Radio Wave Scattering from Lampposts in Microcell Urban Mobile Propagation Channel

Transcription

1 14 CTI TRANSACTIONS ON LCTRICAL NG., LCTRONICS, AND COMMUNICATIONS VOL.7, NO.1 February 29 Rado Wave Scatterng from Lampposts n Mcrocell Urban Moble Propagaton Channel Mr Ghorash 1, Jun-ch Takada 2, and Tetsuro Ima 3, Non-members ABSTRACT The rado wave scatterng from lampposts n urban areas s analyzed. The lamppost s modeled as a fnte-length conductng cylnder and the approxmate theoretcal values of ts bstatc radar cross secton (RCS) are compared to those expermental values obtaned from a propagaton channel measurement campagn n two urban envronments. In the theoretcal dervaton t s assumed that two waves, drect and ground-reflected, are ncdent to the lamppost, whereas only drect scatterng s assumed due to the drectve recever (Rx) antenna. The CDF of the theoretcal RCS of the cylnder and those of the lamppost derved from measurement data exhbt a close agreement. Keywords: Lamppost, radar cross secton, urban propagaton channel. 1. INTRODUCTION Conventonal propagaton predcton tools, such as ray-tracng algorthms, only account for specular reflectons from buldng wall surfaces and dffractons from buldng corners and roof-top edges. But ths s an oversmplfcaton of the complex archtecture of the propagaton channel n the urban areas. In fact, not only approxmatng the nteracton of a buldng wall n the urban areas to a specular reflecton s too optmstc for cellular operatng frequences, but also there are many nteractng objects other than buldngs n the wreless channel n these envronments wth a collectve sgnfcant mpact to the recevng sgnal [1]. To mprove the accuracy of propagaton predcton tools, several researches were accomplshed to nclude the buldng rough surface scatterngs [2],[3],[4]. Moreover, the scatterng from trees was addressed n [5],[6]. However, the scatterng effect of objects such as lampposts, traffc lghts and Manuscrpt receved on January 3, 28 ; revsed on June 13, The author s wth The Center for Research and Development of ducatonal Technology Tokyo Insttute of Technology W9-18 Ookayama, Meguro-ku, Tokyo , Japan, -mal: 2 The author s wth Internatonal Development ngneerng Department Tokyo Insttute of Technology W9-18 Ookayama, Meguro-ku, Tokyo , Japan, -mal: 3 The author s wth R&D Center, NTT DOCOMO, INC., 3-5 Hkarno-oka, Yokosuka-sh, Kanagawa , Japan, -mal: sgnboards have not been analyzed or modeled carefully even though ther contrbuton to the channel are consderable [1]. In ths paper we analyze the scatterng effect of the lamppost n the urban envronments by obtanng ts radar cross secton (RCS) expermentally and analytcally. For ths purpose we use the experment results of a measurement campagn n Yokohama, Japan n a mcrocell scenaro [7]. Ths s done through dervng RCS of the lampposts from measurements n real scenaros n comparson to the analytcal results. Secton 2. provdes the analytcal modelng of the lamppost as a conductng cylnder to get ts RCS values. In secton 3. coherent and ncoherent two-ray ncdence models are ntroduced to take the measured RCS varatons nto account. In secton 4. we ntroduce the measurement set up and values, whereas secton 5. ntroduces the smulaton and ts predctons. Secton 6. provdes dscussons and comparsons and fnally secton 7. s the concluson. 2. LAMPPOST RCS MODL The defnton of RCS s as follows [8]: For a gven scatterng object, upon whch a plane wave s ncdent, that porton of scatterng cross secton correspondng to a specfed polarzaton component of the scattered wave. The RCS s usually shown by σ and can be formulated as [9]: L σ = ncdent wave x 2a θ s 2 lm r r 4πr2 r 2 (1) z φ Fg.1: The ncdent and scatterng parallel electromagnetc wave to a long but fnte-length conductor cylnder. θ s s y

