Vous êtes sur la page 1sur 7

KIMURA and HORIKOSHI

IEICE TRANS. : PREDICTION


ELECTRON., OF MILLIMETER-WAVE
VOL. E82-C, NO. 7 JULY 1999 MULTIPATH PROPAGATION CHARACTERISTICS
1253

PAPER Special Issue on Microwave and Millimeter-Wave Technology

Prediction of Millimeter-Wave Multipath Propagation


Characteristics in Mobile Radio Environment

Kazunori KIMURA a) and Jun HORIKOSHI, Members

SUMMARY This paper presents a prediction of the millimeter- meter or above, we can effectively use the statistical param-
wave multipath propagation characteristics in the typical urban envi- eters such as height of obstacles and structure occupation-
ronment. To analyze the propagation in an outdoor environment, the
three dimensional model based on the geometrical optics and the uni- density in the area. On the other hand, in the case of propaga-
form geometrical theory of diffraction is employed. Prediction by the tion path length being short, prediction using the ray tracing
three dimensional ray tracing method needs a detailed map, which is effective. The ray tracing method may be also effective for
records locations and shapes of obstacles surrounding a transmitter and the prediction of the millimeter-wave from its large attenua-
a receiver. It is usually difficult to create a complete map because tre-
tion loss. Height of the transmitter, receiver and obstacles
mendous data is necessary to describe the area structure. We propose,
in this report, a three dimensional propagation model to predict the can not be neglected in this case, then we attempt to use the
millimeter wave propagation characteristics by using the information three dimensional (3D) ray tracing to predict those of the
available from only a map on the market. This approach gives us much millimeter wave. To accomplish this method, we must create
convenience in the actual design. The modeled results are demonstrated a map database which includes location and shapes of ob-
and furthermore comparison are made between the simulated results
and the experimental data. stacles with their complex dielectric constants. We propose a
key words: millimeter wave, power delay profile, propagation simple but effective simulation model that enables a system
modeling, 3D ray tracing, uniform geometrical theory of diffrac- designer to predict a propagation characteristics only from
tion the map information on the market.
To develop the proposed scheme we assume the next
1. Introduction three important principles in making our model.
(1) A reflection surface such as a wall, a rooftop and the
Revealing propagation characteristics is a most basic and in- ground is treated as the ideal plain surfaces. On the other
evitable work in designing a wireless communication sys- hand, a diffraction edge is treated as the ideal wedge.
tem. Number of experimental results and propagation mod- (2) Simulation error in the reflection loss and the diffrac-
eling have been reported on the indoor propagation charac- tion loss caused by irregularity of walls, wall-corners and
teristics of the millimeter wave, because it is recommended etc. can be compensated simply by the compensation coef-
world-widely to be applied to the indoor communications [1], ficient, which can be obtained by adding the equivalent re-
[2]. As the millimeter wave is also applicable in the outdoor flection waves around the reflection point or the diffraction
environment, its propagation characteristics have to be re- corner.
vealed for effective design of a broadband system. Unfortu- (3) There is only one reflection point or one diffraction
nately, there is only a few data for that purpose [3]. We, in corner on a ray path and three types of ray paths are consid-
this paper, try to reveal the multipath propagation character- ered as effective, i.e. the direct ray, the reflected ray and the
istics of the millimeter-wave in the urban area. To attain this diffracted ray.
purpose, we will establish a basic propagation model enabling In the latter part of this report, simulation results are
us to stand at some common view point on the outdoor propa- shown and effectiveness of the analysis is described. Proce-
gation characteristics of the millimeter wave. Prediction of dure creating the Map Information Database and newly
propagation characteristics is actively investigated from vari- developed 3D ray tracing model will be explained in Sects.
ous approaches in the UHF and microwave frequencies and 2 and 3, respectively. Simulated results and comparison with
many of them lead us to the excellent prediction. In spite of the measured ones will be demonstrated in Sect. 4. Consider-
them, it is usually difficult to obtain an effective information ations on the results are also described in Sect. 4. Conclu-
about the multipath propagation characteristics, even if it is sions will be stated in Sect. 5.
only in a rule of thumb, when a propagation distance is rather
small. It is well-known that when a system designer wants to 2. Map Information Database
predict those in a area of propagation distance about a kilo-
To pursue the 3D ray tracing calculation, we are necessary to
Manuscript received December 25, 1998. have 3D vector data of the obstacles such as a building wall,
Manuscript revised March 15, 1999.
a rooftop, a wall corner and a rooftop edge. The complex
The authors are with the Dept. of Electronic Engineering
University of Gunma, Kiryu-shi, 376-8515 Japan. dielectric constant of each environmental object are also
a) E-mail: kimz@jda-trdi.go.jp needed for us to obtain the electric filed as a composite one
IEICE TRANS. ELECTRON., VOL. E82-C, NO. 7 JULY 1999
1254
Building height in story

