WPC!0
27B RZX#|DѾ(HP LaserJet IID (Additional)HPLAIIAD.PRSx
@,tS%mX@"^:LPt:YYt:M:?tttttttttt??q嗙Zr۫ӣd>dt:w}fhJqEH|Eu~cdOjohettt2nisv"^2ACd2LLd2B26dddddddddd66bę|LXxmęA6Ad2bjXlbBam65h6mmmlLLAmZXYWddd"^2AEd2LLd2B26dddddddddd66aĂxMb|t~~V5Vd2fkWnY@aq;>j;qdnlUVDo[_YWddd"^)68S)@@S)7)-SSSSSSSSSS--Rqvtg?I{kd[mv6-6S)RXIZR7Q[-,V-[[[Z@?6[KpIJHSSSa8DocumentgDocument Style StyleXX` ` `
2pkk{a4DocumentgDocument Style Style. a6DocumentgDocument Style StyleGX
a5DocumentgDocument Style Style}X(#
a2DocumentgDocument Style Style<o
?A.
2vt3 Aa7DocumentgDocument Style StyleyXX` ` (#`
BibliogrphyBibliography:X
(#
a1Right ParRight-Aligned Paragraph Numbers:`S@I.
X(#
a2Right ParRight-Aligned Paragraph Numbers C @` A. ` ` (#`
2
i
a3DocumentgDocument Style Style
B
b
?1.
a3Right ParRight-Aligned Paragraph NumbersL!
` ` @P
1. ` ` (#
a4Right ParRight-Aligned Paragraph NumbersUj` ` @a. ` (#
a5Right ParRight-Aligned Paragraph Numbers
_o` ` @h(1) hh#(#h
2L
a6Right ParRight-Aligned Paragraph Numbersh` ` hh#@$(a) hh#((#
a7Right ParRight-Aligned Paragraph NumberspfJ` ` hh#(@*i) (h-(#
a8Right ParRight-Aligned Paragraph NumbersyW"3!` ` hh#(-@p/a) -pp2(#p
a1DocumentgDocument Style StyleXqq
l^)I. ׃
2+~0Doc InitInitialize Document Style
0*0* I. A. 1. a.(1)(a) i) a) I. 1. A. a.(1)(a) i) a)DocumentgTech InitInitialize Technical Style.
k I. A. 1. a.(1)(a) i) a) 1 .1 .1 .1 .1 .1 .1 .1 Technicala5TechnicalTechnical Document Style)WD(1) . a6TechnicalTechnical Document Style)D(a) . 2>a2TechnicalTechnical Document Style<6
?A.
a3TechnicalTechnical Document Style9Wg
21.
a4TechnicalTechnical Document Style8bv{2a.
a1TechnicalTechnical Document StyleF!<
?I.
2l^a7TechnicalTechnical Document Style(@Di) . a8TechnicalTechnical Document Style(Da) . PleadingHeader for numbered pleading paperP@n $]XX` hp x (#%'0*,.8135@8:!aBAAA..'A6Q554<<G/!/x<<91<0#5@# :]@8;9/-!@3K62.<<=<(,!A5O10/<<23,3.'+3*:(%*G$$"$ &"8&"#"&- $$$GG2|!hr "^7$C:$;$$$$$$$$$$;;;#F713:2,::5.D7:+:7'/;7F773G$#&'##'%:''''' 0 $$$GG"^7$C3$;$$$$$$$$$$;;;!F1/260+9700?57+70'-4.?1,1G$$$ &&$:'%$$' /$$$GG"^ J0ZO%%0P 0000000000PPP/_JBDNC[JN:NJ5?OJ_JJE `0/3*40 /52N5554%% 5,A++*000``"^!J0PH&&0P 0000000000PPP/_<>?E?;HI$+==TCE;E=5:D9O729&&`00.(0&+3.K3-/.&$3)<+(%000``2#!3""=#"^ J0ZE%%0P 0000000000PPP,_B?CIA:NK#%A@THJ:JA5>P(4(,PPPPPPPPPP,,Inhpxl`|:=lkw{`{kWduhmbmD,DP(QQGTD/DU*)P)VSQP5;,VGjB@?PPP"^(/7{Pv(??P(4(,PPPPPPPPPP,,Ndfhsibvx;GeeorarfW`q_ZS_?,?P(PLBO@.GU.*M)|UKNL><,UDdHC>PPP"^(47{P}}(==P(4(+PPPPPPPPPP,,Mhiovi`x|>Nyeu}c}p\dufpeiD+DP(RUFXG3NZ/2U/ZPXVDD6YIgLGEPPP2/#y$$j
%"^')\<^^...+*000``?xxx,Nx6X@8;X@
N@N:,Ӽ4L p# ,
K7qC2,*Xq4L P: XP
L7oC2,Xo4L p# X,
Oy.^8),*j^4L P: PZ?xxx,GJx@g
1 @Zi&HHH,GH@g
1 @Z3```,G~?`@g
1 ?@ZL,GjTa@g
1 a@>?xxx,ax `w;X,
W!D(,*BhD4L P: hP,
Mt,Z5(,*mZ4L P: P,
BP<,*@4L P: Ps4ddd,tCd6X@8;@j
P7mC2,-Xm4L x- XX
Q7jC2,r$Xj4L x7R Xj
SW!A(,NhA4L x- hX
RW!@(,h@4L x7R h,
4(,*(4L P: Pj
4','4L x- X
4&,&4L x7R ,
F7 ,*h"74L P: "Pj
F5 ,Eg"54L x- "X
F3 ,"34L x7R "j
t,W5(,eW4L x- X
t,U5(,U4L x7R
t,Y5(,dY4L p#
W!C(,IhC4L p# h
4(,y(4L p#
