A Propagation Model for Mobile Radio Propagation Loss in an Urban Area at 800 MHz

7/29/2019 A Propagation Model for Mobile Radio Propagation Loss in an Urban Area at 800 MHz

1/14

Electronics and Communications in Japan, Part 1, Vol. 76,No . 10, 1993Translated from Denshi Joho Tsushin Gakkai Ronbunshi, Vol. 75-B-lI, No. 9, September 1992, pp. 638-646

A Propagation Model for Mobile Radio Propagation Loss inan Urban Area at 800 lMHz

Shinich Ichitsubo and Masahiro Kimura, MembersNTT Radio Communication Systems Laboratories, Yokosuka, Japan 238-03

SUMMARYTo understand the propagation structure of the

microcells with a tall and medium-height antenna in anurban area, the fundamental propagation parameters arefound and the structural model of the propagation loss wasconstructed. The fundamental propagation parameterswere obtained based on the multivariable analysis of thepropagation data at 800 MHz measured in the TokyoMetropolitan area. These parameters are the distancebetween the transmitter and receiver, the average buildingheight between the transmitter and receiver, and the base-station height which have high correlation and regressioncoefficients with the propagation loss.

The relationship of parameters obtained from themodel is (20 - a) og@) for the base-station heightcharacteristics and (a 20) log (H) for the building heightcharacteristics if the distance characteristicsare a og (d).The relationships of coefficients of the conventionalestimation equation and the regression equation obtainedby the measured data almost agreed with those of themodel so that the validity of the model was confirmed.

Key words: Radio propagation; propagation model;mobile communication; multivariable analysis.

1. IntroductionIn the system design of land mobile communication

such as the automobile telephone and the cellular

91

telephone, an important topic is estimation of thepropagation loss. Based on th is estimation, the systemparameters such as the location of the base station,transmission power and channel allocations aredetermined. Hence, a practical estimation equation isdesired.

Especially, due to crowded frequency allocationsasthe mobile communication is developed, the traditionalcell radius must be reduced and the use of microcells isconsidered. Accordingly, the propagation loss estimationfor the microcell is being investigated [l-31. Since themicrocell is not defined clearly at present, a cell isassumed to be a microcell if the cell radius is less than 1km even if the base-station height is greater than thesurrounding building height. This paper studies themicrocell for which the base-station height is greater thanthat of surrounding buildings.

A practical estimation equation for the propagationloss is the one in which the definitions of the propagationparameters (such as frequency, locations of the transmitterand receiver, and urban structure) needed for estimationare clear and accurate estimation is possible within a shorttime by a computer. To this end, it is important tounderstand the fundamentalpropagation parameters and tofind the relationship of these parameters with thepropagation loss. In general, the most fundamentalparameter is the distance r which can be related to thepropagation loss via f (n = 3.6 - 3.8 [4, 1. Otherfundamental parameters are discussed only in theSakagami equation using multivariable analysis.

ISSN8756-662 1/93/0010-009@ 1994 Scripta Technica. Inc.


2/14

terrestrialmagne i sm

1 sensorpositionmeter

8OOMlz - edium value800Mlzvehiclespeed

computer sensor[ transmi terl [receiver]

Fig. 1. Measurement system.

For the foregoing reason, this paper selectedpropagation parameters which significantly affect thepropagation loss to clarify the propagation loss structurein an urban area. Based on he relationships among theseparameters, the propagation loss structure was modeled.In the section of the propagation characteristics, the multi-variable analysis used in the formation of the Sakagamiequation also is used and the m easured data in the TokyoMetropolitan area w ere regressed.

The data used for regression were those measured at80 0 MHz in the case where the base-station height isgreater than those of the surrounding buildings and thedistance between the transmitter and receiver is less than1 km. The measured data are regressed by thepropagation parameters w ith significant effects so that anempirical equation was obtained. This empirical equationand conventional estimation equation were compared withthe proposed m odel. The relationship of the characteristiccoefficients agree well with the one indicated by themodel.