7
6
5
4
3
1,2

: Base Station 0m 100m 200m


Fig. 2 Vectors for the reflected ray.
: Guide Rail (Mobile Station)
: Direction of Movement
oped a 3D ray tracing program based on the Snells law and
Fig. 1 Maps created from the map information database. the Kellers law of edge diffraction. We assume in this analy-
sis that there is only one reflection point or one diffraction
that contains reflection and diffraction losses on each ray path.
point on a ray path and three types of ray path are considered
The database, consisted of 3D vector data and the complex
as effective, i.e. the direct ray, the reflected ray and the dif-
dielectric constant, is named as a Map Information Data-
base in the later. fracted ray.
The electric field of a direct wave or a ray path arriving
The 3D vector data are created from 1/1500-scale maps
directly at the receiving antenna depends only on the dis-
on the market, thus easy to obtain. Only the outlines and the
tance between the transmitter and the receiver and is given
positions of buildings are printed on these maps and they do
by
not offer a clear information about details of the building,
and many obstacles such as trees, temporary structures, sign- jkd
boards and etc. They are not printed enough on them. In or- ELos = K a(d) E 0 e (2)
d
der to predict the multipath propagation characteristics, sur- where k is a propagation constant, E0 is the emitted electric
vived ray paths are calculated from the map data only in this field strength at the transmitting antenna, K a( ) is the atmo-
study, i.e. with no investigation in the actual environment or spheric absorption loss and d is a distance between the trans-
field structure. mission antenna and the receiver.
The 3D vector data on the path obstacles are created as The electric field of a reflected ray path is obtained by
follows. Maps are scanned, and coordinates of the corners of
each building accompanied with its number of story are read. jk(d + d')
The corners location can be discriminated from the building's
E Ref = K r K a(d + d') E0 R e d + d' (3)
outline. Buildings are the unique obstacle available from the where d and d' are the distance between the transmitter and
map. Outline data of streets is not used to calculate both the the reflected point, and between the reflected point and the
ray path and the electric field, but only prepared for a clue to receiver, respectively. R is the Fresnel dyadic reflection coef-
identify the 3D vector data obtained from the original map. ficient when the incident ray arrives on a dielectric plain sur-
Figure 1 shows the scanned map presented to predict the face obliquely. R is defined as
propagation characteristics and at which the field measure-
ments were taken [4]. Height of buildings can be estimated R = R//e//i e//r + Rei er (4)
from the number of stories printed in the map by applying
R// = cos sin 2
2
the next relation (5)
cos + sin
H = [3.8N + 2.5] (m) (1)
R = cos sin 2
2
where N is a number of stories read from a map and [ ] repre- (6)
sents the rounding function to the nearest integer [6]. Height cos + sin
of the building is measured from the ground level and it is in which , ei , er and e//i , e//r are the incident angle, unit vec-
assumed that the ground level is flat with no rise and fall. tors perpendicular to the plain of incidence and the unit vec-
Buildings story N is noted in the map only if it exceeds three, tors parallel to the plain of incidence, respectively as shown
then we assume N=1.5 when it is one or two. in Fig. 2. is the complex dielectric constant. To calculate
the electric field of the reflected ray, we make our propaga-
3. 3D Ray Tracing Propagation model tion model as that building walls, their rooftops and the ground
are the only reflection objects. If they are the building walls
The multipath propagation characteristics such as delay pro- or rooftops, the buildings are assumed concrete structure and
files are predicted by calculating the electric field strength their complex dielectric constant can be set to =6.5 0.7j
(we denote it in short electric field in the later) for each ray [6]. If the reflected object is the ground, its complex dielec-
path, that can be obtained from 3D ray tracing results. In tric constant can be set to =1.9 0.4j [7].
order to find reflection and diffraction points, we have devel- In (3), Kr is the compensation coefficient for the reflec-
KIMURA and HORIKOSHI : PREDICTION OF MILLIMETER-WAVE MULTIPATH PROPAGATION CHARACTERISTICS
1255

