FAST PATH LOSS PREDICTION BY USING VIRTUAL SOURCE TECHNIQUE FOR URBAN MICROCELLS Haan M. El-Sallai Radio Laoratory, Intitute of Radio Communication (IRC) Helinki Univerity of Technology P. O. Box 3000, FIN-0015 HUT, Finland Tel: +358 9 451 55, Fax: +358 9 451 15 E-mail: hel#radio.hut.fi Atract - Thi work propoe a new principle for fat path lo prediction and cell coverage calculation for uran microcell in high rie uilding The technique i aed on finding a virtual ource located in line of ight with the moile tation (MS). Thi technique map the out-of-ight propagation prediction prolem to a line-of-ight propagation prediction prolem. The received power at MS i given in a cloed form expreion. Baed on comparion performed with experimental reult, it eem that reakpoint i valid in ide-treet propagation. The propoed virtual ource principle open a door for a new fat computation technique for path lo modeling in the microcellular I. Introduction Currently, the concept of microcell i ued to increae the capacity of radio network and overcome the crowding of the radio pectrum [1]. Microcell generally refer to mall cell (< 1 km) with the antenna that are elow the urrounding uilding rooftop. Thi placement of antenna reult in a different propagation mechanim from that of cellular communication ytem. The treet orientation and individual lock of uilding make a difference in ignal reception. With the increaing demand for cheaper and etter wirele ervice, it i ecoming very important to optimal deign the cell geometrie and deploy the minimum numer of ae tation to provide the maximum poile coverage. Therefore, to calculate the coverage, a fat path lo prediction tool i of primary importance. Thi tool i eential for efficient planning of current and future radio network. II. Background It i important to develop reliale planning tool for future roadand communication ytem in uran microcell. In network planning different empirical propagation model have een ued to calculate the coverage area of the ae tation []. Thee model give good prediction when the tranmitter i aove the rooftop and give poor reult when applied for microcell. However, thee model do not take into account the exact geometry of the urrounding uilding and the actual wave propagation phenomena. In uran microcell, the uual approach to tudy channel propertie i to model the propagation y uing the ray-optical approximation. Thee model uually ue the ray tracing algorithm to compute the ray from the tranmitter to the receiver via different propagation mechanim, which uually are reflection, diffraction and ome comination of them. The ray tracing technique ha ome drawack. Firtly, the technique i very tediou to implement. Therefore, we have to take another approach in our prediction, which will e decried in thi paper. Secondly, the ray tracing require large dataae of the propagation environment, which may not exit or i expenive to get, although only part of it i eential in the propagation prediction. Thirdly, the lengthy time pent in ray tracing computation i a major prolem with moile radio propagation prediction for uran microcellular environment. Fourthly, the ue of different value of permittivity and conductivity in the calculation of Frenel reflection coefficient make large difference in the prediction of multiple reflection. Different ray tracing technique have een introduced to overcome
the lengthy computation time ut they are till not fat enough. In thi work, a new principle for fat path lo and cell coverage prediction of uran microcellular environment i propoed. Thi new model predict propagation for moile tation (MS) traveling in out-of-ight ide treet. The propoed technique i aed on uing virtual ource principle. The previou work [3],[4] propoed a virtual ource technique for parallel treet propagation prediction prolem and mapping it to ide treet propagation prediction prolem. Thi work implifie the prolem more y placing the propoed virtual ource in line-of-ight with the moile tation. III. Propoed Technique Thi new model predict propagation for moile tation (MS) traveling in out-of-ight ide treet. The propoed new technique i aed on uing virtual ource concept. The virtual ource i located in a poition having line-of-ight with the MS. Thu, thi new technique map the out-of-ight propagation prediction prolem into line-of-ight propagation prediction prolem. The technique provide the received power at the MS along the ide treet in cloed form expreion taking into account the information aout the BS and the MS poition in microcellular Side-Street Propagation Prediction The poition of the ae tation with repect to the urrounding uilding ha influence on the dominant propagation mechanim. Two poiilitie of ae tation location can e conidered with repect to the ditance etween idewall and the ae tation. In the firt cae, the ae tation i on (or cloe to) the idewall, the econd cae, the poition i at ome ditance from the ide-wall. In the firt cae, it i expected that the dominant propagation mechanim i the forward multiple diffraction. Therefore, the ide treet corner, on the ide where the ae tation i located, may work a a ource (virtual ource), VS I in Fig. 1, for radio wave propagation in the out-of-ight ide treet. When the ae tation i cloe to or on the idewall, the propagation mechanim i dominated with low grazing incidence ray, thu, the forward multiple-diffraction i expected to e ignificant for wave propagation in to the ide treet. Utilizing the reult preented in [5] and making them applicale for microcellular environment, a cloed form formulation could e found for the received power at the moile tation due to thi virtual ource. w BS r VS II w m x VS I R MS Fig. 1. A microcellular environment plan view When the ae tation i located at ome ditance from the ide-wall, the dominant propagation mechanim for the ignal received at the moile tation in the ide-treet depend on the ditance of the moile and ae tation to the interection [],[6]. When the moile i in the out-of-ight region and cloe to the junction the reflection mechanim dominate [6]. When the moile i further away from the junction, the diffraction mechanim dominate [],[6]. The ditance at which the diffraction ecome dominant over the reflection depend alo on the ditance from the tranmitting antenna to the turning corner, ut alo depend on the frequency, treet width, and the poition of the tranmitting and receiving antenna with repect to the idewall. The formulation preented in thi work i aed on the principle that the power leakage in to the ide treet i regarded a a virtual ource for the propagation of wave in to the ide treet. There could e different approache for the derivation of thi power