2. Fundamental Propagation Parameters2.1. Propagation loss measurementBased on the measured data, the fundamentalparameters in an urban area are derived by the multipleregression analysis among the m ultivariable analysis. Toobtain the propagation data in an urban area, themeasurement was camed out in Tokyo . The base stationswere installed on the towers or on the roofs of NITTokyo Central Network Center, NTT Kyobashi Equip-ment Center and NTT Kuramae Network Center (to becalled Chiyoda Station, Kyobashi Station and Kuramae

Station). The m easurement was carried out by driving thearea within a radius of 1 km. The frequency was 813MHz. The medium value of the 10-m short section wasrecorded digitally on a tape recorder together with thelocation coordinate. Sleeve antennas were used for thetransmitter and the receiver. Figure 1 shows themeasurement system while Table 1 lists the measurementparameters.

When a base station was installed at ChiyodaStation, Nihonbashl District with m any high rise buildingswas measured. For Kyob ashi Station, Ginza District withcriss-crossed streets was measured. On the other hand,from Kuramae District, the region east of Ueno RailwayStation was measured where the height of buildings waslower than two other districts. The m easured area is fan-shaped with a radius of 1 km centered at the base station.There is no area shadowed by the tower or the stationbuilding. The total distance of measurement was 10 to 20km per base station. Figure 2 shows an example ofmeasurement course for Kyobashi Station as the basestation.

2.2.

Theparameter

Parameter selection

ten parameters intended for propagationselection are shown below. The frequencyparameter was not considered at this time. Also,when acell is constructed, the radiation pattern of the antenna isan mpo rtant param eter. If the propag ation structure withan omnidirectional antenna can be understood, then thecase with a directional antenna can be handled byincorporating the antenna pattern. Hence, this parameterwas not considered.

92


3/14

Table 1. Measurement parameters

Item MeasurementparametersFrequencyPowerBase-station height (transmission)

Mobile-station height (reception)Transmit and receive antenna

813.275 MHz20 W or 1 WChiyoda tower 86.0 mrooftop 37.5 mKyobashi rooftop 36.1 mKuramae tower 64.2 m

rooftop 39.4 mMeasurement vehicle 3.0mSleeve antenna

(a) Distance between the transmitter and thereceiver ( d ) . Distance between the transmitter and thereceiver in a two-dimensional plane.

...

(b) Distance between the transmitter and thereceiver (L). Distance between the transmitter and thereceiver in a three-dimensional plane.

, , , - , , , .1 1 1 1 1 1 1 , 1, , , , . , , , ,__ .____ . . _____ ._____ ._____ . . _ . _ . . . _ . . _ . . ___ . .

(c) Base-station height (hb). Ground height of thebase station.

(d) Area average building height (Har): Averagevalue of the heights of the buildings in the area within 1km radius from the base station.

(e) Building heigh t between the transm itter and thereceiver (Ho). Average value of the heights of thebuildings on the line connecting the base station with themobile station on the two-dimensional plane (on theground).(f ) Height of buildings near the mobile station(Hm).Average height of the buildings within 50 m x50 m from the mobile station on the base-station side.(g) Height ratio of the buildings in front of and inthe rear of the mobile station (Hm) n the base-stationside of the mobile station to that of the building height onthe other side.(h) Street width (W). Street width including thesidewalks (and is equal to the spacing of the buildings).(i) Street angle (0). The angle between thedirection of the street and that of the base station from themobile station.(i) Line-of-sight (LOS) . Existence or nonexistenceof the clear path between the transmitter and the receiver.Existence: LOS = 1, Nonexistence: LOS = 0.

93


4/14

Table 2. Regression results for propagation parameters

Parameters

T-R distanceT-R building heightBase-station heightStreet widthStreet angleLine-of-sight

Function

These parameter values were derived from thebuilding-stre et database. Because the database has no lowbuilding data, the building data were limited to thosehigher than 15 m. With the propagation loss as theobjective function and the propagation parameters as theexplaining functions, a multiple regression analysis wascarried out. The standard regression coefficient k, partialcorrelation coefficient rp and partial F value of eachparameter were obtained. Th e propagation loss was theaverage over 50 m derived from the 10-m short sectionmedium values obtained in the measurement. Thestandard regression coefficient k is the value when thestandard deviation of the explaining functions is nor-malized to unity. The magnitude of this parameter candetermine that of the variation of the propagation lossversus the variation of the parameter value. The partialcorrelation coefficient rp indicates the partial correlationbetween each explaining function and the propagationloss. Th e magnitude of this value determines the cor-relation between the parameters and the propagation loss.The standard regression coefficient k and the partial cor-relation coefficient rp take values between -1 and 1. Th eeffect is larger if the absolute value is larger. The partialF value detects the significance of each parameter.