tion loss at the reflected point. The reflection loss for the
+0 < d i < 2 (7)
actual surface of the building is larger than that of the ideal
plain surface, because surfaces of a building are considered
where d is the distance between the transmitter and the dif-
as a irregular surface at a wave length of the millimeter wave.
fracted point, d' is the distance between the diffracted point
To obtain the reflected field, detailed description of the build-
and the receiver, R is the Fresnel dyadic reflection coeffi-
ing surfaces is required as a millimeter wave length scale. As
cient, Kr is the compensation coefficient for the reflection
it is very hard to do this satisfactory in the actual surface, we
loss (Kr = 8 dB). The first term is the diffracted ray when
define the compensation coefficient for the reflection loss Kr .
the incident ray comes obliquely into the wedge of the smooth
In spite of the reflection loss depend generally on the inci-
and finite conductivity surface. The second term is 0 or the
dent angle, it can be assumed as Kr = 8 dB [8].
reflected ray when it is incident obliquely on the assumed
The electric field for a ray path generated from the dif-
surface. In the first term, D is the dyadic conductivity wedge
fracted point on the building edges is calculated by applying
diffraction coefficient and is given by
the uniform geometrical theory of diffraction [9]. Prediction
by the ideal wedge diffraction is sufficient to calculate on the i d i d
D = 0 0Ds Dh (8)
lower frequency band such as UHF and the microwave. How-
ever, irregularities of the edge is not negligible to predict the where Ds and Dh are the soft and hard scalar diffraction coef-
diffracted field using the theory of the ideal wedge diffrac- ficient [9], [10]. They are expressed by
tion, because the wave length of the millimeter wave is al- exp( j / 4)
most the same or less than the order of irregularities of the Ds,h =
2n 2k sin 0
building edge. Accordingly, we further assume that the re-
flected ray generated by irregularities of the building edge + ( d i )
exists besides the diffracted ray, when the transmitter and the + cot
2 n
[ +
F k L a ( )
d i
]
receiver are located in the line-of-sight region. This obscurely
reflected ray travels on the same path as the diffracted one. ( d i )
The exact condition of irregularities for the building edge is + cot
2 n
[
F k L a ( )
d i
]
necessary to calculate this reflected ray and its electric field
with accuracy. But it is very hard to do this sufficiently in the + ( d + i )
actual edge. To obtain this reflected ray and its electric field + Rn / / cot
2n
[ +
F k L a ( +
d i
]
in a easier manner, the dielectric plain surface on the edges
( d + i ) i
[ ]
are assumed and its electric field are calculated using the regu-
+ R0n/ / cot F k L a ( +
d
(9)
lar reflection theory as with the compensation coefficient ap- 2 n
peared just before. The electric field arriving at the receiver
by way of diffracted point on the edges can be obtained from where F[ ] is a Fresnel integral, L is a distance parameter
the next equation (geometry is depicted in Fig. 3), expressed by

