Indonesian Journal of Electrical Engineering and Computer Science V ol. 7, No . 1, J uly 2017, pp . 164 169 DOI: 10.11591/ijeecs .v7.i1.pp164-169 164 Estimation of Millimeter W a ve Atten uation Due to Rain Using 2D Video Distr ometer Data in Mala ysia Manhal Alhilali 1 , Jafri Din 1* , Mic hael Sc h ¨ onhuber 2 , and Hong Yin Lam 3 1 Wireless Comm unication Centre , F aculty of Electr ical Engineer ing, Univ ersiti T eknolo gi Mala ysia 81310 Johor Bahr u, Johor , Mala ysia 2 Joanneum Research Gr az, A ustr ia 3 Depar tment of Electr ical Engineer ing T echnology , F aculty of Engineer ing T echnolog y , Univ ersiti T un Hussein Onn Mala ysia 86400 Batu P ahat, Johor , Mala ysia 1 *Corresponding author , e-mail: jafr i@utm.m y Abstract The increasing use of millimeter w a v e frequency bands requires a good understanding of the atmo- spher ic channel. In equator ial regions , r ain pla ys the centr al role in the impair ment of millimeter w a v e prop- agation. Using large amounts of precise data collected b y a tw o-dimensional video distrometer in Mala ysia, the r ain-specific atten uation of both v er tically and hor iz ontally polar iz ed w a v es at the 38 GHz frequency w as computed b y applying the T -matr ix technique . Good ag reement is obser v ed betw een these computations and fit ted po w er la w models from neighbor ing areas , b ut the ITU-R Recommendation P .838-3 significantly underestimates the r ain-specific atten uation. The impor tance of including r aindrop axial r atio in the calcula- tion is re v ealed b y diff erences among the local fitted coefficients f or the v er tically polar iz ed estimations . The po w er la w fit is pro vided, and the measurement is v er ified. K e yw or ds: Millimeter W a v e ,Rain Atten uation, Equator ial Regions , Raindrop Siz e Distr ib uti on Cop yright c 2017 Institute of Ad v anced Engineering and Science . All rights reser ved. 1. Intr oduction As the w or ld increasingly demands the higher bandwidths that millimeter w a v es can pro- vide [1], precise prediction and estimation of their atten uation b y atmospher ic eff ects is cr ucial [2]. In equator ial regions lik e Mala ysia, hea vy precipitation is the dominant f actor that reduces link quality and a v ailability [3, 4]. T o address this issue , a thorough understanding of precipitation drop siz e distr ib ution (DSD), shape , f all v elocity and macro-ph ysical st r ucture is required. The DSD is par ticular ly impor tant , because r aindrops ar e compar ab le in siz e to w a v elengths in the millimeter w a v e r ange; it is also kno wn that h ydrometeors are ob late spheroids with flattened bases , causing the r ain-specific atten uation of millimeter w a v es to depend on their polar ization. Linear ly polar iz ed w a v es are atten uated to a g reater deg ree when the polar izatio n is hor iz ontal r ather than v er tical [5], and depolar ization of the signal occurs when the polar ization is or thog- onal. A tw o-dimensional video distrometer (2D VD) can capture cross-sections of the drops [6], and theref ore allo ws accur ate measurement of these eff ects . Ho w e v er , that accur acy depends on the a v ailability of not only r ain data b ut also models that can estimate the introduced atten u- ation as w ell as atten uation da ta to v alidate these estimates . This ar ticle presents the estimated r ain-specific atten uation of linear ly polar iz ed millimeter w a v es f or terrestr ial links , using a y ear of data collected b y a f our th-gener ation 2D VD installed at the souther nmost city of peninsular Mala ysia [7]. Data gathered b y co-located meteorological station and the older 38 GHz link [8] are used to v alidate the measurements . Receiv ed F ebr uar y 19, 2017; Re vised Ma y 26, 2017; Accepted J une 18, 2017 Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS ISSN: 2502-4752 165 2. Distr ometer Measurements The 2D VD uses tw o line-scanning camer as to capture contour images and f all v elocities of an y par ticles pass through its 10 10 cm sensing area, pro viding inf or mation about shape , axial r atio , and canting angle . The relationship of r aindrop axial r atio to equiv alent diameter has been the f ocus of man y