The parameters related to the distance, viz. thedistance d, and the distance L have m utually high corre-lations. Hence, they cannot be regression analyzed onceas the explaining functions; a multiple regression wascamed out by using one parameter at one time. I t wasfound that the regression residual is smaller, the result fitsthe measured data better and m ake the standard regressioncoefficient k, he partial correlation coefficient rp and thepartial F value the highest if the distance L is used.

StandardregressioncoefficientK0.710.40

-0.22-0.14-0.13-0.05

Partialcorrelation coef-ficient y,0.780.60

-0.38-0.25-0.25-0.09

Partial F

284597330512111316

The parameters related to the buildings, i.e.,building height Har, H o, Hm, nd building height ratioFB , are highly correlated. Hence, the multiple regressionwas carried out by using one parameter at a time. Thefunction forms of each parameter used for the multipleregression were the true value and the logarithmic[log(H)]. They are evaluated by a function form with ahigh partial F value. It was found that the building heightHo is the most significant parameter.

The distance L is used as a parameter on thedistance while the building height Ho is used as a param-eter on th e height. With such a choice, a multiple regres-sion analysis was carried out. The function forms of eachparameter were logarithmic for the distance L and theheight H , while they are true value [WJ and logarithmic[log(W)] or the street width. For the street angle 8, thepower [([90$]/90)"(n = 1, 2, 3, 4)] and sinusoid [sin(8)]are used.

Finally, the line-of-sight LOS was true value(LOS 0. 11. Table 2 shows the results of the multipleregression analysis in terms of the most significantfunction forms. Even in the microcell, the distance L sthe maximum parameter and the partial correlationcoefficient is 0.78. The partial correlation coefficient forthe building height H , is 0.60. That of the base stationheight h, is -0.38. The negative value of the partialcorrelation coefficient indicates the reduction of the lossas the base-station height hb is increased. The partialcorrelation coefficient of the street width and the streetangle is -0.25. Although they are related to thepropagation loss, th e correlation is small. Depending onthe existence of the line-of-sight, the difference of thepropagation loss is empirically large. However, there are

94


5/14

a * h b I \("Fig. 3. Building-induced propagation loss.

Fig. 4. Distance Characteristics and height o i Bslocations where the received level increases due to reflec-tion and diffraction even if there is no line-of-sight.Hence, the partial correlation of the propagation loss andthe line-of-sight is low and no significant effect is seen.

characteristics.

Based on the data obtained by the presentmeasurement, the fundamental propagationparameters andtheir function forms are selected. From Table 2, they arethe following three: the distance log@), the buildingheight log(Ho), and the base-station height log(&). Toexpress the propagation loss structure in terms of a simplemodel, these three fundamental propagation parametersare used.

3. Model by Characteristic Relationship3.1. ResumptionsIn terms of the three fundamental propagationparameters, a model of the propagation loss in an urban

The propagation loss Loss is given by

(3) The height of the buildingsHo nd sizes in theurban area are identical and the distributions are uniform.From this presumption, the waves I and I1 with identicalincident angle 4 to the ground and the frequency haveidentical lossLB(6 Ho) caused by the buildings if only thebase-station heights are different. Figure 3 shows thewaves I and 11. If the loss by the buildings is LB(q5, I fo) ,the free-space loss is Lossf = 20 log(L) + C (C : con-stant), and the distances between the transmitter and thereceiver for the waves I and I1 are L , and L,, then thepropagation losses Loss(1) and Loss(I1) of the waves I and11are

area is considered. First, the relationship (4 model)between the distance characteristics and the base-stationpresumptions. Next, the structural mode adding thebuilding height characteristics is presented.

Loss( I 1=ZO.Iog(L ,)+LB(4,H")+ c (3 )Loss(I1) =20*log(L*)+LB(q5.h)+c (4)eight characteristics is shown under the following three3.2. Relationship between the characteristics (4

Presumptions. model)(1) The distance characteristics are in accordancewith $.If the distance between the transmitter and the re-

ceiver is L, the distance characteristic coefficient isa, ndthe constant is C , then the propagation loss is given by

(2) The propagation lossLoss is the sum of the freespace loss Lossf and the loss by the buildingsLB(4,Ifo).