L = si s d sin 2 i
i d
EDif = K a(d + d') E0 D d' e jk(d + d') (10)
d(d + d') d' s +s
and
jk(d + d')
+ K r K a(d + d') E0 R e d + d'
2nN ()
0 d i a () = 2cos 2 sin 2 i (11)
2
in which N + and N are the integers which most nearly sat-
isfy the equations [9], [10].
(2 - n)
2nN + () =
2nN () = (12)
Tx
= d i
R0 // and Rn // are the reflection coefficients for perpendicu-
lar or parallel polarization for the zero face and the n face.
Rx All the edges of a building, also rooftop edges, is in-
cluded in our analysis. The dielectric constant for the dif-
fracted ray is assumed as =6.5 0.7j similar to the reflected
ray.
The power delay profile can be obtained by the next
procedure: (1) Converting the above three electric field (i.e.,
direct ray, reflected ray and diffracted ray) into the received
Fig. 3 Vectors for the diffracted ray.
IEICE TRANS. ELECTRON., VOL. E82-C, NO. 7 JULY 1999
1256

power and then (2) Putting each received ray in order of its the mobile station moves with constant speed and on the other
arriving time. Let n be a number of the rays obtained from a hand, the mobile station is fixed at a specific point and the
pair of the transmitter and the receiver, the power delay pro- horn antenna is rotated with its angle of azimuth and eleva-
file is obtained as tion. The parameter of the measurement is shown in Table 1
2 and these are selected as the same used in the prediction.
n

Ei The predicted and measured power delay profiles using
P(t) = P0 Git Gir (t i) (13)
i=1 4 E0 the omnidirectional antenna is shown in Figs. 4 and 5. The
results of the prediction and the measurement are obtained
where P0 is the transmitted power, Ei is the electric field of under condition that the mobile station moves with constant
the i-th ray among the direct ray, the reflected ray and the speed and the omnidirectional antenna is used. In both plots,
diffracted ray. i is the delay time of i-th ray that calculated x, y and z-axes are the delay time, the distance between the
from its path length. Git is the antenna gain of the transmitter base and mobile station and the received power level, respec-
which depends on the direction of emission, Gir is the an- tively. The measured profiles, which are obtained in every
tenna gain of the receiver and denotes the delta function. 100 mm, is the average of 60 power delay profiles over the
100 mm interval. The analytical prediction of the power de-
4. Results and Discussion lay profiles are performed with the same situation as the
measurements. But it is usually difficult to compare the cal-
To estimate our proposing model, prediction of the power culated results with the measured one such as the delay time
delay profile should be made under the same condition with and the received power, because the predicted results are ob-
the measurement. The power delay profiles are measured by tained as like the impulse shape. Accordingly, in order to re-
using the auto-correlation characteristics of the pseudo ran- shape the predicted power delay profiles as like the measured
dom sequence modulated by BPSK. Resolution ability of this one, the impulse-shaped delay profiles (calculated by Eq. (13))
system is 13.6nsec in the time domain. The power delay pro- are convolved with the ideal impulse response of the pseudo
files are measured on the urban area (Setagaya Tokyo) as random sequence applied in the measurement. The predicted
shown in Fig. 1 [4], [11]. The measurement on this sites is power delay profiles are calculated in every 5 mm interval
made in the line-of-sight regions. The transmitting base sta- and we take an average of 50 power delay profiles over the
tion is set up on the roof of the five-storied building as shown 100 mm interval. As can be seen from the results of the pre-
by the black-circle plot in Fig. 1. Vertically polarized biconical diction and the measurement, the wave form of the predicted
antenna is used as the transmitting one. Its directivity is om- power delay profiles gives approximately agreement with the
nidirectional in the horizontal plane and 36 degree beam-width
in the perpendicular direction. The experimental antenna gain
is 0 dBi. The receiving mobile station runs a distance of 20
m. The course of the mobile station is depicted by the solid
Received level (dBm)

line also in Fig. 1. Two kinds of antennas are prepared for


receiving. One is omnidirectional (same one with
-60
the transmitting antenna) and the other is a vertically polar- 125
-80
ized horn antenna with a beam-width of 21 degree in azi-