2 Rado Wave Scatterng from Lampposts n Mcrocell Urban Moble Propagaton Channel 15 where s 2 s the power per unt area n scattered wave at the recever (Rx) antenna whch s n the polarzaton of the Rx antenna, 2 s the power per unt area n the wave ncdent on the scatterer and r r s the scatterer to Rx separaton. To get the analytcal RCS of the lamppost we approxmate t to a long but fnte-length conductor cylnder. The approxmated theoretcal value of the RCS for a perfect conductor fnte-length cylnder of crcular cross secton can be derved by calculatng the nduced current on the surface of the cylnder. It s formulated as a functon of the ncdent and scatterng angles [1]: σ = 4L2 sn 2 { θ s sn[ kl 2 (cos θ } s + cos θ )] 2 π sn 2 kl θ 2 (cos θ s + cos θ ) ( 1) n e jn φ J n(ka sn θ s ) H n (2) (2) (ka sn θ ) n= where L and a are the cylnder length and radus respectvely wth ts axs of symmetry along the z axs. Moreover θ and θ s are the ncdent and scatterng waves co-elevatons and φ s the azmuth dfference between the ncdent and scatterng waves. J n and represent Bessel functon of the frst knd and Hankel functon of the second knd respectvely. Fgures 2 and 3 show the cylnder RCS varatons as a functon of coelevaton θ and azmuth dfference φ respectvely, for whch L = 6.5 m, a =.8 m, and θ s = 9. It s observed that the cylnder RCS dmnshes rapdly as the ncdent coelevaton changes from 9 (horzontal ncdence), whereas RCS varatons due to the azmuth dfference s small except for the backscatterng zone. Consequently, n the next secton we wll address the lamppost RCS varatons n the urban areas by ntroducng a two ray model whch mposes the coelevaton varatons n the supermposed ncdent wave. H (2) n 3. TWO-RAY INCIDNC MODL For the ncdent wave to the lamppost we use the two-ray model [11], [12]. That s we consder the d- RCS [dbm 2 ] θ Fg.2: The RCS of a conductor cylnder as a functon of ncdent wave coelevaton θ, for L = 6.5 m, a =.8 m, θ s = 9 and φ =. RCS [dbm 2 ] φ Fg.3: The RCS of a conductor cylnder as a functon of ncdent and scatterng waves azmuth dfference φ, for L = 6.5 m, a =.8 m, θ = 9 and θ s = 9. Transmtter Antenna o 9 r d r g g d s Fg.4: Two ncdent paths transmtted by a dpole half wavelength antenna at the cylnder. rect ncdent wave from the transmtter (Tx) antenna wth the coelevaton of θ d = 9 and the groundreflected ncdent wave wth a coelevaton of as s llustrated n Fg. 4. Here we only consder the drect scatterng wave, more dscussons on ths wll be provded n secton Coherent waves When the two ncdent waves coherently mpnge at the lamppost, the composed ncdent angle s gven by the superposton of two propagaton vectors of the drect ncdent wave and the ground-reflected wave. As the Tx antenna s a half wavelength vertcally nstalled dpole, the transmtted sgnal s vertcally polarzed (θ component n the sphercal coordnates) and there s not any radaton along horzontal polarzaton vector (φ component). However, due to the grazng angle of the ground-reflected wave, the electromagnetc feld at the reflecton pont ncludes a parallel (to the ground) component. Havng the Fresnel parallel R and vertcal R coeffcents for the ground, the ground-reflected electromagnetc feld at the ncdent pont can be calculated as: g = h θ ( ) R e jkr g g = h θ ( ) R e jkr g r g sn( 9) (3) r g cos( 9) (4)

3 16 CTI TRANSACTIONS ON LCTRICAL NG., LCTRONICS, AND COMMUNICATIONS VOL.7, NO.1 February 29 n whch s the equvalent electrc feld radated by the sotropc antenna, r g s the travelng dstance for the ground-reflected wave and h θ s the dpole feld pattern defned as: h θ (θ) = cos[ π 2 cos(θ)] sn(θ) (5) θ moreover the ground Fresnel coeffcent for parallel and vertcal polarzatons R and R depend on relatve delectrc constant of the ground ɛ r and the grazng angle accordng to R K = cos( 9) k ɛ r sn 2 ( 9) cos( 9) + k ɛ r sn 2 ( 9) (6) where k = 1/ɛ r for R K = R and k = 1 for R K = R. The relatve delectrc constant s ɛ r = ɛ r j6λσ, n whch the values for relatve permttvty ɛ r and conductvty σ for an average ground are 15 and /m