F6 ,`"64L p# "x0@"H20Xa#4L p# Ӽ#July 1992`T$pDOC: IEEE P802.1192/83Ѓ
xRyO#^4L P: jP#
SubmissionHPage ՟
#Xq4L P: XP#5*Wideband Propagation Measurements for Wireless Indoor Communication
!
8Peter B. Papazian and Robert J. Achatz
>U.S. Department of Commerce
V .National Telecommunications and Information Administration
7Institute for Telecommunication Sciences
E325 Broadway
r=Boulder, Colorado 80303 USA
B>Telephone: (303) 4973498
X
Introduction
Because of the importance of wireless local area networks to the U.S. economy, the
Institute for Telecommunication Sciences (ITS) has implemented a wideband measurement
system to aid in standards development. This report presents a preliminary view of indoor
channel propagation measurements and the metrology used in collection and analysis of indoor
multipath interference data. At present the development of acquisition and analysis metrology
is not complete. It is hoped that cooperative efforts with industry can be used to guide further
Xdevelopments.
XHMeasurement System
X<<d
dddd
The ITS wideband probe is used to measure indoor channel propagation characteristics.
The system, which has a bandwidth of 1 GHz, has previously been used for outdoor millimeter
wave propagation experiments in urban and vegetated settings as described in [1], [2], and [3].
For indoor measurements, the probe was modified to transmit a 1.5 GHz carrier with BPSK
modulation. The modulation code is a 127 bit pseudo random sequence with bit rate adjustable
between 100 and 500MB/s. Wideband, vertically polarized, omnidirectional antennas were used
at the transmitter and receiver. The system measures cophase and quadphase channel response
using a sliding correlator. By adjusting the sliding correlator's clock offset frequency, channel
responses were measured every 100 ms and stored on a digital analog tape recorded (DAT) for
later computer processing. A block diagram of the system is shown in Figures 1 and 2. The#0*$$detector sensitivity is limited by its correlation noise level. The detector's dynamic range is
determined by the 127 bit PN word length and is 42 dB. Dynamic range of the system is determined by the DAT, which employed a 16 bit A/D converter with a range of 90 dB. By using
variable gain video amplifiers in the DAT, the signal voltage was maintained above the noise
floor at all measurement sites.
X.Data Processing
X
dddd
<d<dThe DAT medium allows complete storage of cophase and quadphase channel responses
for 2 hours of continuous measurements. In our tests, the receiver was stationary while the
transmitter was moved on a cart at a constant speed along a predetermined path. Data are
X
stored in records that contained the two channels of digital data representing n discrete channel
responses recorded for a particular profile. A DSP card is used to interface the DAT to a PC,
which enables data transfer for processing and statistical analysis. Although impulses were collected every 100 ms, Doppler spread and coherence time were not included in the processing.
