Sie sind auf Seite 1von 5

358 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. VT-26, NO.

4, NOVEMBER 1977
The Attenuation of UHF Radio Signals by Houses
PAUL I . WELLS, MEMBER, IEEE
Abstruct-The results of a measurement program that was con-
ducted to determine the attenuation of UHF radio signals penetrating
to the inside of a typical house is presented. This program is part of a
study to determine the feasibility of using direct satellite communica-
tion to disseminate disaster warning messages. The measurements were
made in a manner to determine the building attenuation as a function
of frequency, construction type, climate, and the elevation angle to the
signal source. Attenuation measurements were made in five cities:
Boulder, CO, Duluth, MN, Kansas City, MO Little Rock, AR, and Hous-
ton, TX. The measurements were made at three frequencies, 860 MHz,
1.550 MHz, and 2569 MHz, using the ATSd geosynchronous satellite as
a signal source. Most measurements were made on two principal house
types; wood frame with a wood outside surface, and wood-frame with
a brick-veneer outside surface. The average measured building attenua-
tion for all houses and al lfrequencies was 6.3 dB.
INTRODUCTION
T
HE MEASUREMENT program discussed in this paper is
one part of a broader study to assess the feasibility of
using a geosynchronous satellite for direct broadcast of warn-
ings of impending natural disaster such as tornadoes, floods,
etc. The feasibility study is being conducted by the National
Aeronautics and Space Administration (NASA). The purpose
of this measurement program was to study the additional
attenuation of ultra-high frequency (UHF) radio signals pene-
trating to the inside of a typical house.
Building attenuation measurements were made at three
frequencies in the UHF band, 860 MHz, 1550 MHz, and 2569
MHz, using the Applications Technology Satellite No. 6 (ATS-
6) as a signal source. Measurements were made between Octo-
ber 1974 and J une 1975 on several typical wood-frame wood-
siding, and several wood-frame brick-veneer houses in each of
five geographical locations. Measurements were also made on
two houses during the period the ATS-6 was being moved
from its geostationary position of 94"W longitude to a new
position over East Africa.
The measurements were made by sampling the signal level
at 16 locations outside the house and then sampling the signal
level at 16 locations in each of two rooms in the house. Due to
satellite scheduling constraints, measurements were not made
at all three frequencies on all houses. The data samples were
then compared using statistical analysis techniques to deter-
mine the average attenuation and the standard deviation of
the data samples. A more complete discussion of the measure-
ment program and the statistical analysis techniques used to
analyze the data have been reported elsewhere [ 11 .
Manuscript received May 16, 1977.
The author is with the Institute for Telecommunication Sciences,
Office of Telecommunications, U.S. Department of Commerce, Boul-
der, CO 80302.
TABLE I
CITY LOCATIONS AND ANGLES TO ATS-6
City Locati on Angl e -. t c ATS- 6
E:eva:lon AZl rnut:i
Dul uth, MS 92'11'Ti' 36. 1' 182. 5'
46O50' N
Uoul der, CO 1 0 5 1 3 ' W 4 2 . 3 ' 1 6 2 . 9 O
40'00'N
Kansas C i t y, 1.40 94O35'W 4 4 . 7 '
39' 06' N
1 7 9 . 1 O
Li t t l e Rock, AX 92'13'15 4 9 . 1 3 ~ 1 8 3 . 1
3 4 O2 3 ' N
i i ouston, TX
29 38' K
95O16'1.J 5 5 . 4 0 1 7 7 . 4 O
BUILDING ATTENUATION MEASUREMENTS
Operating time on the ATS-6 satellite was made available
for the building attenuation measurements by the Goddard
Space Flight Center (GSFC) of NASA. The schedule for
operating time was coordinated with the ATS Operations Con-
trol Center (ATSOCC) in the ATS Project Office at GSFC.
Five cities in quite different climatic regions were selected
for the measurements. These are listed in Table I along with
the geographic coordinates and elevation and azimuth angles
to the ATS-6 signal source.
In this study, the measurements were limited to the single-
family class of structure. Within this class of structure there
are a number of basic types of building materials used and
different combinations of these materials. Also, there are
many variations in building styles, number of floors, split-
level configurations, etc. The most common combinations of
basic structural materials are
1. wood frame with wood (or similar) siding;
2. wood frame with brick veneer;
3. wood frame with metal siding (primarily on older,
4. solid masonry walls (brick, cement block, stone, etc.);
5. mobile homes, usually wood frame with sheet aluminum
remodeled homes);
siding.