Next, under the foregoing presumptions, therelationship between the distance characteristics and thebase-station height characteristics is shown. As shown inFig. 4, the waves I and 111 are radiated from the base-station height hb and the wave I1 is radiated from thebase-station height ahb. The distance between thetransmitter and the receiver for the wave I is Lo while forthe wave I1 is aLw The incident angle of th e waves I andI1 is 4, and that of the wave 111 is 6' . In the case ofahb a!do,he distance between the transmitter and thereceiver for the wave 111 is approximately4. ll waves

95


6/14

have an identical frequency f. If the loss of wave Icaused by the buildings is LB(t$, Ho), the distancecoefficient is a and the constant is CI, then the lossLoss(1) of the wave I is

B S

If Eq. (5) is expressed by the distance characteristics asin Eq. (l) , henM o b i l e

Fig. 5. Propagation model.Since the first term of Eq. (6) indicates the distancecharacteristics, the second term indicated by [ 1 becomesconstant if the base-station heights are identical. Thevalue in [ ] appears not to b e a constant since the distanceLo between the transmitter and the receiver changes.However, it is a constant since 4 in LB(4, Ho) lsochanges with Lo. In the case where the base-stationheights are identical, Eq. (6) becomes as follows in termsof a constant C2:

When the base-station height is changed to ah, inreference to the base-station height h,, the loss differenceis -(a - 20)log(a4 ,). Henc e, if the distance characteristicis expressed as alog(L), then the base-station heightcharacteristic is -(a - 20)log(hd.

Loss( I ) =a.log(L.o)+ c2 (7) 3.3 Propagation loss str u ctu r e modelThe propagation loss Loss(I1) of the wave I1 can beexpressed as follows similarly to Eq. (6) since the lossfrom the buildingLE(4, Ho) is identical to the wave I:

LOSS( I ) =2O*Iog(a . L o )+L B ($, Ho)+ C1

The coefficient relationship between the distancecharacteristics and the base-station height characteristicsis a: -(a 20). A model is constructed in which thebuilding height characteristics are added to the former.The propagation loss increases as the building height isincreased. Also, to maintain the coefficient relationship(4 model) of the distance characteristics and the base-station height characteristics is maintained, the followingitem (4) is assumed.

= Q . l O g ( a * L o ) + ( ( 2 0 - a ) ' l O g ( Q ' I - o )+ LB ( 9, H")+Cl) (8)

For the wave 111, the base-station height is identicalto the wave I while the distance is longer. Hence, thepropagation loss Loss(II1) is as follows in terms of thedistance characteristic alog(L,) in Eq. (7).

(4) The loss LB(q5, Ho) due to buildings is propor-tional to the logarithm of the propagation distance rb.As shown in Fig. 5, the loss LE(q5, H,) due to

buildings in the case of the base-station height of h,,building height of H,, and the distance Lo between thetransmitter and the receiver is as follows if the character-istic coefficient of the pro pagation distance rb in the build-ing part is q :

Loss (111) =Q ' log(a . L o ) +c2= u.log(a*Lo)+ ( ( 2 0 - a ) * l o g ( L , )

+ L B ( d ,HI,)+C1) (9)To obtain the loss difference due to the base-stationheight, the loss difference of the waves 111 and I1 is Ln(d,H O ) = ~ ' l o g ( ? ~ h )derived: = q ' l o g [ L O * / j l l / h h ]

= a . l o g ( L A + rl'log(l-ln)- q . l og( l z* ) (11)Since the distance characteristic coefficient is a nd thebase-station height characteristic is -(a 20), we let q =

96


7/14

Table 3. Regression coefficients

alog ( d ) 38.21 Partid Fropagation Regressioncharacteristics coefficient

3206

ylog (11")Constant

-17.6 33031og(ho)19.9 88215.5

-

FactorRegressionResidualTotal sum

Table 4. Detection of meaningfulness of reg ression

Degrees of freedom Square sum Unpartial dispersion Dispersion ratio3 68826 22942 1448

1759 68826 391762 238890

Also, if the frequency is f and the constant is C3,the free-space loss Lossfin Presumption (2) is as follows:

Lossf=20'log(La~+20~log~f)c 3 (13)Hence, the propagation loss becomes as followsfrom Eqs. (12) and (13):

In the assumed propagation model, the distancecharacteristic is a log(Lo), the base-station heightcharacteristic is -(a 20) log(h&, and the building heightcharacteristic is (a - 20) log(Ho). The distancecharacteristic coefficient a s a = 20 -t E if it is the sumof the free-space loss coefficient of 20 and the additionalpart E. The base-station height characteristic coefficient0 is -E while the building height characteristic coefficienty is expressed by E . Here, E is called the urban spacecoefficient in contrast to the free-space co efficient since it

is related to the parameters of urban structures such asdistance, base-station height and building height.