This may be added in the future. However, we did include the ability to spatially average the
data, which would minimize large variations in delay spread from individual scatterers.
XThe first step in processing is to separate the n distinct channel responses recorded on a
particular profile. To do this an autocorrelation is performed on the data to accurately determine the impulse repetition rate. Once this is accomplished, the peak of the first response
(minus sufficient samples to account for the pulse rise time) is located to determine a beginning
point of the record. Then the repetition rate is used to identify each succeeding response. Let
XCOi()),(i=1,2..n) and Quadi()),(i=1,2..n) be the n distinct cophase and quadphase voltage time
Xsequences for each channel of a data profile. The powerdelay profile (PDP), Pi()), is the
X[Ԛchannel impulse function squared, #Xm4L x- -XX#Ii())2#Xq4L P: XP#. For the ith record in a data profile, this is
X" 1dddddddd(1)1dddddddd(1) I!x)ddddvNddx5P_i (tau) ~=~ CO_i (tau)^2 + Quad_i(tau)^2 Xq4L P: XPXq4L P: XPXq4L P: XPvP6ivCO6ivQuad6iv(jv)v(v)2v(Vv)2v)Vv)v)vIvI` "T$(1) "0*$$ԌXThe PDP can also be averaged spatially by computing a running average of n adjacent impulse
responses as the transmitter cart is moved, equation (2). The power delay is then normalized by
total received power to obtain the averaged probability density function of excess delay, equation (3) [4, 5].
3XtA x3
dddd3
kdd x>cIota_i^n (tau)^2~= ~1 over n~ Sum From { alpha= i} To { i +
n-1} Iota_ alpha(tau)^2 (i=1,2.....j) Xq4L P: XPXq4L P: XPXq4L P: XP)86{Y)benbi(\ninn6ii3
j(U)I2(j11() I2L(?1,29 .k . . .
.k
)36*0I>` "T$(2)
X
a xPdddd<dd, xSp_i^n(tau)^2~= ~{Iota_i^n(tau)^2} OVER {Int_0 ^ infinity
Iota_i^n(tau)^2 d(tau)} Xq4L P: XPXq4L P: XPXq4L P: XPpnvini nid()2H(H)260()2{(+)4)H\H)[$))8Y_tGL` "T$(3)
XNThe mean delay, T0in, is then calculated using the statistical definition of first moment,
equation (4).
X/ xdddddd x-T0_i^n~=~Int_0^Infinity tau p_i^n(tau) d(tau)Xq4L P: XPXq4L P: XPXq4L P: XPT0 nipc n<id8GL60(V)()g))@)/` "T$(4)
X[ xdddd
dd xWTd_i^n ~ = ~ LEFT [ INT _0^INFINITY(tau- T0_i^n)^2
p_i^n(tau) d(tau) RIGHT ] ^{1/2}Xq4L P: XPXq4L P: XPXq4L P: XPTd niT0 nip niEdA8Au |GL
60()=2R()(Y ) I1
I/6
I2)))` "T$(5)
XThe delay spread Tdin is defined as the square root of the variance, which is also the
square root of the second central moment, equation (5). This integral is performed as a summation with maximum integration time equal to the sum of the word lengths of the averaged
impulses or up to the time for any individual impulse when the signal decays to a predefined
threshold level.;"0*$$The signal level attenuation versus distance is tracked using peak power and energy per
impulse. These quantities are normalized to 0 dB with the transmitter and receiver separated
Xby 3 m.
Peak power is the maximum of the PDP for each impulse response. Energy is the integral of the PDP equation (6). This is also the average cw power over the system bandwidth.
Xt6 x
ddddddX x2P_{i _Total}^n ~ = ~ int_0^inf P_i^n (tau) d(tau) Xq4L P: XPXq4L P: XPXq4L P: XPP niddYTotalPC niydN8GL60(6)()) )6` "T$(6)
The correlation function [6] is computed using equation (7).
