m 9.5 GHz 00 MHz 0 00 mw 38.6 dbi.9 degrees 35 dbm. db db 00 khz GHz the LOS link can be extracted from these plots as explained in [0]. The DEM of the study area is shown in Fig.. DEM can broadly be defined as a digital representation of the continuous variation of elevation over space [0]. Elevation can be any continuous variable that depends on geographic coordinates []. It is also customary to use the term ' DEM ' for what can be called ' gridded DEM ' so that the more general term should then be ' Digital Terrain Model ' (DTM). The digital elevation model is an extremely useful product of a GIS for land evaluation and production of maps []. It can be seen that more detail information on the topographical feature of our study area is revealed by the DEM map (see Fig. ). Our link is surrounded by different geographical features such as road, river, vegetation and different undulating terrain as can be observed from Fig.. All these contribute in various ways to the signal degradation as was observed from the clear-air signal level measurement over 6.73 km, 9.5 GHz link. Though one can get a rough estimate of terrain height distribution of our study area from the DEM, it is still difficult to get the exact height of each point from the DEM. This is why we have produced a contour map of our study area using the same software as shown in Fig.. The contour map has the advantage of showing the exact height for each point in the study area. We have divided our path length of 6.73 km into an interval of km from the transmitter to the receiver as can be seen from the two plots. This is done so that the procedure described in [0] can be implemented for our study area. The procedure described in [0] for detail link design is stated as follows:

3 where hr is the altitude in meter of the receiver antenna, he is the altitude in meter of the transmit antenna and d is the path length in kilometer. From the profile of the terrain along the path, obtain the terrain heights, h, at intervals of km, beginning km from one terminal and ending km to km from the other. Using these heights, carry out a linear regression with the method of least squares to obtain the linear equation of the average profile: (5) h (x) = a 0 + ax where x is the distance along the path from the transmitter. The coefficients a0 and a can be calculated from the relations []: a0 = ( a = Fig. A digital elevation model (DEM) of the study area ) h a x /n (6) xh ( x h)/n x ( x) /n (7) As a first step, determine the geoclimatic factor and the where the summations are over the number, n, of profile height samples. From (5), calculate h (0) and h (d), the heights of the average profile at the ends of the path, and the heights of the antennas above the average path profile: h = h e h (0) (8a) h = h r h (d) (8b) For paths where the point of specular reflection is fairly obvious (such as on paths over water, partially over water, or partially over flat, level terrain), the height above the reflecting surface should be used for h and h. Next, we calculate the average grazing angle φ (mrad), corresponding to a 4/3-earth radius model for refraction (i.e., ae = 8500 km) from Fig. Contour Map of the Study Area h +h = m ( + b ) d magnitude of the path inclination, p (mrad), using the following relation: (9) Where K = dN () m= dn dn dh h 65m (3) p = he d h (0) h () h +h The path inclination is determine as follows: hr d 4a e (h + h ) b (4) m+ cos 3m 3 3 Arcos 3 3m (m+)3 ()

4 In calculation of the coefficients m and ς, the variables a e, d, h and h must be in the same units. The grazing angle φ will be in the desired units of milliradians if h and h are in meters and d in kilometers. If desired, the distances d e and d r from terminals e and r to the point of specular reflection on the average profile can be determined from: and h > h d e = ( ± b)d/ h < h (3a) h > h d r = ( b) d/ h <h (3b) such calculations can be useful in choosing a suitable regression interval on the path profile. Finally, calculate the percentage of time, P, that the fade depth, A(dB), is exceeded in the average worst month using (4) P Kd 3.3 f 0.93 ( ).. 