)
-100
muth and 36 degree in elevation and with 19 dBi antenna (m
-120 120 e
nc
0
sta

gain. In the measurement using the omnidirectional antenna, 500


Di

Delay
time 1000
(nsec 115
) 1500
Table 1 Parameters for the prediction and the measurement of power
delay profiles. Fig. 4 Predicted power delay profiles using the omnidirectional an-
tenna.
Frequency 54 GHz band
Biconical anttena
Azimuth:Omnidirectional
Received level (dBm)

Base station
Elevation:36 degree
(Transmitter)
Antenna gain:0dBi
Height of antenna:20m
-60
Biconical anttena 125
-80
Azimuth:Omnidirectional
)
m

-100
e(

Elevation:36 degree -120


nc

Antenna gain:0dBi 0 120


sta

Mobile station Height of antenna:2m


Di

Delay500
(Receiver) Horn anttena time
Azimuth:21 degree (nsec1000
) 115
1500
Elevation:36 degree
Antenna gain:19 dBi Fig. 5 Measured power delay profiles using the omnidirectional an-
Height of antenna:2m
tenna.
KIMURA and HORIKOSHI : PREDICTION OF MILLIMETER-WAVE MULTIPATH PROPAGATION CHARACTERISTICS
1257

measured power delay profiles except for the delay signals at and the mobile station is 127.1 m and the angle of elevation
about 700 ns. Figures 6 and 7 show the average delay and the is 0 degree. In both plots, x, y and z-axes are the delay time,
delay spread distribution obtained from Figs. 4 and 5. In com- the angle of azimuth and the received power level. In the same
parison by the 90% value of the average delay, the predicted way as the results using the omnidirectional antenna, the de-
result is 13.0 nsec and the measured one is 9.9 nsec. In value lay signals at 700 ns are errors in the prediction. This delay
of the delay spread, the predicted result is 52.0 nsec and mea- signals arrive from about 160 degree direction. In addition
sured one is 46.0 nsec. The predicted average delay and de- to these errors, there are some errors in the prediction (at
lay spread give good coincidence with the measured results. 1400 nsec in the direction of 60 degree, and at 800 nsec in
In order to investigate the relation between the predicted the direction of 90 degree). But the delay time and the arrival
power delay profiles and the measured one in detail, the power direction of the predicted delay signals around the direct sig-
delay profiles are predicted and measured using the horn an- nal agree approximately with the measurement results. The
tenna when the mobile station is fixed. The horn antenna is predicted and measured direction patterns of the average de-
rotated to any direction in azimuth and elevation. The 0 de- lay is shown in Figs. 10, 11, and the direction patterns of the
gree direction of the antenna rotation is the forward direction delay spread are shown in Figs. 12, 13. In these plots, x and
of mobile moving for the azimuth angle. The angle of azi- y-axes are the angle of azimuth and the angle of elevation.
muth is varied clockwise from 180 degree to +180 degree. The z-axes are the average delay and the delay spread. In
And the 0 degree direction of the antenna rotation is the hori- these results, the predicted results agree approximately with
zontal plane for the elevation angle. The angle of elevation is the measured one for the direction of the peaks. But the pre-
varied upward from 30 degree to +75 degree. The predicted dicted value of the peaks are as large as the measured one,
and measured power delay profiles using the directional an- which suffers from prediction error of the delay signals.
tenna is shown in Figs. 8 and 9. These results are obtained We compare the power delay profile and the ray paths
under condition that the distance between the base station on this location and it is shown in Figs. 14 and 15. Figure 14
shows the map of the obstacles and the rays calculated by the
100 3D ray tracing program. It is projected on a x-y plane as a
folded 3D data. Figure 15 shows the comparison of the pre-
dicted power delay profile and the measured one using the
80 Predicted
omnidirectional antenna. The ray number (I-VII) in Fig.14
Cumulative probability (%)

corresponds to the peak number in Fig. 15. The errors in the


60 predicted delay signals at 700 nsec in Fig. 4, Figs. 8 and 15
Measured are caused by the ray path No.V. This delay signal dont
present in the measurement. We consider that the difference
40
between the prediction and the measurement about those de-