X$xdddd:jddx.R(Delta f) ~ = ~ F ~LDBRACK p_i^n(tau) RDBRACKXq4L P: XPXq4L P: XPXq4L P: XPvRmvfvF vpnCiv(v)v(v)v5v)@vvv ` "T$(7)
The correlation bandwidth is then defined as the bandwidth at which the correlation
drops by 50 percent.
XCalibration
System calibration and delay spread processing algorithms were checked using two
coaxial transmission lines to simulate a multipath signal in the laboratory. The calibration setup
is shown in Figure 3. The delta time for signal arrival was calculated using cable manufacturer
specifications. The attenuation of a 1.5 GHz cw signal was also measured. These results are
listed in Table 1. Figure 4 shows the measurement results for one nonaveraged impulse
response from a 10 s calibration test. Table 2 lists the measurement results verses the known
characteristics of the unequal delay lines. Delay measurements for a random impulse were
found to be accurate to about 5 percent and amplitude measurements agreed to about
7percent.
d!0*$$4Table 1. Delay Line Characteristics at 1.5 GHz
^
ddx
!4ddx ^
>
%Cable # Xb
D6Vp(m/s) IL(m) XbYTd(ns) <kA(dBm)
1(FWG)
D62.40E+8 EI15.5 Z64.6 l<l5.67 2(RG55)
D61.73E+8 wJ1.5 Z 8.67 l<l2.17
$Table 2. Calibration Results for Known Delay t and Peak Signal Attenuation A
;Measured at 100M B/s and 500 MB/sЃ
c
!4ddx
Addx c b
]%7Calculated UMeasured
XS100, 500 MB/sx IoPercent Error
Io100, 500 MB/sb
t(ns)] v%:55.9] 59, 58] 5.5, 3.7 x Ar U%: 0.45r 0.417, 0.483r 7.3, 7.3 ]
X,A minimum delay spread measurement was also made. To do this a delay spread profile
for a direct signal with no multipath was recorded for 10 s. The theoretical minimum delay
spread for the system under these conditions was calculated for a triangular pulse 127 bits high
with a base 2 bits wide. A threshold detection level of 20 dB was then used to perform the
integration required in equation (5). Results from this test are given in Table 3.
0*$$= Table 3. Measured Versus Calculated Minimum Delay Spreads for Direct Signal at 100 and
;500 MB/s Using a 20 dB ThresholdЃ
Y
Addx
addxF` ` Y b
] g K X>CalculatedĐC2
&A(ns) Xa#Xm4L x- -XX#Measured#Xq4L P: XP#ѐeVnc(ns)b
B F g
X0Td100 BQ4.2C2
d3.6eVB r Xr Td500
BQ0.8C2
\ d1.1eVr :
X
Measurements
'
XDelay spread measurements were made at three locations. Site #1 was a long hallway
with cinder block walls at the Department of Commerce Boulder Laboratories. The hallway
was flanked by offices on one side and laboratories on the other. Site #2 was the Commerce
auditorium, also located at the Boulder Laboratories. This room is wood, paneled with metal
seats, and a raised stage area. Site #3 was an open floor plan office with soft partitions at the
US West Research Facility in Boulder, CO. Floor plans, transmitter, receiver, and data profile
locations are shown in Figures 5, 6, and 7. In all cases the transmitter was moved on a cart
along the profiles with a fixed receiver location. The receiver was positioned at a height of 2.1
m (7 ft). The transmitter height was 1.6 m (5.3 ft). This gave a clear line of sight (LOS)
between the transmit and receive antennas in the hallway and the auditorium. In the soft partitioned office, all profiles except #1 and #2 were obstructed line of sight (OLOS). This obstruction was due to the soft partitions, which were 6 ft high and randomly spaced office furniture
(see Figure 6).