0 A/0 p (4) where the symbols have their usual meaning. It should be noted that the latest revision of the ITU-R recommendation on the above subject has incorporated and embedded in (4) the grazing angle φ and by virtue of this latest revision, (4) can now be written as (see [3]): Exceedance Probability (%) pw Durban_Meth A=30 db Jan Feb Mar Apr May Jun July Aug Sept Oct Nov Dec Months pw Durban_Meth A = 30 db Fig. 3 Percentage of Time that Fade Depth A = 30 db is exceeded in the average worst month in Durban Links for ITU-R methods and (9.5GHz) f hla/0 P Kd ( p ) 0 (5) IV. RESULTS The procedure described above was followed and after useful information have been extracted from the DEM and the contour map of the study area (see Fig. and Fig. ) we are TABLE MULTIPATH MODELING PARAMETERS FOR ITU-R METHOD LINE OF SIGHT LINK DESIGN Distance from HEIGHT Transmitter PROFILE Other Parameters (km) h(x) Value Parameter Value (m) 0 h(0) 78 a h() 80 a h() 50 h r 65 3 h(3) 0 h e 0 4 h(4) 0 h 4 5 h(5) 80 h 0 6 h(6) 80 n h(7) h(d) 45 φ Fig. 4 Clear-air Signal Level Measurement over 6.73km LOS Link over 4 hrs on June able to calculate the required multipath modeling parameters highlighted in () (5) and the results are shown in table, Fig. 3 and Fig. 4 respectively.

5 V. DISCUSSION Multipath propagation modeling has been presented using both clear-air signal level measurement and detailed ITU-R line of sight link design process. The detail ITU-R method was achieved by extracting needed topographical information from the digital elevation model (DEM) map plotted using the Argis software. The DEM map and the contour map (See Fig. and Fig. ) reveal the detailed information about the link characteristic and implementation of ITU-R method two was possible. The different multipath modeling parameters obtained using this process is shown in table. After the multipath modeling parameters have been obtained, the line of sight link implementation was done using the path link parameters stated in table. This helps to calculate the percentage of time that a particular fade depth A is exceeded for the average worst month in Durban as shown in Fig. 3. It should be noted that the fade depth shown here is for A =30 db for all the months for the two ITU-R methods. While the first ITU-R method is the multipath process that does not involve detail link information such as the grazing angle φ, the second ITU-R method incorporate the grazing angle into the calculation as seen above. The first method has been implemented elsewhere [4] and that is why we have made a comparison with the second method (see Fig. 3). Comparing these two ITU-R methods shows that they both have similar pattern distribution for the different months in Durban for the chosen fade depths (see Fig. 3). The worst months in Durban are February and August as shown in Fig. 3 for both methods, this is in agreement with our previous results in [4]. This shows that the link designer need to make adequate plan for the months of February and August in Durban to avoid link outages in these seasonal months. While the pattern distribution is similar for all the months for both methods in Durban at the chosen fade depth, it can be observed that the first ITU-R method underestimate the percentage exceedance probability value for all the months for the chosen fade depth in Durban. This show that the second ITU-R method is an improved method compared to the first ITU-R method. The clear-air signal level measurement plot at the receiver is shown in Fig. 4. The expected signal level at the receiver is -4 dbm as explained in section, but it can be seen that the actual signal levels are 4 db below the expected free space value of 4dBm. Several factors contribute to this; k- factor fading arises due to the fact that the value of k used in the design is.33, while the median value of k-factor for Durban is., with a value of k = 0.5, exceeded 99.9% of the time[4, 5]. The worst month for k-factor fading is February (with value of k 0. exceeded 99.9% of the month), while