leakage. For implicity, the following aumption are made in derivation of the leakage power: Firt, neglecting entering power due to diffraction and reflection in the main treet (i.e., where the ae tation i located). Second, the virtual ource i placed in the center of the ide treet. Third, it i aumed that it ha the ame offet ditance x from the idewall in the main treet. Fourth, it i alo aumed that it ha the ame height a the ae tation antenna height. Thu, the aim i to find the total direct radiated power that enter the ide treet. Conidering thi total power entering the ide treet a the tranmit power of a virtual ource (VS II ), the virtual ource i located in line of ight with the moile tation (ee Fig. 1). Since the virtual ource repreent the power entering the ide treet, thu, it tranmit only in the dedicated ide treet. For the oppoite part of the ide treet, different virtual ource with different tranmitting power i needed. The virtual ource power i derived utilizing angular information of direct wave that propagate in to the ide treet from the ae tation. Thi model i valid for the cae when the ae tation i located at ome ditance from the idewall. The received power at the moile tation due to the virtual ource can e given in a cloed form expreion a α Pr = α xw λ Pt, π 4π r R xw λr Pt, π 4π r R R r, (1) R > r where P r i the received power y iotropic antenna at MS, P t i the tranmitted power y iotropic antenna at the ae tation (BS), λ i the wavelength, x i the ditance etween the BS and the idewall, for the oppoite part of the ide treet x i replaced with w m -w, where w m i the main treet width, w i the ide-treet width, r i the ditance etween the BS and the center of the ide-treet where the MS i traveling, R i the traveling ditance in the ide treet from the virtual ource, α i a treet parameter, it i introduced here to pecify treet characteritic ince there are no two identical treet. The parameter r i the reak point ditance [7], defined a 4hthr r = () λ where h t i the ae tation antenna height and h r i the moile tation antenna height. IV. Numerical Reult In order to validate the given expreion for received power due to the virtual ource, we have computed the path lo for everal uran treet grid in Manhattan and Tokyo for comparion with pulihed meaurement in open literature [8]. Experimental data In Manhattan data were collected at 900 MHz. The tranmitting and receiving antenna height were 30 ft and 6 ft, repectively. Polarization wa vertical. In Tokyo data were collected at 1.5 GHz with vertical polarization and tranmitting and receiving antenna were 5.3 m and 3 m, repectively. Comparion The aove-mentioned experimental data and geometrical parameter from the correponding city treet plan view are ued in computation. Comparion etween the prediction and the meaurement for oth Tokyo and Manhattan how good agreement. Having a cloed form expreion make computation fat ince there i no earching for ray that join the BS and the MS. Running a program written in MATLAB in tandard PC, the computation lat le than a econd for more than 600 point along MS traveling route. Figure how the model prediction and Manhattan meaurement reult [8]. The tranmitter i located at Lexington Avenue and the moile tation i traveling along the 51t treet (i.e., ide treet) a hown in Fig. 3 in [8]. The meaurement reult ( ) in Fig. are otained from Fig. 7 in [8]. The olid line repreent computation reult of formulation provided in Eqn. (1). The geometrical parameter are w =1 m, w m = 36 m, x= 14 m,
and r = 40 m. The treet parameter α=1 for oth ide of croing treet. 60 10 000 0 0 1000 000 3000 4000 5000 Ditance, feet Fig.. Path lo along a ide treet (51t in Manhattan) with w =1 m, w m =36 m, x=14 m, r =40 m. It can readily e een from Fig. that the reakpoint eem to e valid for the ide-treet propagation when the virtual ource i ued. The dotted line repreent the calculation reult of Eqn. (1) utilizing the econd power law lo. One can ee that the meaurement reult how fourth power law lo eyond the reak point. Uing fourth power law lo how good agreement with the meaurement at ditance eyond the reakpoint. Figure 3 and Fig. 4 depict the agreement with meaurement reult carried out along ide treet in Tokyo. The olid line repreent the computation reult of expreion of Eqn. (1). The ar repreent the meaurement reult. The meaurement, in Fig. 3 and Fig. 4, are otained from Fig. 4 and Fig. 5 in [8], repectively. The geometrical parameter are w m = 33 m, x= 5.5 m, w =15 m, r = 5 m for the ide treet of Fig. 3 and w =45 m, r = 710 m for the ide treet of Fig. 4. The treet parameter i different from that ued in Manhattan calculation ince the two 1 environment are quite different, α = i π ued for the two ide treet AC and BE a defined in Fig. in [8] and α = π i ued for the two ide treet AD and BF a defined in Fig. in [8]. 10 400 300 00 0 100 00 300 Ditance, m Fig. 3. Path lo along ide treet near Shinahi tation Tokyo with w =15 m, w m =33 m, x=5.5 m, r =5 m. 10 150 00 0 100 00 300 Ditance, m Fig. 4. Path lo along ide treet near Shinahi tation Tokyo with w =45 m, w m =33 m, x=5.5 m, r =710 m.