Delay spreads were calculated from the received PDP using no averaging and a 20 dB
threshold. Signal levels below threshold (20 dB below the peak signal) on a PDP are set to
zero. This must be done so delay spread is not a function of integration time. #0*$$ԌBecause significant delays were encountered at the US West office on OLOS profiles,
the 100 MB/s data were selected for reporting. This choice was made by monitoring the tail of
each PDP. The ratio of the average value of the tail of the PDP and the peak signal of each
PDP should be at least 20 dB. An example of this calculation is in Figure 8 and is labeled PDP
Amplitude Ratio. The delay profile and PDP's associated with the minimum and maximum
delay spreads for US West profile 4 are also shown in Figures 8. The correlation function for
the maximumminimum impulses from Figure 8 are shown in Figure 9. As can be seen, the
FFT's (correlation function) of PDP's with larger delays decay more rapidly to the 3 dB points
and have a smaller correlation bandwidth (BW). However, it can be seen by inspection of the
max and min delay spreads and the corresponding 3 dB points of the correlation function that
delay spread and correlation BW do not have a precisely inverse relationship in our calculations.
For comparison of these results with calibration data, the 3 dB points for direct signals
with no delay spread are given in Table 4.
[%Table 4. Measured System Bit Rate Versus 3 dB Point of Correlation FunctionЃ
Bfor a Direct SignalЃ
Y
addxF` `
ddxT Yr
XaBit Rate (MB/s) >4100 X-200 t500
XF3 dB Point(MHz) >4112 X-225 t450
When making estimates of the correlation BW, the 3 dB points from Table 4 can be
considered as upper limits for channel BW measurements. Correlation BW's approaching these
numbers are really underestimates for the channel. To correct this, the system transfer function
can be deconvolved in the processing. This would extend the usable correlation BW measurement ability of the system. This has not been done as yet but can be if the upper limits are
considered important.#0*$$ԌDelay spread, correlation BW, and attenuation versus position graphs have been compiled for all data profiles. These graphs display the calculated results for each PDP at 100 ms
intervals again using a 20 dB threshold and no spatial averaging. The time sequence was then
converted to distance by assuming a constant cart speed.
Figures 10, 11, and 12 are comparisons of delay spread profiles and correlation BW.
Figure 10 shows the two LOS profiles at the US WEST office. Figure 11 displays the OLOS
profiles from US WEST, and Figure 12 shows the LOS profiles measured at the Department of
Commerce (DOC) sites. The somewhat inverse relationship between delay spread and CBW
can be seen on all plots. In general, the delay spread magnitudes are dominated by the geometry of the measurement site. For LOS paths delay spread decreases slightly with increasing
distance in the narrow DOC hallway (Figure 13) and has smaller delay spreads than the LOS
measurements in the square plan auditorium (Figure 13). LOS paths in the US WEST office
experience larger delay spreads up to 70 ns caused by possible multipath through windows and
scattering off partitions.
OLOS paths at US WEST (Figure 11) exhibit the largest delays, which increase somewhat linearly with distance. Here we see maximum delay spreads around 200 ns on profiles 2,
3, 4, 5, 7, and 8.
Cumulative distributions and histograms of delay spread were compiled and are shown in
Figures 14, 15, and 16. Median values of delay spread are 8.5, 24, and 65 ns for the hallway
auditorium and soft partitioned office respectively.
Position data were used to convert attenuation versus profile position to attenuation versus log euclidean distance [7] from the transmitter. Scatter plots were made for each profile,
and the average slope was determined using a linear regression algorithm. Results are given in
Figures 16, 17, and 18, and the average power law coefficients are summarized in Table 5.
0*$$H)Table 5 Power Law Coefficients and Averages for LOS and OLOS Paths
Y
ddxT
}
ddx Y b
+ Location Type/# Power Law
Coefficientsb
+J DOC Hallway LOS 1.41 JJ DOC Auditorium LOS .898 JJ US WEST Office LOS/1 1.69 JJ
LOS/6
.929 JJ OLSO/2 2.58
JJ z OLSO/3z 3.51 JJ _ OLSO/4_ 4.74 z JJ D OLSO/5D 7.49 _ JJ ) OLOS/7) 3.37 D OLOS/8 7.75 ) JJ Average LOS 1.23 Jw OLSO 4.91 w
An inspection of these coefficients shows an average below free space attenuation (coefficient 2) for indoor LOS paths. However, the attenuation coefficients for OLOS paths vary
between 2.58 and 7.75. This would seem a large range, but is explained by the number of partitions blocking line of site on each path. It can be seen from Figure 6 that path 2 has the fewest
partitions separating the transmitter and receiver and the smallest attenuation coefficient for the# 0*$$OLOS paths. The coefficient then increases in a somewhat predictable pattern on paths 7, 3, 4,
5, and 8.