20
Received level (dBm)

0
0 10 20 30 40 50 -60
100
Average delay (nsec) -80
)
eg
-100
Fig. 6 Cumulative probability of average delay.
(d

0
-120
h
ut

0
im

Delay500 -100
Az

100 time
(nse1000
c)
Measured 1500
80 Fig. 8 Predicted power delay profiles using the directional antenna.
Cumulative probability (%)

60
Received level (dBm)

Predicted
40
-60
100
-80
g)

20 -100
e

0
(d

-120
h
ut

0
im

500 -100
Az

0 Delay
0 20 40 60 80 100 time 1000
(nsec 1500
Dealy spread (nsec) )
Fig. 7 Cumulative probability of delay spread. Fig. 9 Measured power delay profiles using the directional antenna.
IEICE TRANS. ELECTRON., VOL. E82-C, NO. 7 JULY 1999
1258

400
IV
Average delay (nsec)

350 VI
III Rx V
300 Tx I
250 II

200 VII
150
100

)
on (deg
50 60
0 30
-180

Elevati
-120 -60 0
0 60 -30
Azimuth (d 120 180
eg)

Fig. 10 Direction pattern of the predicted average delay. 0m 50m 100m

Fig. 14 Ray paths, shown as the distance between the transmitter and
400 the receiver is 127.1 m.
Average delay (nsec)

350
300
-80
250
Measured
200
I

Received level (dBm)


150 -90 Predicted
100
)
on (deg

50 60 II
-100
0 30 III V
-180 -120
Elevati

-60 0
0 60 -30
Azimuth (d 120
eg) 180 -110 IV

Fig. 11 Direction pattern of the measured average delay. VI VII


-120
400 0 500 1000 1500
Delay time (nsec)
Delay spread (nsec)

350
300 Fig. 15 Comparison of the predicted and the measured profiles.
250
200
lay signals results is resulted from the Map Information
150
Database. Other information expect location and features of
buildings are not included in the Map Information Data-
)

100
on (deg

50 base, because it is reconstructed from the maps on the mar-


60
0 30 ket. In the actual environment of the measurement, there are
Elevati

-180 -120 0 some trees around the base station and the mobile station. It
-60
Azimu0th (d 60 120 -30 may be reasonable to consider that one of those trees ob-
eg) 180
structs the ray path No. V and thus the delay signals at 700
Fig. 12 Direction pattern of the predicted delay spread. nsec can not be measured. This information is not included
in the Map Information Database. Accordingly, we can
400 claim that the prediction error is resulted from the unknown
factor which we can not find in the Map Information Data-
Delay spread (nsec)

350
300 base.
250
200 5. Conclusions
150
In this paper we present a propagation model for prediction
)

100
on (deg

50 60
of the millimeter wave power delay profiles in a urban site.
0 30 The main objective is to clarify what a modeling could be
-180
Elevati

-120 -60 0
0 proposed to predict the delay characteristics of the millime-
Azimuth (d 60 -30
eg) 120 180 ter-wave in the outdoor environment. We stand on a principle
that a map on the market is only available in predicting the
Fig. 13 Direction pattern of the measured delay spread. delay characteristics and try to make an effective propaga-
tion model based on the commercial map.
The Map Information Database is made using the
KIMURA and HORIKOSHI : PREDICTION OF MILLIMETER-WAVE MULTIPATH PROPAGATION CHARACTERISTICS
1259