respectvely [13], and λ s the operatng wave-length. For grazng angles however the reflecton coeffcents can be approxmated to R = R = 1 [11],[12]. On the other hand, for the drect ncdent wave, the propagaton drecton s horzontal and the followng relatons exst: d = d φ = (7) d = d θ = e jkr d r d (8) where r d s the the travelng dstance for the drect ncdent wave. The ncdent angle of the total electrc feld can be obtaned from superposton of the two ncdent waves that s: θ = π { 2 + d + } tan 1 g d + g (9) Fg. 5 shows the coelevaton of the superposed ncdent wave as a functon of the ground-reflected wave coelevaton. Also n Fg. 6 the RCS of the cylnder obtaned from the superposed ncdent wave s llustrated. It s observed that the RCS ncreases for grazng ground reflectons. The reason can be understood by nspectng Fg. 5, n whch for grazng ground reflecton coelevatons the coelevaton of the superposed ncdent wave decreases back toward 9. In Fg. 6 the total coherent RCS usng R = R = 1 s also added for comparson. It s observed that the total RCS range usng average ground reflecton coeffcents and R = R = 1 does not show any sgnfcant dfference. As a consequence and for the sake of smplcty, we wll use these coeffcents n the smulatons dscussed later n ths work. 9 θ g Fg.5: Incdent coelevaton angle of the superposed drect and ground-reflected waves as a functon of ground reflecton coelevaton. RCS [dbm 2 ] Average ground R =R = 1 9 θ g Fg.6: The total RCS of the conductor cylnder n whch the coherent drect and ground-reflected waves are mpnged as a functon of ground reflecton coelevaton. L = 6.5 m, a =.8 m, θ s = Incoherent waves It s observable that the total ncdent and scatterng waves can be obtaned as: = d + g (1) s = s d + s g (11) Where s d and s g are the scatterng electrc felds due to drect and ground-reflected ncdent waves respectvely. Accordng to the defnton, the total RCS σ t s: s σ t = lm r r 4πr2 d + s g 2 r d + (12) g 2 In the ncoherent ncdent and scatterng, the total ncdent power s the ncoherent superposton of the two ncdent waves and scattered waves are also ncoherently added. The total RCS for ths case therefore s equal to: σ t = s lm r r 4πr2 d 2 + s g 2 r (13) d 2 + g 2 After smple manpulatons, the total RCS expresson turns to the followng formula: σ t = σ d + Γ g σ g 1 + Γ g (14)

4 Rado Wave Scatterng from Lampposts n Mcrocell Urban Moble Propagaton Channel 17 Table 1: xperments Specfcatons f c 3.35 GHz Sgnal BPSK wth PN-9 sequence Chp rate 5 Mcps Tx Power 4 dbm Antenna Sleeve (2.2 db) Antenna heght 3 m Antenna Patch array (24.5 db), 1 beamwdths n azmuth and elevaton sdelobe level -26 db Rx Antenna heght 3 m Azmuth scan 3 step (12 ponts for full azmuth) Tx-Rx confguraton 6 m separated, lne-of-sght (LoS) σ t [dbm 2 ] Fg.7: The total RCS of the conductor cylnder for the ncoherent drect and ground-reflected waves as a functon of ground reflecton coelevaton. L = 6.5 m, a =.8 m, θ s = 9. where the RCS for the drect ncdent wave separately s σ d and that of the ground reflecton s σ g. Moreover, Γ g s the rato between the ncdent ground reflecton power and the ncdent drect path power: Γ g = g 2 / d 2. Ths factor s dfferent from Fresnel coeffcent as the drect wave and the groundreflected wave travel dfferent dstances to the scatterer. Fg. 7 llustrates the total ncoherent RCS for a cylnder of L = 6.5 m, a =.8 m, θ s = 9 and φ =. It s observable that for the ncoherent case, by ncreasng the ncdent ground-reflected coelevaton the total RCS of the cylnder changes no more than a few dbs. 4. MASURMNTS In ths secton the measurement campagn n urban areas s descrbed. We wll derve the RCS of a number of lampposts n two locatons of urban areas n Yokohama, and then wll compare the values wth those obtaned from two ray models of secton Set up and scenaro Table 1 shows the specfcatons of the experments. The detaled descrpton of the measurements, data analyss