the month of August has k-factor value exceeded 99.9 % of the time of 0.9. Note also that the median value of k for August is.7 as opposed to the. for February [5]. Thus this type of fading may contribute.5 db over the path (see [4]). On the other hand, due to the rugged, hilly, and non flat nature of the intervening terrain (see Fig. ), multipath fading contribute about db. Water vapour attenuation is another contributor with highest contribution in summer with an average pressure of about 7mb (see Table 3 in [4]), giving an average attenuation of 0.34 db/km, or. db over the 6.7 km path. On the other hand in winter, the average water vapour pressure is about 3mb (see Table in [8]), resulting in attenuation of about 0.3 db per km, and 0.9 db over the 6.7 km path [5]. Thus, water vapour attenuation contributes about db in winter and. db in summer. Due to the coastal nature of Durban, as well as the industries, fog attenuation is also a contributor. At the operating frequency of the link of 9.5GHz, an average attenuation of 0.dB/km is expected, resulting in a value of 0.7 db along the propagation path [6]. This thus account for the difference in the clear-air signal level measurement shown in Fig. 4 from the expected value of 4dBm. VI. CONCLUSION This paper has presented a multipath propagation modeling using digital elevation model (DEM) and contour map to extract useful topographical terrain information. This topographical terrain information made it possible to implement the ITU-R detailed line of sight link design process as described in [0]. This ITU-R detailed LOS link design method is then compared with its initial planning purpose method counterpart. It was discovered that the initial method underestimate the percentage of time that a particular fade depth (A = 30 db) is exceeded in the average worst month in Durban compared to that of the ITU- R detailed method. Hence the ITU-R method two (detailed link method) is considered to produce an improved result compared to its initial planning counterpart. Clear-air signal level measurement plots were also done using the line of sight link set up between Howard college campus and Westville campus of University of KwaZulu-Natal. The plots reveals that the receive signal level at the receiver differs from the expected free space loss value of 4dBm by 4 db. This difference is attributed to various clear-air radioclimatic variables such as; the k-factor value used for the design is different from that of the Durban environment. Also contributing to this difference is the multipath, water vapour, terrain, fog and industry pollution among other things. The results in this paper will be found very useful by radio link designers in South Africa. ACKNOWLEDGMENT The Authors wish to thank the South African Weather Service for availing the radiosonde data used in this presentation.

6 REFERENCES [] T.Tjelta, R.L. Olsen and L.Martin, Systematic development of new multi-variable techniques for predicting the distribution of multipath fading on terrestrial microwave links, IEEE Transactions on Antenna and Propagation, vol. 38, pp , Oct [] K. Morita, Prediction of Rayleigh fading occurrence probability of line-of-sight microwave links, Rev. Elec.Comm.Lab.(Japan), vol. 8, pp.80-8, Nov.-Dec [3] W.T. Barnett, Multipath propagation at 4, 6, and GHz, Bell Syst. Tech. J., vol. 5, pp. 3 36, Feb. 97. [4] A. Vigants, Space diversity engineering, BellSyst.Tech. J., vol. 54, pp.03 4, Jan [5] K.W. Pearson, Method for the prediction of the fading performance of a multisection microwave link, Proc. Inst. Elec. Eng., vol., pp , July 965. [6] J.E. Doble, Prediction of multipath delays and frequency selective fading on digital radio links in the UK, Inst. Elec. Eng Dig., no. 6, 979. [7] L.V. Nadeneko, Calculation of signal stability in line-of- sight radio-relay systems, NIIR Proc., vol. N5, 980. [8] CCIR Report 338-6, Propagation data and prediction methods required for line of-sight radio-relay systems, Report of the CCIR, Annex to vol. V, XVII Plenary Assembly, International Telecommunication