4. Regression Equation forPropagation LossTo compare the proposed structural model with themeasured results, the measured data example was ana-lyzed by multiple regression with th e fundamental propa-gation parameters. The measured data used for multipleregression analysis are identical to those presented insection 2. The total number of the 50-m averagepropagation loss data is 1763. The function form of eachparameter used for regression is as follows, if thepropagation loss is Loss, the distance characteristiccoefficient is a, he base-station height characteristiccoefficient is 0, the building height characteristic is y, andthe constant is C:

The result of regression is

97


8/14

Table 5 . Evaluation of regression

Evaluation item ValueMultiple correlation coefficientDetermining coefficientI Standard deviation of residual [ 6.24 dB I

The regression coefficient and the partis F value ceach characteristic by regression are listed in Table 3.The significance detection of regression is in Table 4whereas the regression evaluation is in Table 5. The Fvalue of regression is 1448 and the F value with a riskfactor of 0.5 percent, degrees of freedom of 3 and thenumber of data 1763 is F(3, 1759; 0 .5 percent) = 4.3.The regression is meaningful since the F value ofregression is larger than the F value with a risk factor of0.5 percent (1448 > 4.3). The determining coefficient is0.71 and the multiple regression coefficient is 0.84. Theregression residual from the measured value is 6.24 dB inthe standard deviation.

To study the convergence of each regressioncoefficient for the number of propagation data, it isassumed that the data used for regression are those of nspaced ones. The relationship between the number ofdata and the regression coefficient is shown in Fig. 6. Ifthe number of data is less than 10, then the regression isnot significantly with the risk factor 0.5 percent. As thenumber of data is increased, the regression coefficientbecomes stabilized.

In the present measurement example, the coefficientsof each characteristic were a = 20 + 18.2, p = -17.6and y = 19.9 as shown in Eq. 16). If the urban spacecoefficientE is assumed to be 18, then in the propagationmodel the distance characteristic coefficient is a = 38 ,the base-station height characteristic coefficient is /3 = -18and the building height characteristic coefficient isy = 18. The relationship of each characteristic coef-ficient obtained from the empirical equation (16) almostagrees with the value indicated by the propagation lossmodel.

On the other hand, if it is assumed from the propa-gation loss model that the coefficients of each character-istic are a = 38, /3 = -18 and y = 18, then the results ofregression of measured data are

1 0 100 1000Data numberFig. 6. Data number vs. coefficient for multiple

regression analysis.

The regression residual is 6.26 dB in the standarddeviation and is almost equal to the value in Eq. (16).

In the measured data used for comparison, thedistance between the transmitter and the receiver was lessthan 1 km at 800 MHz and hence the validity of themodel in this range was confirmed. Even if the distanceis more than 1 km, the distance characteristic isconsidered to follow the r" ule shown in Presumption (1)[ 4, 61. Hence, the applicable range of distance can beextended. Also, in terms of the application range of thefrequency, the frequency characteristic (0.4 to 2 GHz) istreated independently without relationship to the distanceand the building height in the Okumura-Hata equation [4 ,61 and the Sakagami equation [7]. Hence, the relationshipbetween the characteristic coefficients indicated by thestructural model is assumed to hold in the region of 0.4to 2 GHz.

5. Comparison of Conventional EstimationEquation and Present ModelThe conventional estimation equations for the urban

propagation loss and the empirical equation (16) (meas-ured) in this paper are compared with the proposedmodel. As the conventional equations, we use theOkumura-Hata equation (0-H), the Sakagami equation(SAKAGAMI), the Morita equation (MORITA) [8],

98


9/14

Te6P

ochescoe

moeqo

Dsa

B

g

[4965og(hb)

OmHa

log(d)

[-18-65o(d.)

log(h*)

Sm

Ma

[A2+A4log()

log(d)

A+A2log(d)

+[A3+A4log(d)Og(o)

WashkmI

A-P

I3og(d)

L

I36.8og(d)

PnemnI32og(d)

Budnhg

[A3+A4log(d)

loH

19og(u)