and scatterer dentfcatons can be found n [1],[7]. Measurements were accomplshed n a street of 26 m wde (locaton 1) and another nearby street of 18 m wde (locaton 2) n the dense Table 2: The RCS of lampposts of the locaton 1, measurement ponts 1 and 2 (P1, P2). σ m [m 2 ] No P1 P NA Table 3: The RCS of lampposts of the locaton 2, measurement ponts 1, 2 and 3 (P1, P2, P3). σ m [m 2 ] No P1 P2 P v NA NA urban areas of Yokohama, Japan. The drectonal data was acqured and by nvestgatng the azmuthdelay-power spectrum we were able to dentfy the scatterer objects n azmuth-delay doman. The sgnfcance of the scatterng from dentfed objects such as lamppost, sgnboard, etc. s estmated to be up to 4% of the non-lne-of-sght receved power [1]. The mportant relevant pont s that the Tx dpole antenna was rotated wth a dameter of.5 m and constant rotaton speed of 5 rpm to create dynamc uncorrelated fadng [1],[7]. As a consequence, the transmtted multpath are averaged. We use ths to choose the approprate two-ray model n the smulatons. In the next subsecton we wll use the outcome of the lamppost dentfcaton to get ts RCS n each of the measurement locatons Measured RCS For locaton 1, the number of dentfed lampposts s 6, whereas 13 lampposts were dentfed n locaton 2. To obtan the RCS of lampposts from the mea-

5 18 CTI TRANSACTIONS ON LCTRICAL NG., LCTRONICS, AND COMMUNICATIONS VOL.7, NO.1 February 29 surement data we can use the radar equaton [14]: P ra = P tag t G r λ 2 (4π) 3 r 2 t r 2 r σ m (15) where: P ta : transmtted power at the Tx antenna, P ra : receved power at the Rx antenna, G t : gan of the Tx antenna, G r : gan of the Rx antenna, r t : dstance between Tx antenna and the scatterer, r r : dstance between Rx antenna and the scatterer, λ : operatng wave-length, σ m : RCS of the scatterer obtaned from measurement data. Table 2 shows the the RCS of the dentfed lampposts n locaton 1 derved for 2 measurement ponts, and Table 3 shows the RCS of the dentfed lampposts n locaton 2 obtaned for 3 measurement ponts. Remarkable s that the lampposts n each measurement locaton are unform whereas the shape of lampposts n locaton 1 s dfferent from those of locaton 2. Therefore, as the RCS value s a functon of the scatterer s shape, the varatons of RCS n each table s unexpected Analytcal RCS values Neglectng multpath effect and consderng the equal Tx antenna and Rx antenna heghts, t can be assumed that the ncdent and scattered waves are both horzontal. In ths case the RCS, σ, can be calculated usng θ = θ s = 9, φ =, a =.8 m, L = 6.5 m for locaton 1 and a =.7 m and L = 4.5 m for locaton 2. Wth these parameters the value of σ for lampposts of locaton 1 s 241 m 2 and of locaton 2 s 11 m 2. These values however can not explan the varatons n measured RCS of tables 2 and 3. In the next secton, we employ those models of secton 2. to reproduce these varatons. 5. SIMULATION Varatons of the measured RCS from the analytcal values of secton 4.3 can be due to changes n the coelevaton of the ncdence, and the ncdent and scatterng azmuth dfferences. The former can be modeled accordng to analyses n secton 3. whereas the latter can be assumed a random value. In the smulatons the two-ray model s consdered for the ncdence and only drect scatterng wave s accounted because the Rx antenna s drectve and the ground reflecton s not wthn the Rx antenna beam-wdth as the scenaro s llustrated n Fg. 8. Moreover even though ncdent multpath superpose coherently at the scatterer, as a consequence of averagng the multpath by rotatng the Tx antenna n the experments, the superposton of the ncdent multpath at σ t [m 2 ] Fg.9: The total RCS value σ t for lampposts of locaton 1 as a functon of the ncdent angle of ground reflecton wave. σ t [m 2 ] Fg.1: The total RCS value σ t for lampposts of locaton 2 as a functon of the ncdent angle of ground reflecton wave. the scatterer n ncoherent. The RCS values of the ncoherent two-ray model σ t for the lampposts of locaton 1 and of locaton 2 are presented n fgures 9 and 1 respectvely as a functon of ground-reflected wave coelevaton n lnear scales. It s observed that consstent wth what nferred from the measurement scenaro, the range of these values are close to those of tables 2 and 3. Ths s n contrast to two-ray coherent model dscussed n secton 3. 