6FREFERENCES
[1]XE. J. Violette, R. H. Espeland, K. C. Allen, "A diagnostic probe to investigate propagation at millimeter wave lengths," NTIA Report 83128, August 1983, (NTIS Order
No.PB84-104223).T$
[2] XE. J. Violette, K. C. Allen, et al., "Millimeterwave urban and suburban propagation
measurements using narrow and wide bandwidth channel probes," NTIA Report 85184,
1985 (NTIS Order No. PB 86147741).T$
[3]XE. J. Violette, Felix Schwering, et al., "MillimeterWave Propagation at Street Level in an
Urban Environment," IEEE Trans. on Geoscience and Remote Sensing, Vol. 26, No.3,
May 1988.T$
[4]XD. C. Cox, "Time and frequency domain characterizations of multipath propagation at
910MHz in a suburban mobileradio environment," Radio Sci., Vol. 7, No. 12., pp.10691077, December 1972.T$
[5]XD. M. J. Devasirvatham, "Multipath time delay spread in the digital portable radio environment," IEEE Comm. Magazine, Vol. 25, No. 6, June 1987.T$
[6] XJ. G. Proakis, "Digital Communications," (McGrawHill 1983), pp. 458500.T$
[7]XS. E. Alexander, "Radio propagation within buildings at 900 MHz," Electron. Letters,
X1982, 18, pp. 913914.T$
M
0*$$
1Figure 1. Wideband delay spread system transmitter.
h3Figure 2. Wideband delay spread system receiver.#0*$$
k:Figure 3. Calibration test setup.
Figure 4. Measured calibration data. Single correlation at 100 and 500 MB/s, 60 ns delay.#0*$$3'3'Standard'3Standard - Dup Long'3Standard - Dup Longф
'3Standard - Dup Long'3Standard - Dup Long'3'3Standard- Dup Longф
'3'3Standard- Dup Long3'3'Standard- Dup Long(ф
5 ,Figure 5. Floor plan and data profile, Wing 4, Radio Building.
0*$$
ht
Figure 6. US WEST Room 3200 floor plan with delay spread profile locations and receiver location.
0*$$ht
3Figure 7. Floor plan, Radio Building auditorium.e0*$$
Figure 8.44US WEST Profile No. 4, PDP's of max, min delay spreads, delay spread profile,
44and PDP amplitude ratio.N0*$$
Figure 9.44US WEST Profile No. 4, correlation functions for min and max PDP delay
44spreads, correlation BW profile.N0*$$
Figure 10.44Delay spread and correlation BW profiles. US WEST Rm. 3200 (LOS): Profiles
44(a) 1 and (b) 6.70*$$
Figure 11.44Delay spread and correlation BW profiles. US WEST Rm. 3200 (OLOS):
44Profiles (a) 2, (b) 3, and (c) 4.70*$$
Figure 12.44Delay spread and correlation BW profiles. US WEST Rm. 3200 (OLOS):
44Profiles (a) 5, (b) 7, and (c) 8.70*$$
Figure 13.44Delay spread and correlation BW profiles. Commerce Radio Building (LOS):
44(a) hallway and (b) auditorium.70*$$
.Figure 14a. Histogram of delay spreads, Commerce hallway.0*$$
1Figure 14b. CDF of delay spreads, Commerce hallway.0*$$
-Figure 15a. Histogram of delay spreads, Commerce auditorium.0*$$
0Figure 15b. CDF of delay spreads, Commerce auditorium.0*$$
/Figure 16a. Histogram of delay spreads, US WEST office.0*$$
2Figure 16b. CFD of delay spreads, US WEST office.0*$$
Figure 17.44Attenuation scatter plots, US WEST. Profiles: (a) 1, (b) 2, (c) 3, (d) 4, (e) 5,
44(f) 6, (g) 7, and (h) 8.70*$$
3Figure 18. Attenuation scatter plot, auditorium.
V4Figure 19. Attenuation scatter plot, hallway.
0*$$