10/14


11/14


12/14

the base-station height h,, Fig. 7 shows the distancecharacteristic coefficient a for each estimation equationwith the base-station height as a parameter. As the base-station height is increased, th e distance characteristiccoefficient a ends to decrease. From Fig. 7, th e distancecharacteristic coefficient a s 36 to 38 .Figure 8 shows the base-station height characteristiccoefficient p with the transmitter-receiver distance as a

parameter. The base-station height characteristic in theMORITA equation and the Walfisch-Ikegami equation isnot expressed as a function of p log(h,), the buildingheight is assumed to beH o = 20 m as a general value, andthe base-station height characteristic coefficient p is shownin Fig. 9 with the base-station height as a parameter. Inthe Okum ura-Hata equation, p takes a large value whereasp in Wa lfisch-Ikegam i is small. In generation, p is -1 8 to-20.Figure 10 shows the building height characteristiccoefficient y with the transmitter-receiver distance as a

parameter. In the Sakagami equation, y = 7.5 as the sumof the coefficients of the building height H, along th estreet and the building height (H) around the mobilestation. In the M orita equation,y is expressed in terms ofthe base-station height and the transmitter-receiver dis-tance. In the Walfisch-Ikegami equation, y is th e valuefor the building height of Ho = 20 m.From Fig. 10, it is found that the values of y arescattered in the range of 7 to 80. The reason is that thedefinitions of the building height is different in differentestimation equations and is not uniquely determined unlike

the distance and the base-station height. The buildingheight in the Sakagami equation uses the height H, longthe street and the one for the buildings around the mobilestation ( I f ) . In the ca se of the W alfisch-Ikegami equation,the heights are identical in the area. In the Mo ritaequation, the averag e value generated by the simulation isused. In the present equation, the building height Ho isthe average of the buildings between the transmitter andreceiver.Table 7 shows the propagation characteristic coef-ficients in the case where the distance d = 1 km, the

base-station height h, = 30 m and the building heightHo = 20 m. To compare the relationships between coef-ficients, the values of a + /3 and a y are shown . In thestructural model, a + p and a y are 20. The value ofa + p is 21 in the Okumura-Hata equation, 16 in theSakagami equation, 29 in the Morita equation, 16 in theA-P equation, 17 in the Lee equation and is hence about20. If, for instance, the distance d is 1 to 3 km and thebase-station height h, is 20 to 60 m, the value of a + p

is 16 to 23 in the Okumura-Hata equation. The coef-ficients p and y in the Walfisch-Ikegami equation are largebecause of the functions of the base-station heightcharacteristic and the building height characteristic interms of log(] + h, - Ho). When the base-station heightcharacteristic coefficient p is derived, the function of thebase-station height characteristic is differentiated withrespect to log(hb). Hence, the slope f3 becomes large if h,and H, are close to each other. The value of a - yfluctuates from equation to equation.

The relationship between th e distance characteristiccoefficienta nd the base-station height characteristic p ineach estimation equation almost agrees with the structuralmodel. The building height characteristic coefficient yfluctuates from equation to equation and cannot becompared with the structural model.

6. ConclusionsWith a view to understanding the fundamentalpropagation characteristics in an urban area, the funda-mental propagation parameters are presented based on thepropagation loss data in 800 MHz in Tokyo. A model forthe propagation loss was established based on theseparam eters. The applicab le range of this model is whenthe base station is taller than the surrounding buildings.The relationships among characteristics obtained from themodel were compared with the conventional estimationequations and the regression equation obtained from themeasured data. The measured data are within the cell of

less than 1 km radius and at 800 MHz. The findings aresummarized below.( 1 ) The fundamental parameters of the propagation

loss in the urban area are the transm itter-receiver distance,base-station height and average building height betweenthe transmitter and receiver in addition to frequency.

2) In the proposed model, if the distancecharacteristic is 01 log(& then the base-station heightcharacteristic is (20 - a)lo g(h b) and the building heightcharacteristic is (a 20 ) log(Ho).

(3 ) The coefficients of the distance, base-stationheight and bu ilding height characteristics computed basedon the measured results in Tokyo are 38.2, -17.6 and 19.9.They agree almost with the relationships betweencharacteristics shown by the proposed modei. Also, theurban space coefficient I s about 18.

102


13/14

(4) When the conventional estimation equations andthe proposed model are compared, the relationshipsbetween he coefficients for the distance characteristic andthe base-station characteristic almost coincide.