Of these basic structure types the two most common are
the wood frame with a wood or similar material exterior, and
the wood frame with a brick veneer. Since the scope of the
measurements were necessarily limited, these two structure
types were selected for attenuation measurement in al l the
selected cities. Only a few samples of other types were
measured.
It is noted in Table I that the elevation angle from the se-
lected cities to the ATS-6 vanes from about 36" to 55". How-
WELLS: ATTENUATI ON OF UHF SIGNALS BY HOUSES 359
TABLE I1
ATS-6 DOWNLI NK POWER CALCULATI ONS
Freq.
Power E I W Loss
(MHz) (watts) ( dBw) (dB)
Path Folarlzation
LOSS
dB
Available
(dB!-I)
860 80 51 1 8 2 . 3 3 - 104.3
1550 40 51 187.4 3 -109.4
2569 1 2 53 192. 2 3 - 112. 2
ever, any effect due to elevation angle would be so entwined
with climatic and regional effects as to be inseparable.
It was determined that the best opportunity for evaluating
the effect of changes in elevation angle would be to make re-
peated measurements on one or more houses in Houston
during the move of the ATS-6 from 94"W longitude eastward
to a new location over Eastern Africa.
In considering the possible wave propagation aspects of the
changing elevation angle it was thought that the attenuation
by trees at lower elevation angles may have a far greater effect
than the penetration loss of the house. With this in mind, two
houses were selected in Houston for measuring the effects of
the change in elevation angle. Both houses were of the same
basic structure type, wood-frame with brick-veneer exterior.
The two houses were less than 2 kmapart: one was in an open
area with no nearby trees, and the second was well immersed
in trees whch were 15 to 2 1 m (50 to 70 ft) tall.
As was stated above, the ATS-6 satellite was used as a signal
source for the building attenuation measurements. The trans-
mitting antenna on the ATS-6 satellite, which was used for
these measurements, was a parabolic reflector 9.14 m (30 ft)
in diameter. The transmitter power output at each frequency
and the downlink power calculations in effective radiated
power are shown in Table 11.
The transmission from the ATS-6 was right-hand circularly
polarized at all three frequencies. The receiving antennas used
for the building attenuation measurements were, in all cases,
linearly polarized. A linearly polarized antenna will intercept
only one half of the power available in the circularly polarized
field. This 3-dB polarization loss is accounted for in Table 11.
The primary antenna used for the building attenuation
measurements was a vertically polarized conical monopole
above a ground plane. The diameter of the ground plane was
approximately one wavelength at the lowest operating fre-
quency, 860 MHz.
The antenna impedance was measured over a frequency
range from 500 MHz to 4000 MHz using a swept frequency
and an automatic impedance analyzer. At the operating fre-
quencies, the VSWR was 1.7 at 860 MHz, 1.3 at 1550 MHz,
and 1.15 at 2569 MHz. The antenna pattern was measured at
each operating frequency. In the horizontal or H-plane the
gain was constant to within 20.3 dB at 860 MHz and 1550
MHz, and to within ?1 dB at 2569 MHz.
The vertical or E-plane, pattern at 860 MHz is shown in Fig.
1 with a peak power gain of 5 dB as shown. The E-plane
patterns at 1550 MHz and 2569 MHz were similar in shape and
had peak power gains of 5 dB and 3 dB, respectively.
A secondary antenna having horizontal polarization was
used at 860 MHz during the building attenuation measure-
I
\
'\
90
Fig. 1. Conical monopole antenna, E-plane pattern at 860 MHz. The
power gain is relative to isotropic.
ments. This antenna consisted of four dipoles properly spaced
on a printed circuit board. The dipoles were positioned at
0.475 wavelengths above a ground plane which was about 1.7
wavelengths in diameter. At 860 MHz, the VSWR was approxi-
mately 1.8. The antenna pattern was measured at 860 MHz.
In the horizontal or E-plane the gain was constant within
k0.3 dB. The vertical or H-plane pattern was similar to Fig. 1
and had a peak power gain of 4 dB.
In the measurement of the level of the received signal for
determining the building attenuation two levels of sampling
were used. First, since the received signal level is different at
different antenna locations it was necessary to measure the
received signal at a number of locations both outside the
house and inside each room selected. The analysis of prelimi-
nary measurement data showed that any number of samples
greater than ten was adequate. Measurements were made at 16
antenna positions, two antenna heights at each of eight loca-
tions. The same locations and antenna heights were used for
each operating frequency.