1 whch predcts values much smaller than measured lampppost RCS. To nvestgate the goodness of the exercsed model we conducted a Monte-Carlo smulaton by assumng that lampposts are dstrbuted randomly n each of the measurement locatons. Hence t s assumed that φ s unformly dstrbuted over the nterval [, 15 ], because the measurement s n lne-of-sght scenaro and no lamppost can locate n between Tx and Rx to make a larger φ. The ground reflecton coelevaton s assumed n the nterval [9, 18 ], however ts most probable regon s assumed [93, 15 ] correspondng to a Tx to the lamppost separaton of 12 m to 3 m. Moreover, smulaton results show that for locaton 1 best results are obtaned wth %9 of the lampposts n ths regon, whereas ths value for locaton 2 s %8.

6 Rado Wave Scatterng from Lampposts n Mcrocell Urban Moble Propagaton Channel 19 Fg.8: Two ncdent waves at the lamppost, whereas only drect scatterng wave s consdered as the Rx antenna s drectve. 6. DISCUSSION As t was expressed n prevous secton, the ncoherent two-ray model was employed amd coherent scatterng n the channel because the Tx antenna was rotated durng the measurements to average the transmtted multpath. Ths wll cease to ncoherent superposton of the ncdent multpath at the scatterers. In addton, the RCS values obtaned from the ncoherent two-ray ncdence model present a close match to the measured values n the urban envronments. The CDF of the analytcal RCS values σ t derved from Monte-Carlo smulatons and those of measured values σ m are presented n fgures 11 and 12 for locatons 1 and 2 respectvely. It s observed that by usng the ncoherent two-ray model for the ncdent wave to the lamppost, an excellent agreement between the CDFs of the measured and smulated values are achevable. Small dfferences n the low RCS values for locaton 2 n Fg. 12 s observable whch may be explaned by a two-ray scatterng model. arler we argued that we do not consder the two-ray model for the scatterng, nonetheless there can be some lampposts wth the scatterng ground reflectons wthn the beam-wdth of the Rx drectve antenna. If two waves (the drect and the ground-reflected) were supermposed destructvely at the Rx [12], a lower measured scatterng power at the Rx would be observed and consequently the measured RCS would be smaller. 7. CONCLUSION The study of scatterng from objects such as lampposts s a crtcal ssue n the analyss of the wreless propagaton channel n the urban envronments. Lampposts are wdely used n such areas and exhbt a sgnfcant potental for scatterng rado waves. In ths paper, we employed a two-ray ncdence model to explan the rado wave scatterng propertes of the lampposts n the urban wreless rado channel. A comparson of the CDF of the RCS obtaned from smulaton and those calculated from the measurement data shows that the two-ray ncdence model can predct the scatterng of rado waves from lamp- CDF Smulaton Measurement RCS [m 2 ] Fg.11: The CDF of the total RCS for the lampposts of locaton 1. CDF Smulaton Measurement RCS [m 2 ] Fg.12: The CDF of the total RCS for the lampposts of locaton 2. posts n the urban areas. References [1] M. Ghorash, J. Takada, T. Ima, Identfcaton of scatterng objects n mcrocell urban moble propagaton channel, I Trans. on Antennas and Propagaton, Vol. 54, No. 11, pp , Nov. 26. [2] H. Budarto, K. Horhata, K. Haneda, J. Takada, xpermental study of non-specular wave scatterng from buldng surface roughness for the moble propagaton modelng, IIC Trans. on Communcatons, Vol. 87-B, No.4, pp , Aprl 24.