For figure, 3, and 4, when the moile tation i traveling along the interection and i in a location having line of ight with ae tation (BS), a line-of-ight model [9] i ued in calculation rather than the virtual ource. The zero of the ditance lael in the acia of figure, 3, and 4 are from the center of the interection. V. Concluion Thi work propoe a principle for a new fat path lo and cell coverage prediction technique for microcellular The new technique propoed in thi work make the out-of-ight ide-treet propagation prediction prolem impler y mapping it to a line-of-ight propagation prediction prolem. The received ignal power at the MS i given in a cloed form expreion, which make computation very fat. By comparion with meaurement, it eem that the reakpoint i valid in the ide treet. The virtual ource principle preented in thi work open a door for new fat computation technique for propagation modeling in uran microcellular Acknowledgement Thi work i a part of reearch project of the Intitute of Radio Communication (IRC). The author thank Prof. Pertti Vanikainen and Dr. Wei Zhang for valuale dicuion. The author i alo grateful to Nokia Foundation for the financial upport. Reference [1] W. C. Y. Lee, Smaller Cell for Greater Performance," IEEE Communication Magazine, pp. 5-71, Nov 1991. [] H. L. Bertoni, W. Honcharenko, L. R. Maciel, and H. H. Xia, UHF propagation prediction for wirele peronal communication, Proc. IEEE, vol. 8, pp. 1333-1359, Sept. 1994. [3] H. M. El-Sallai, W. Zhang, and P. Vainikainen, "An efficient technique for modeling wideand propagation in an uran parallel treet microcellular environment," in Proc. IEEE Antenna and Propagat. Soc. International Sympoium (AP-S' 99), Orlando, Flourida, pp. 1490 1493, July 11-16, 1999. [4] W. Zhang, J. Lähteenmäki, P. Vainikainen, and H. M. El-Sallai "Fat two dimenional diffraction modeling for moile radio propagation prediction in uran micro-cellular environment" in Proc. 8 th European Microwave Conf. Amterdam, The Netherland, pp. 36-367, Oct. 1998. [5] H. H. Xia, "A implified analytical model for predicting path lo in uran and uuran environment," IEEE Tran. on Veh. Tech., vol. 46, pp. 1040-1046, Nov. 1997. [6] V. Erceg, A. J. Rutako Jr., and R.S. Roman, Diffraction around corner and it effect on the microcell coverage area in uran and uuran environment at 900 MHz, GHz, and 6 GHz," IEEE Tranaction on Vehicular Technology, vol. 43, no. 3, pp. 76-766, Aug 1994. [7] H. H. Xia, H. L. Bertoni, L. Maciel, A. Linday-Stewart, and R. Rowe, Radio propagation meaurement and modeling for line-of-ight micrcellular and peronal ytem, in IEEE Antenna and Propagat., vol. 41 no. 10, pp. 1439-1447, 1993. [8] S. Y. Tan and H. S. Tan, A micro-cellular communication propagation model aed on the uniform theory of diffraction and multiple image theory, IEEE Tran. Antenna and Propagat., Vol. 44, pp.1317-136, Oct. 1996. [9] Haan. M. El-Sallai and P. Vainikainen, "Modeling and imulation of wideand radio channel characterization for an uran line-of-ight microcell," in Proc. IEEE Vehicul. Technolg. Conference VTC'99 fall, 19-, Sep. 1999, pp. 383-387, Amterdam, the Netherland.