studies [9, 10, 11, 12]. The 2D VD data computed from natur al r aindrops , sho wn in red in Figure 1, ag ree with [11] in tha t the axial r atio lies along the upper bound of the Beard-Chung model [10] and with [12] in that drops are slightly more spher ical. 0 1 2 3 4 5 6 7 8 9 Eq. diameter [mm] 0.4 0.5 0.6 0.7 0.8 0.9 1 Mean axis ratio Measured Mean Axis Ratio BC with upper and lower bounds Figure 1. Relationship betw een natur al r aindrop mean axial r atio and equiv alent diameter , com- pared with the upper and lo w er bounds of the Beard-Chung model [10]. The DSD , N ( D i ) , is the n umber of drops per unit v olume per unit drop diameter , which can be calculated from the drop data recorded b y the 2D VD using N ( D i ) = 1 t D m k X k =1 1 A k v k ; (1) where t is the integ r ation time per io d (in this case , one min ute), m k resemb les the n umber of drops within th e drop siz e inter v al D , and A k and v k are the eff ectiv e area and the measured f all v elocity of the drops , respectiv ely . While , the r ainf all r ate R in mm/h calculated b y dividing the r ain amount computed from t he quotient of the drop’ s v olume and eff ectiv e area b y the corresponding time inter v al using R = 3600 1 t n X i =1 V i A i ; (2) where n represents the total n umber of the visib le drops within the per iod t and V i is the drop v olume . 3. Rain-Specific Atten uation Calculation The r ain-specific atten uation depends on r ain r ate , one-min ute-integ r ated DSD , drops shape , and w a v elength and polar ization of the oper ating frequency . The f orw ard scatter ing ampli- tude can be computed using the T -matr ix method [13], assuming drop shape is an ob late spheroid with an axial r atio as discussed in section 2., atten uation f or hor iz ontal and v er ti c a l ( H and V ) po- lar ization in dB/km can be defined as H ;V = 8 : 686 10 3 41 X i =1 Im( f H ;V ( D i )) N ( D i ) D i ; (3) Estimation of mmW a v e Atten uation Due to Rain Using 2D VD data in Mala ysia (Manhal Alhilali) Evaluation Warning : The document was created with Spire.PDF for Python.
166 ISSN: 2502-4752 where is the w a v elength and f H ( D i ) and f V ( D i ) are f orw ard scatter ing amplitudes f or H and V polar ization, respectiv ely . Figure 2 sho ws an e xample of r ain-specific atten uation time ser ies f or the 38 GHz fre- quency from one e v ent occurred on 29 th of Apr il 2016. At the top of the figure , one-min ute r ain in- tensity calculated from distrometer data is sho wn in mm/h, f ollo w ed b y the calculated r ain-specific atten uation f or H and V polar ization at 38 GHz in the middle of the figure and finally the diff erence betw een the tw o ( H V ) at the bottom of the figure . 06:30 06:45 07:00 07:15 07:30 Apr 29, 2016    0 50 100 150 200 Rainrate[mm/h] 06:30 06:45 07:00 07:15 07:30 Apr 29, 2016    0 20 40 60 Sp. atten. (H-pol) [dB/km] 06:30 06:45 07:00 07:15 07:30 Apr 29, 2016    0 20 40 60 Sp. atten. (V-pol) [dB/km] 06:30 06:45 07:00 07:15 07:30 Time [UTC] Apr 29, 2016    0 2 4 6 Sp. diff. atten.[dB/km] (a) (b) (c) (d) Figure 2. 2D VD measurements of one e v ent at 38 GHz; (a)one-min ute r ainf all intensity , (b) r ain- specific atten uation f or H P olar ization, (c) r ain-specific atten uation f or V polar ization and (d) dif- f erential r ain-specific atten uation. Alter nativ ely , the r ain-specific atten uation can be appro ximated using a po w er la w rela- tionship , which uses coefficients as a function of w a v elength and polar ization to relate the cal- culated r ain-specific atten uation in dB/km to the calculated one-min ute r ain r ate in mm/h, giv en as H ;V = k H ;V R H ;V ; (4) where k and are the po w er la w coefficients and R is the r ain r ate in mm/h. The ITU-R Rec- ommendation P .838-3 [14] also p ro vides a set of k and v alues f or the frequency r ange from 1 - 1000 GHz. Ho w e v er , due to high diff erences in the regional precipitation char acter istics , plenty of authors recommended diff erent local alter nativ es; within the study region, locally der iv ed v alues are presented f or K uala Lumpur [3] and Sin gapore [15]. F or the one y ear of 2D VD measurements IJEECS V ol. 7, No . 