above noted map and it is assumed that there is only one Kazunori Kimura received the B.E., M.E.
reflection point or one diffraction point on a ray path and degrees in electronic engineering form Gunma
university of Gunma, Japan in 1990 and 1992,
three types of ray path are considered as effective. The pre-
respectively. He joined Technical R&D Insti-
diction can be achieved by using the proposed propagation tute, JDA and has been engaged in the research
model. The power delay profiles around the direct signal show and development of radio communication sys-
the approximate coincidence with the measured results us- tems. In 1995, he was matriculated a doctoral
ing the omnidirectional and the directional antenna. Also it is course in Gunma university of Gunma, Japan,
clarified that the unknown particular obstacles affect seriously and now is engaged in the research of milli-
meter wave propagation.
on the prediction of the delay profiles when they are at the
near range, and careful considerations should be paid to this
countermeasure. It is a next work how we can make an effec-
tive map information database as possible as easily. Jun Horikoshi received B.E. and M.S.
degrees from Gunma University in 1966 and
1968, respectively and Dr. of Engineering
References from Tokyo Institute of Technology in 1988
He joined the Department of Electronic En-
[1] A.R.Tharek and J.P.McGeehan, Indoor propagation and bit gineering, Gunma University as an Assistant
error rate measurements at 60GHz using phase-locked oscilla- Professor and is now an Associate Professor.
tors, Proc. 38th IEEE VTC, pp.127-133, Philadelphia, 1988. He have been engaged in the research of the
[2] A.R.Tharek and J.P.McGeehan, Indoor propagation and bit digital mobile radios (especially development
error rate measurements at 60GHz using phase-locked oscilla- of digital modulation-demodulation scheme)
tors, Proc. 38th IEEE VTC, pp.127-133, Philadelphia, 1988. and radio wave propagation characteristics.
[3] L.M.Correia and J.R.Reis, Wideband characterization of the His main interest is a basic technology development for the high-speed
propagation channel for outdoors at 60 GHz, Proc. PIMRC personal mobile communications, e.g. clarification of the millimeter-
96, pp.752-755, Taipien, Oct. 1996. wave propagation characteristics and application of digital signal pro-
[4] K.Kimura, T.Abe, H.Sawasa, and S.Inoue, Measurement of cessing to the wireless communication area. Mr.Horikoshi is a member
propagation characteristics of millimeter-wave in mobile radio of the IEEE.
environment, IECIE Gen. Conf. 97, B-1-27, 1997.
[5] T.Takeuchi, T.Hano, S.Yoshida, and F.Ikegami, A feasibility
study on theorical prediction of multipath delay profiles in ur-
ban mobile radio environments, IECIE Trans., vol.J73-B-II,
no.11, pp.779-785, Nov. 1990.
[6] K.Sato, T.Manabe, T.Ihara, Y.Kasashima, and K.Yamaki, Mea-
surement of reflectivities of interior construction materials at
60GHz, IEICE Technical Report, AP93-36, 1993.
[7] J.Awaka, H.Yamasaki, A.Takahashi, K.Okamoto, and T.Ihara,
Scattering properties of soil measured by a 60GHz
scatterometer, IEICE Technical Report, AP90-120, 1991.
[8] A.Satoh and E.Ogawa, An evaluation method for the reflec-
tion coefficient of building walls, IECIE Trans., vol.J72-B-II,
no.5, pp.207-217, May 1989.
[9] R.G.Kouyoumjian and P.H.Pathak, A uniform geometrical
theory of diffraction for an edge in a perfectly conducting sur-
face, Proc. IEEE, vol.62, no.11, pp.1448-1461, Nov. 1974.
[10] A . G. K a n at a s , I . D . Ko u n t o u r i s , G. B. Ko s t a r a s , a n d
P.Constantinou, A UTD propagation model in urban
microcellular environments, IEEE Trans. Veh. Technol., vol.46,
no.1, pp.185-193, Feb. 1997.
[11] K.Kimura and J.Horikoshi, Millimeter-wave multipath propa-
gation delay characteristics in the outdoor mobile radio envi-
ronment, IECIE Trans. Commun., vol.E81-B, no.8, pp.1696-
1699, Aug. 1998.

Vous aimerez peut-être aussi