The second level of sampling was the number of data points
read at each location. Since the instantaneous output of the
receiver can vary due to such causes as receiver noise, a car
passing in the street, etc., eight data samples at one second in-
tervals were recorded at each antenna location. In the data
processing, the eight data samples were averaged to determine
the level of the received signal at each antenna position.
The measurement procedure discussed above was repeated
for each frequency scheduled for that house. To the degree
possible, a random order of the sequence for making measure-
ments at the three frequencies was used. The use of a random
sequence of measurement reduced the likelihood of the data
being biased by measurement procedure or equipment behav-
ior. The number of frequencies measured was dependent on
360 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. VT-26, NO. 4, NOVEMBER 1977
the amount of satellite time scheduled for that individual
house. The same antenna positions and heights were used at
all frequencies.
Prior to making measurements on a house, the receiving and
data recording system was calibrated using a signal generator.
The signal generator was connected to the system at the
antenna terminal, thus all components except the antenna
were included in the calibration. The output level of the signal
generator was verified each time using a calibrated power
meter. The receiving system was calibrated over the range from
the minimum detectable signal to -80 dBm.
EXPERIMENT RESULTS
Repeat measurements of building attenuation were made in
13 rooms in six different buildings. Repeat measurements were
made primarily at frequencies of 2569 MHz and 1550 MHz,
but one set was run at 860 MHz, vertically polarized, and 860
MHz, horizontally polarized. The pooled standard deviation
within each frequency was computed. There was no evidence
of a difference due to frequency so an overall pooled stand-
ard deviation of 1.5 dB was computed. By this we mean that
the random error in measuring the attenuation of a room has
an estimated standard deviation of 1.5 dB.
Each of the 203 measurements is classified according to
city (5), frequency (4), construction type of building (5), and
type of insulation used in the building (7). Each room was also
classified according to effective window surface area (six
groups) and degree of exposure (nine groups); i.e., classified by
degree of sheltering by other parts of the building. The roof
conditions of the building were also the basis for classification.
Snow (three levels) and rain (three levels) were considered.
In the analysis of these data, the statistical technique of
analysis of variance was used. The purpose of the analysis was
to determine the relative contribution of each of the eight
factors listed above to the overall variability of the measured
building attenuation data. Thus the initial analysis design is an
eight-factor analysis of variance (ANOVA). However, it is
badly unbalanced with many combinations of factor levels
missing due to practical reasons ranging from budget limita-
tions to the uncooperativeness of mother nature and the fact
that construction and insulation types may be different in
different climate areas.
Analysis of the data was made possible by Rummage, a
general linear model computer package written by Del Scott
of the Pennsylvania State University and implemented at the
National Bureau of Standards in Boulder, CO, by J anet R.
Donaldson of the NBS Statistical Engineering Laboratory.
Rummage, which is capable of analyzing such unbalanced
designs, is a modified form of the MAD package [2] .
Initial analysis of all the data and all eight factors showed
that all factors except room aperture, snow, and rain had
statistically significant effects on building attenuation. It
should not be concluded, however, that rain or snow on a roof
do not have an effect since the number of measurements with
anything but dry roofs was very small. As always, failure to
reject a null hypothesis of no effect can only be interpreted as
an indication of insufficient evidence and not that the hypoth-
esis is true.
The adjusted R 2 (the combined correlation function
squared) for the eight-factor linear model was 0.73. That is, 73
percent of the observed variability in the attenuation measure-
ments is explained by the eight-factor linear model. The resid-
ual standard deviation of the measured attenuations about the
fitted model was 2.4 dB; this is the combined effect of the
variability of the measurements and the variability of indi-
vidual buildings. Since it is only slightly larger than the 1.5-dB
standard deviation estimated for the variability of the measure-
ments, we conclude that this model represents this set of data
rather well.
Because of the numerous missing factor level combinations,
analysis of all data to obtain quantitative estimates of the
effect of each factor is impossible. A subset of the data was
selected by omitting data from building and insulation types
that were isolated cases. These are considered separately. Re-
tained in the data set were only two building types (wood
siding and brick veneer), two insulation types (blown-in ceiling
and ceiling and walls, no metal), and two room exposure types
(one or more exposed walls and no exposed walls). The factor
city is highly confounded with insulation type and could
not be used in the model since the insulation type was differ-
ent in the different cities.
Analysis of the subset of 147 data points using the reduced
four-factor ANOVA model [frequency (4 levels), construction
type (2 levels), insulation type (2 levels), and room position (2
levels)] shows that all but room position are statistically signi-
ficant (at the 0.05 level of significance). However, this factor
was retained in the model, and the effect estimated since, with
the finer classification, it was shown to be significant. The esti-
mated effects are given in Table 111.