1, J uly 2017 : 164 169 Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS ISSN: 2502-4752 167 the one-min ute-integ r ated DSD is used to der iv e the specific atten uation at 38 GHz f or H and V polar ization, and the resulted po w er la w fits presented in T ab le 1 and compared in Figure 3 with ITU-R P .838-3 and the locally der iv ed coefficients v alues . T ab le 1. P o w er la w cur v e-fitting coefficients and their goodness of fits f or the relationship betw een r ain r ate and r ain-specific atten uation at 38 GHz computed from 2D VD data. P olar ization k R 2 Hor iz ontal 0.400 0.935 0.997 V er tical 0.367 0.928 0.987 0 50 100 150 Rainrate [mm/h] 0 10 20 30 40 50 Sp. atten. (H-pol) [dB/km] 2DVD data (38 GHz H) Singapore (38 GHz H) Fitted (38 GHz H) ITU-R P.838-3 (38 GHz H) Kuala Lumpur (38GHz H) 0 50 100 150 Rainrate [mm/h] 0 10 20 30 40 50 Sp. atten. (V-pol) [dB/km] 2DVD data (38 GHz V) Singapore (38 GHz V) Fitted (38 GHz V) ITU-R P.838-3 (38 GHz V) Kuala Lumpur (38GHz V) Figure 3. Specific atten uation from 2D VD in natur al r ain, with der iv ed po w er la w fit, compared with other fits from the ITU-R Recommendation P .838-3 [14] K uala Lumpur data [3] and Singapore data [15]. In gener al, the ITU-R Recommendation P .838-3 r ain-specific atten uation v alues unac- ceptab ly lo w compared with those from the 2D VD data, while a closer ag reement w as achie v ed b y using the locally der iv ed coefficients . The significant diff erences can be accounted f or b y the use of theoretical DSDs f or diff erent regions in the ITU-R Recommendation P .838-3, whereas the locally der iv ed v alues are based on actual DSDs fitting using one-min ute-integ r ated data. Fur- ther more , the drop axial r atio significant necessity in deter mining the r ain-specific atten uation f or H and V polar ization, leads to noticeab le diff erences in V polar ization v alues calculated from the 2D VD data compared with other locally der iv ed v alues . Additionally , there is a diff erence betw een the K uala Lumpur and Singapore v alues , the latter of which are in line with the 2D VD hor iz ontal v alues; w e attr ib ute this to the closer pro ximity of the 2D VD to Singapore than K uala Lumpur . Estimation of mmW a v e Atten uation Due to Rain Using 2D VD data in Mala ysia (Manhal Alhilali) Evaluation Warning : The document was created with Spire.PDF for Python.
168 ISSN: 2502-4752 4. Comparison of Atten uation Measurements The data sho w that the ITU-R Recommendation P .838-3 underestimates the r ain-specific atten uation b y 6-25%. T ab le 2 compares the r ain-specific atten uation v alues estimated f or se- lected r ain r ates using the fitted coefficients from 2D VD data with those from the ITU-R Recom- mendation P .838-3 v alues . T ab le 2. Compar ison of t he 2D VD and ITU-R Recommendation P .838-3 estimations of r ain- specific atten uation dB/km at 38 GHz f or H and V polar ization, presented according to r ain r ate mm/h. Rain r ate (mm/h) P838-3, H 2D VD , H P838-3, V 2D VD , V 1 0.400 0.400 0.384 0.387 5 1.654 1.801 1.522 1.634 10 3.0483 3.444 2.754 3.110 25 6.840 8.112 6.030 7.277 50 12.605 15.510 10.908 13.846 100 23.231 29.65 3 19.733 26.343 T o f ur ther quantify the 2D VD r ain-specific atten uation results and e xamine the long-ter m eff ect of r ain on terrestr ial links , a compar ison of its complementar y cum ulativ e distr ib ution func- tion (CCDF) with the fitting, as w ell as that of the ITU-R Recommendation P .838-3, is presented in Figure 4. 0 10 20 30 40 50  [dB/km] 10 -2 10 -1 10 0 Time Percentage [%]  Fitted 2DVD DSD ITU-R 838-3 Singapore Fitting Figure 4. Compar ison of r ain-specific atten uation CCDFs of H P olar ization at 38 GHz. 5. Conc lusion This ar ticle repor ts data gathered b y a 2D video distro meter in Mala ysia f or the first time . The r ain-specific atten uation of hor iz ontally and v er tically polar iz ed millimeter w a v es has been computed from actual r ain data using T -matr ix calculations . The fitted po w er la w has been com- pared with other fits from neighbor ing areas and the ITU-R Recommendation P .838-3. While the T -matr ix calculations fitted po w er la w are in good ag reement with the local fitted la ws , there is significantly lo w er ag reement with the ITU-R recommendation, which might be due to the high v ar iability of DSD in equator ial regions . This under lines the impor tance of local po w er la w coeffi- cients f or the estimation of r ain-specific atten uation. The r aindrop axial r atio is also impor tant; the discrepancies appeared in other fits of the v er tically polar iz ed data due to the use of older equip- ment and modeled axial r at ios which do not represent the actual shape of r aindrops . The local estimations can theref ore pro vide b etter predictions as the w or ld mo v es to w ards higher-frequency demands . IJEECS V ol. 7, No . 