From the foregoing results, it is found that thefundamental propagation structure in an urban a m an beexplained by the proposed m odel. Also, dependingon theestimation equations based on the model, the number ofpropagation parameters is small so that the propagationloss estimation in an urban microcell can be c a m 4 out.The model used here pays attention to the relation-ships among propagation characteristics. No physicalqualitative reasonings are provided, thus in the future,physical discussions are needed. Also, hemeasured dataused in the present study are a portion of those obtained

in Tokyo. More comparisons are needed at many meas-urement points. Further, in addition to the propagationstructure, it is necessary to extract effective propagationparameters from more measured data so that theestimation accuracy is improved.

Acknowledgement. The authors thank Dr.Kuramoto, Mr. Sakamoto, Dr. Hata of NTT RadioCommunication Systems Laboratories for guidance. Theyalso thank Mr. Sakagami, former staff member of "lTRadio Communication Systems Laboratories.

REFERENCES1. S.Kozono and A . Taguchi. Propagation character-istics of low base-station antenna on the urban road.I.E.I.C.E. Trans. (B-11), J72-B-II, 1 , pp. 3441(Jan. 1989).2. M. Kaji and A. Akeyama. UHF-band radio propa-gation characteristics in built-up areas for landmobile system using low antenna height base station

3 .

4.

5.

6.

7.

8.

9.

10.

1 1 .

12.

I.E.I.C.E. Trans. (B), J70-B, 5, pp . 6 0 2 6 1 0 ( M a y(1987).Y. Nagata, Y. Furuya, E. Moriyama, T.Saruwatari, I. Kamiya, and S. Hattori. 2-GHz bandradio propagation characteristics in microcellularmobile systems under tall building environments.Papers of Technical Group on Antennas and Propa-gation. I.E.I.C.E., Japan, AP90-84 (1990).Y. Okumura, E. Omori, T. Kawano, and K.Fukuda. Experimental study of propagation charac-teristics on the land mobile radio waves. Researchand Development Report, 16, 9, pp . 1705-1764(1967).H. Hata. Empirical formula for propagation loss inland mobile radio service. IEEE Trans. Veh.J. Walfisch and H. L. Bertoni. Theore tical modelof UHF propagation in urban environments. IEEETrans. Antennas & Propag., AP-36, 12, pp . 1788-1796 (1988).S. Sakagami and K. Kuboi. Mo bile propaga tion lossprediction for arbitrary urban environments.I.E.I.C.E. Trans (B-11), J74-B-II, 1 , pp . 17-25(Jan.1991).K. Morita, I. Higuchi, and S. Sato. Formula forestimatingmean propagation loss in quasi-microwaveurban mobilecommunications. I.E.I.C.E, 1991 FallNational Convention, B-9.Urban transmission loss models for mobile radio inthe 900- and 1800-MHz bands. EURO-COST 231TD (90) 119 Rev. 1 (Jan. 1991).F. Ikegami, S. Yoshida, T. Takeuchi, and M.Umehira. Propagation factors controlling meanfield strength on urban streets. IEEE Trans.Antennas & Propag., AP-32,8, pp . 822-829 (1984).A. Atefi and J. D. Parso n. Urban radio propaga tionin mobile radio frequency bands. IEEE Conf.W. C. Y. Lee. Mobile Communication En&' w r -ing, ch. 3, 4. McGraw-Hill, New York (1982).

Te~hnol. ,VT-29, 3, pp . 317-325 (1980).

Publ., 262, pp . 13-18 (1986).

103


14/14

AUTHORS (from left to right)

S h iNch i Ichitsubo graduated in 1985 from the Dept. of Communication Engineer, University of Electro-Comm unications, and obtained an M.S. rom Kyushu University in 1988. In that year he joined NTT Radio CommunicationSystems Laboratories. Since then, he has been engaged in research on radio propagation of m obile communication.

Masahiro Kimura graduated in 1983 from N agano Technical College and oined Nippon Telegraph and Telephone.Since then, h e has been engaged in constructio n and maintenance of radio facilities with the Japan Y outh Overseas TechnicalCooperative Corps, and research and development of mobile commun ication. Presently, he is with NTT RadioComm unication Systems Laboratories.

104

A Propagation Model for Mobile Radio Propagation Loss in an Urban Area at 800 MHz

Documents