Table I11 can be interpreted in the following way. The aver-
age attenuation is 6.3 dB. A frequency of 2569 MHz exhibits
1.16 - 0.39 =0.77 dB more attenuation than a frequency of
1550 MHz, and wood houses exhibit 1 .I 6 dB less attenuation
than brick-veneer houses. The predicted attenuation for a
room having any combination of levels can be obtained by
summing the effects. For example, the predicted attenuation
at 2569 MHz is 9.14 dB into an unexposed room in a brick-
veneer house with insulation in the ceiling and walls (worst
case). The standard deviation of this predicted value is 0.23
dB. The predicted attenuation for the opposite extreme condi-
tion (exposed room, wood siding, blown-in ceiling insulation,
860-MHz vertical polarization) is 2.93 dB with a standard de-
viation of 0.27 dB.
The residual standard deviation for this model is 3 dB,
which is consistent with the 2.4 dB for the full model. The
interpretation that should be placed on it is that measurements
on individual buildings of any one type will vary with a stan-
dard deviation of 3 dB. The variation is due both to random
measurement error and to the variabhty of the buildings. For
example, the predicted attenuation for the worst-case building
above was 9.14 2 0.46 dB (+2 standard deviations). To ac-
count for the variability among buildings, we might add *2 X 3
dB to cover roughly 95 percent of all buildings of this type,
WELLS: ATTENUATION OF UHF SIGNALS BY HOUSES
TABLE 111
ESTIMATED EFFECTS
r.-.'sra?e 6.30 dB
I'regLI*.::; 2563
1553
1.16
0.39
-1.69
0.14
- 0. 58
0. 58
- 0. 80
.,1__5 0. 80
- 0. 30
0. 30
855 7
865 !I
Cons:r-z'L;=? ;.;3t5. s :i:rc
Srl ci ';i7Ber
1n;al ar.c-. Blo:.r-ln c2111r.c:
Gel?;:.; 334 1 -
a30- ?c; l t l c- . ?:,." ,~s:? ::3::5
I, ^ _.i - :.x;csei . , ; z7. 1s
implying that an attenuation of 9.14 +[(0.46)2 +(6)2]1/2 =
15.2 dB should be allowed.
HIGH ATTENUATION CASES
One class of structure that exhibited high attenuation was
the mobile home. The attenuations observed for two mobile
homes averaged 23.6 dB, but other attempts to measure
mobile homes were not successful because the signal could not
be detected inside the structure. All that could be determined
is that the attenuation exceeded about 25 dB.
Measurements were made on two wood-frame wood-exter-
ior houses in Kansas City that exhibited high attenuation. On
these houses the interior walls and ceiling were of Sheetrock
backed with aluminum foil. These two houses exhibited an
average attenuation of 17.1 dB. This is about 11-dB higher
attenuation than other houses of the same basic structure
type.
ELEVATION ANGLE
As discussed earlier, attenuation measurements were made
on two selected houses in Houston, TX, during the transient
period of the ATS-6. The measurements were made at 1550
MHz.
The results of these measurements are shown in Figs. 2 and
3. Fig. 2 shows the level of the received signal outside the two
houses. The data points shown at 55" elevation angle were
data taken prior to the start of the move of the ATS-6. It is
noted that there is some decrease in outside signal level at the
house without trees as the elevation angle decreases. However,
it should be noted on Fig. 1 that the relative gain of the receiv-
ing antenna is about +5 dB at 50" to 60" elevation angle and
decreases to about 0 dB at 0". Thus most of the decrease in
received signal level is explained by the decrease in antenna
gain.
The important information in Fig. 2 is the relative differ-
ence in the level of the received signal outside the two houses,
one with no trees and the other surrounded by large trees. The
datz indicate a difference of 12 to 15 dB. On the final day of
measurement with an elevation angle about 3", the level of the
signal in the trees was below the receiving system threshold,
which was about -128 dBm.
Fig. 3 shows the building attenuation of the house without
trees as a function of elevation angle. Each point shown is the
a
Fig. 2 . Outside measurements at Houston, TX? during transit of
ATS-6.
L- L"- . - L!: - - --- . ; sGLE. 3E:EESZ
Fig. 3. Building attenuation versus elevation angle for the house
without trees.
average of the measured attenuation in two rooms in the
house. It can be seen that for this hosse the elevation angle is
not a significant factor. In the Houston house in the trees, the
initial building attenuation measurement at 55" elevation angle
gave an average of 10.3 dB attenuation. On subsequent mea-
surement attempts, the level of the signal inside the house was
so low that reliable attenuation data could not be obtained.