Union, Geneva, Switzerland, pp , 990. [9] K. Morita and K. Kakita, Fading in microwave relays, Rep. ECL, NTT, Japan, vol. 6, pp , Sept [0] R.L. Olsen and T.Tjelta, Worldwide Techniques for predicting the multipath fading distribution on terrestrial L.O.S. links: Background and Results of Tests, IEEE Transactions on Antennas and Propagation, vol. 47, pp , Jan.999. [] T.J. Afullo and P.K. Odedina Effective Earth Radius Factor Characterization for Line of Sight Paths in Botswana, Proceedings of IEEE AFRICON 004 Conference, ISBN , pp [] D.C. Baker and A.J. Palmer A model for the Fraction of time Availability of the Effective Earth Radius Factor for Communications Planning in South Africa The Basic Model, Transactions of SAIEE, vol. 93, pp. 7, 00. [3] T.J. Afullo, M.O. Adongo, T.Motsoela and D.F. Molotsi Estimates of Refractivity Gradient and k-factor Ranges for Botswana Transactions of SAIEE, vol. 9, pp. 6, 00. [4] P.K. Odedina and T.J. Afullo Effective Earth Radius Factor (k Factor) Determination and its Application in Southern Africa Proceedings of the Second IASTED International Conference on Antennas, Radar and Wave Propagation, pp. 7, July 9, 005, Banff, Alberta, Canada. [5] T.J. Afullo and P.K. Odedina On the k-factor Distribution and Diffraction Fading for Southern Africa To appear, on Special Edition, SAIEE Transaction, Sept., 006. [6] M.O Fashuyi, P.A Owolawi and T.J Afullo, Rainfall Rate Modelling for LOS Radio Systems in South Africa, Africa Research Journal of the South Africa Institute of Electrical Engineering, Vol. 97 No., March 006, pp.74-8 ISSN No [7] K. Naicker, Rain attenuation modelling for line-ofsight terrestrial links, MSc. Thesis, University of KwaZulu-Natal, Durban, 006. [8] M.O.Fashuyi and T.J. Afullo Rain attenuation prediction and modeling for line-of-sight links on terrestrial paths in South Africa American Geophysical Union: Radio Science Journal Vol. 4, RS5006, doi:0.09/007rs00368, 007. [9] D. M. Pozar, Microwave Engineering, nd ed., chap., John Wiley, New York [0] P.A. Burrough, Principles of geographical information systems for land resources assessment, Oxford University press. xiii, New York, pp [] S.L. Lystad, T.G. Hayton. A.K. Marsh and T.Tjelta, Interpolation of Clear-air Parameters Observed at Non-regular Observation Locations in Proc. URSI Comm. F Open Symp. on Climatic Parameters in Propagation Prediction (CLIMPARA 98), Ottawa, Canada, pp. 5 6, 7 9 April 998. [] N.R. Draper and H. Smith, Applied Regression Analysis. New York: John Wiley and Sons, 98. [3] ITU Radiocommunication Study Group III: Propagation Data and Prediction Methods Required for the Design of Terrestrial Line-of- Sight Systems, ITU-R P.530, [4] P.K. Odedina and T.J. Afullo, Use of Spatial Interpolation Technique for the Determination of the Geoclimatic Factor and Fade Depth Calculation for Southern Africa, Proceedings of IEEE AFRICON conference 007 ISBN: X. IEEE Catalog number: 04CH37590C, September 6 8, 007, Namibia. [5] G.O. Ajayi, S. Feng, S.M. Radicella, and B.M. Reddy (Eds.), Handbook on Radiopropagation Related to Satellite Communications in Tropical and Subtropical Countries, pp. 4, ICTP Press, Trieste, Italy 996. [6] M.P.M Hall, L.W.Barclay, and M.T. Hewitt, Propagation of Radiowaves, chap. 4, IEE press, London, 996. Peter K. Odedina holds a B.Sc. in Physics (Electronics Specialization), from, Federal University of Technology Akure (FUTA), Nigeria. An M.Sc. in Electrical Engineering from University of KwaZulu-Natal Durban, South Africa, and currently pursuing a PhD degree in Electronics Engineering at the same University. He has been an IEEE member for five years. Thomas J. Afullo holds the B.Sc (Hons) Electrical Engineering from University of Nairobi, Kenya, the MSEE from West Virginia University, USA, and the License in Technology and PhD in Electrical Engineering from the Vrije Universiteit Brussel (VUB), Belgium. He has held various positions in industry and university for more then 5 years. He is currently an Associate Professor, Dept. of Electrical Engineering University of KwaZulu-Natal, Durban, South Africa.