7 2 CTI TRANSACTIONS ON LCTRICAL NG., LCTRONICS, AND COMMUNICATIONS VOL.7, NO.1 February 29 [3] P. Pongslamanee, H. Berton, Specular and nonspecular scatterng from buldng facades, I Trans. on Antennas and Propagaton, Vol. 52, No. 7, pp , July 24. [4] V. Degl-spost, F. Fuschn,. Vtucc, G. Falcasecca, Measurement and modellng of scatterng from buldngs, I Trans. on Antennas and Propagaton, Vol. 55, No. 1, pp , Jan. 27. [5] N. Blaunsten, D. Censor, D. Katz, Rado propagaton n rural resdental areas wth vegetaton, Progress In lectromagnetcs Research, PIRS 4, , 23. [6] Y. de Jong, M. Herben, A tree-scatterng model for mproved propagaton predcton n urban mcrocells, I Trans. on Vehcular Technology, Vol. 53, No. 2, pp , March 24. [7] M. Ghorash, J. Takada, T. Ima, Mcrocell urban propagaton channel analyss usng measurement data, Proc. of I Vehcular Technology Conf. (VTC 5 Fall), Vol. 3, pp , Sept. 25. [8] I standard defntons of terms for antennas Antennas and Propagaton, I Standards , June, 22, [9] A. Bhattacharyya, D. Sengupta, Radar Cross Secton Analyss and Control, Artech House, [1] V. DCaudo, W. Martn, Approxmate soluton to bstatc radar cross secton of fnte length, nfntely conductng cylnder, I Trans. on Antennas and Propagaton, AP-14, No. 5, pp , Sept [11] H. Xa, H.L. Berton, L.R. Macel, A. Lndsay- Stewart, R. Rowe, Rado propagaton characterstcs for lne-of-sght mcrocellular and personal communcatons, I Trans. on Antennas and Propagaton, Vol. 41, No. 1, pp , Oct [12] R. Vaughan, J. Bach Andersen, Channels, Propagaton and Antennas for Moble Communcatons, The I Press, 23. [13] W. Jakes (dtor), Mcrowave Moble Communcatons, Wley-I Press, [14]. Knott, J. Shaeffer, M. Tuley, Radar Cross Secton, Artech House, lectroncs ngneerng (I) and the Insttute of lectroncs, Informaton and Communcaton ngneers of Japan (I- IC). Jun-ch Takada receved the B.., M.., and D.. degrees from the Tokyo Insttute of Technology, Tokyo, Japan, n 1987, 1989, and 1992, respectvely. From 1992 to 1994, he was a Research Assocate Professor wth Chba Unversty, Chba, Japan. From 1994 to 26, he was an Assocate Professor wth Tokyo Insttute of Technology. Snce 26, he has been a Professor wth Tokyo Insttute of Technology. Hs current nterests are wreless propagaton and channel modelng, array sgnal processng, UWB rado, cogntve rado, appled rado nstrumentaton and measurements, and ICT for nternatonal development. Dr. Takada s a member of I, IIC, ACS, and the CTI Assocaton Thaland. Tetsuro Ima was born n Tochg, Japan, n He receved hs B.S. and Ph.D. degrees from Tohoku Unversty, Japan, n 1991 and 22, respectvely. He joned the Wreless System Laboratores of Nppon Telegraph and Telephone Corporaton (NTT), Kanagawa, Japan, n Snce then, he has been engaged n the research and development of rado propagaton and system desgn for moble communcatons. He s now Manager of the Rado Access Network Development Department, NTT DOCOMO, INC., Kanagawa, Japan. Dr. Ima s a member of the Insttute of lectrcal and lectroncs ngneerng (I) and the Insttute of lectroncs, Informaton and Communcaton ngneers of Japan (IIC). Mr Ghorash receved B.. from Isfahan Unversty of Technology, sfahan, Iran, and M.. from Amrkabr Unversty of Technology, Tehran, Iran, both n electrcal engneerng n 1993 and 1999 respectvely. In 1999 he entered Tokyo Insttute of Technology for Ph.D. course where he s a swenor researcher from 24. Hs research nterests are rado channel analyss and sgnal processng fro wreless communcaton systems and wreless postonng and trackng. He s a member of the Insttute of lectrcal and