CONCLUSIONS
In this measurement program we have studied the attenua-
tion of UHF radio signals by single-family residence type
buildings. The measurements were designed to determine if the
building attenuation is affected by such factors as frequency,
construction characteristics, elevation angle of the signal
source, and climate. From these data, several conclusions can
be drawn.
1) The position of a room within a house caused a very
minor effect on the received signal level, an average of 0.6 dB.
2 ) The basic exterior construction material had a small but
significant effect on the measured attenuation with brick-
362 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. VT-26, NO. 4, NOVEMBER 1977
veneer houses having about 1.2-dB higher attenuation than the
houses with wood siding.
3) The manner in whch the building was insulated had a
significant effect on the measured attenuation; houses with
insulation in the ceiling and walls had about 1.6 dB more
attenuation than those with insulation in the ceiling only.
Several other variations of insulation type and combinations of
insulation types were encountered; however, not in sufficient
numbers to test the significance of their effect on the building
attenuation. The use of Sheetrock with aluminum backing was
cited as possibly causing high attenuation. The measured
average attenuation for the two houses having aluminum-
backed Sheetrock ranged from 14.6 dB to 22 dB, whch could
be very important in areas where it is widely used.
4) It is readily apparent in Table I11 that the frequency of
the radio signal has a significant effect on the building atten-
uation with the attenuation increasing as the frequency
increases. The average measured atttenuation at 2569 MHz was
about 2.9 dB greater than at 860 MHz. Also, the polarization
of the receiving antenna was significant. At 860 MHz the
average measured attenuation with horizontal polarization
was about 1.8 dB greater than with vertical polarization.
REFERENCES
[ l ] P. I . Wells and P. V. Tryon, The attenuation of UHF radio signals
by houses, Office of Telecommunications Report 76-98, NTIS
Accession No. PB 258447/AS, Aug. 1976.
[2] G. R. Bryce,MAD: An analysis of variance program for unba-
lanced designs,Appl. Statistics, vol. 24,110. 3,p. 350,Mar. 1975.
New Radio Paging System and Its Propagation
Characteristics
MITSURU KOMURA, MEMBER, IEEE, TADATOSHI HAGIHIRA, AND MASANOM OGASAWARA
Abstract-A radio paging system using a 200 band NRZdigital code
for the selective calling signal on new frequency bands is discussed. This
system uses the 250-MHz band, and its propagation characteristics in
urban areas, necessary for the radio system design, was measured in the
Tokyo area. Distance versus median field strength characteristics in this
band approximate that of the 150-MHz band. Building loss is 19.7 dB
and is less than that of the 150-MHz band. Therefore, the new system
will provide nearly the same grade of service as the former system using
the 150-MHz band.
I. INTRODUCTION
N J APAN, The Pocket Bell radio paging service is now
spreading over 49 urban areas with a total of about 600 000
subscribers by 1977. This system uses the 150-Mhz band.
However, this band is so crowded with radio services that new
frequency allocations cannot be made, in spite of a great
demand for this service. Therefore, a new radio paging system,
New Pocket Bell, was developed using the 250-MHz band.
This paper describes the system outline and results of some
prototype model field tests, which were carried out in the
Tokyo area in 1976.
I
Manuscript received March 8,1977;revised May 15, 1977.
The authors are with the Mobile Radio Section, Engineering Bureau,
Nippon Telegraph and Telephone Public Corporation, Chiyoda-Ku,
Tokyo 100, Japan.
11. SYSTEM OUTLINE
Pocket Bell service automatically pages a subscriber who
is out of his office by sending him beep sounds via radio to
contact him at any time in the service area. The New Pocket
Bell gives the same service in addition to the following.
1) This system provides a three times increase in subscriber
2) This system adds several functions.
3) The system covers a larger area.
capacity.
To realize these characteristics, the following techniques were
adopted:
a) 200-band signal is introduced as a selective calling signal
b) dual-call, battery-saving, and two-alarm signal levels;
c) frequency offset techniques;
d) automatic compensation for any phase-shift in calling
signal occurring on the transmission path between base
stations.
instead of two sequential double tones;
Fig. 1 shows offset frequency versus calling rate. In this
system, the radio offset frequency is adjusted to either 250 Hz
or 500 Hz. Typical specifications for this system are shown in
Table I.

Das könnte Ihnen auch gefallen