1, J uly 2017 : 164 169 Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS ISSN: 2502-4752 169 Ac kno wledg ement This w or k has been funded b y Ministr y of Education Mala ysia and Univ ersiti T eknologi Mala ysia under HICOE Univ ersity Gr ant V ot. No . 4J221 and par tially de v eloped in the fr ame w or k of the ESA/ESTEC/Contr act 4000106180/12/NL/NR. Ref erences [1] Rappapor t, T . S ., Sun, S ., Ma yzus , R., Zhao , H., Azar , Y ., W ang, K., W ong, G. N., Schulz, J . K., Samimi, M., Gutierrez, F ., “Millimeter w a v e mobile comm unications f or 5G cellular : It will w or k!” IEEE Access , v ol. 1 335–349, 2013. [2] Shrestha, S ., Choi, D .-Y ., “Rain atten uation statistics o v er millimeter w a v e bands in South K orea, J . Atmos . Solar-T errestr ial Ph ys . , v ol. 152-153, no . No v ember 2016 1–10, 2017. [3] Lam, H. Y ., Luini, L., Din, J ., Capsoni, C ., P anagopoulos , A . D ., “In v estigation of r ain atten- uation in equator ial K uala Lumpur, IEEE Antennas Wirel. Propag. Lett. , v ol. 11 1002–1005, 2012. [4] Malinga, S . J ., Ow ola wi, P . A., Afullo , T . J . O ., “Computation of Rain Atten uation through Scatter ing at Micro w a v e and Millimeter Bands in South Afr ica, Prog. Electromagn. Res . PIER , 2013. [5] Atlas , D ., K er k er , M., Hitschf eld, W ., “Scatter ing and atten uation b y non-spher ical atmo- spher ic par ticles , J . Atmos . T err . Ph ys . , v ol. 3, no . 2 108–119, 1953. [6] Sch ¨ onhuber , M., Lammer , G., Randeu, W . L., “The 2D-Video-Distrometer, Precipitation: Adv ances in Measurement, Estimation and Prediction , Michaelides , S . ed., chap . 1, 3–31, Spr inger , 2008. [7] Cuer v o , F ., Sch ¨ onhuber , M., Capsoni, C ., Lam, H. Y ., Jong, S . L., Din, J . B ., Mar tellucci, A., “Ka-band prop agation campaign in Mala ysia - First months of oper ation and site div ersity analysis, 2016 10th Eur . Conf . Antennas Propagation, EuCAP 2016 , 2016. [8] Badron, K., Ismail, A. F ., Din, J ., Tharek, A. R., “Rain induced atten uation studies f or V -band satellite comm unication in tropical region, J . Atmos . Solar-T errestr ial Ph ys . , v ol. 73, no . 5-6 601–610, 2011. [9] Pr uppacher , H. R., Be ard, K. V . “A wind tunnel in v estigation of the inter nal circulation and shape of w ater drops f alling at ter minal v elocity in air , Quar ter ly Jour nal of the Ro y al Mete- orological Society , v ol. 96, no . 408, 247–256, 1970. [10] Beard, K. V ., Chuang, C ., “A Ne w Model f or the Equilibr ium Shape of Raindrops, J . Atmos . Sci. , v ol. 44, no . 11 1509–1524, 1987. [11] Thur ai, M., Br ingi, V . N., “Drop axis r atios from a 2D video disdrometer, J . Atmos . Ocean. T echnol. , v ol. 22, no . 7 966–978, 2005. [12] Marzuki, Randeu, W . L., K ozu, T ., Shimomai, T ., Hashiguchi, H., Sch ¨ onhuber , M., “Rain- drop axis r atios , f all v elocities and siz e distr ib ution o v er Sumatr a from 2D-Video Disdrometer measurement, Atmos . Res . , v ol. 119 23–37, 2013. [13] Mischenk o , M., T r a vis , L., Lacis , A., “Scatter ing, absor ption, and emission of light b y small par ticles, NASA Goddard Inst. Sp . Stud. , 1–128, 2002. [14] ITU-R, “P . 838-3 Specific atten uation model f or r ain f or use in prediction methods, 2005. [15] K umar , L. S ., Lee , Y . H., Ong, J . T ., “T r uncated gamma drop siz e distr ib ution models f or r ain atten uation in Singapore, IEEE T r ans . Antennas Propag. , v ol. 58, no . 4 1325–1335, 2010. Estimation of mmW a v e Atten uation Due to Rain Using 2D VD data in Mala ysia (Manhal Alhilali) Evaluation Warning : The document was created with Spire.PDF for Python.