1290 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 58, NO.

4, APRIL 2010

Radio-Wave Propagation Into Large Building

Structures—Part 2: Characterization of Multipath
Kate A. Remley, Senior Member, IEEE, Galen Koepke, Member, IEEE, Christopher L. Holloway, Fellow, IEEE,
Chriss A. Grosvenor, Member, IEEE, Dennis Camell, Member, IEEE, John Ladbury, Member, IEEE,
Robert T. Johnk, Member, IEEE, and William F. Young, Member, IEEE

Abstract—We report on measurements that characterize multi- large buildings where difficult radio reception is often encoun-
path conditions that affect broadband wireless communications in tered because of signal attenuation and variability. Our measure-
building penetration scenarios. Measurements carried out in var- ment set-up is intended to simulate a response scenario, where
ious large structures quantify both radio-signal attenuation and
distortion (multipath) in the radio propagation channel. Our study
an incident command vehicle is located near a structure and a
includes measurements of the complex, wideband channel transfer mobile unit is deployed inside. In contrast to the many existing
function and bandpass measurements of a 20 MHz-wide, digitally studies on cell- or trunked-radio systems, in our case we focus
modulated signal. From these, we derive the more compact met- on ground-based point-to-point radio communications.
rics of time delay spread, total received power and error vector Our goal is to provide a large body of measurement data, ac-
magnitude that summarize channel characteristics with a single quired in key responder environments, to the open literature (see
number. We describe the experimental set-up and the measure-
ment results for data collected in representative structures. Finally, [7] for a complete list) for improved communication system de-
we discuss how the combination of propagation metrics may be velopment and design and to aid in technically sound standards
used to classify different propagation channel types appropriate development. A second goal is to share our methodology so that
for public-safety applications. responder organizations and others may carry out these charac-
terizations as desired. A third goal of this program is to provide
Index Terms—Attenuation, broadband radio communications,
building penetration, digital modulation, emergency respon- measurement data that will be useful for verification of network
ders, error vector magnitude, excess path loss, received power, simulations of emergency responder radio links. Such simula-
time-delay spread, vector network analyzer, vector signal ana- tions are being developed by NIST, among others [8].
lyzer, wireless signals, wireless system measurements, wireless Much work has been published describing measurement char-
telecommunications. acterization of multipath in the radio-propagation environment.
Various figures of merit are often used to describe multipath
effects, including excess path loss, frequency selectivity, time
I. INTRODUCTION delay spread, bit error rate (BER) and its variants, and/or error
vector magnitude (EVM). Most of these publications (for ex-
ample, [9]–[13] and references cited therein) describe measure-

T O aid in the development of standards that support reli-

able wireless communications for emergency responders
such as those discussed in [1], the National Institute of Stan-
ments intended to simulate communications via cellular tele-
phone or other wireless systems that rely on a fixed base sta-
tion whose antenna is positioned high above the ground and a
dards and Technology (NIST) has embarked on a project to mobile user located at ground level. Only a few publications de-
acquire data on radio-wave propagation in key emergency-re- scribe measurements that simulate point-to-point radio-commu-
sponder and public-safety environments. In past work [2]–[4], nication scenarios, such as those utilized in many emergency-re-
measurements were made in buildings scheduled for implosion sponder scenarios. Examples include sections of [10], [14] and
to simulate collapsed-building environments. The focus of cur- references cited therein, and [15].
rent work [5], [6] is to study the penetration of radio waves into Part 1 of this work [16] summarizes the statistics of re-
ceived-signal-strength measurements under single-frequency
excitation in twelve large public building structures, based
Manuscript received December 18, 2008; revised August 10, 2009; accepted
September 25, 2009. Date of publication January 22, 2010; date of current ver-
on NIST Technical Note 1545 [5]. In that work, a spectrum
sion April 07, 2010. This work was supported in part by the U.S. Department of analyzer was used to measure the relative received signal
Justice, Community-Oriented Police Services through the NIST Public-Safety strength at a fixed receiver location as a portable transmitter
Communications Research Laboratory. was carried throughout the buildings. The mean and standard
K. A. Remley, G. Koepke, C. L. Holloway, C. Grosvenor, D. Camell,
J. Ladbury, and W. F. Young are with the National Institute of Standards deviation of the measurements in the structures were calculated.
and Technology, Boulder, CO 80305 USA (e-mail: remley@boulder.nist.gov; Frequencies included bands near licensed public-safety bands
kate.remley@nist.gov; koepke@boulder.nist.gov; holloway@boulder.nist.gov; and cell phone bands including 49 MHz, 160 MHz, 225 MHz,
chriss@boulder.nist.gov; camell@boulder.nist.gov; ladbury@boulder.nist.gov;
wfyoung@sandia.gov). 450 MHz, 900 MHz, 1.8 GHz, and a limited set of data at
R. T. Johnk was with the National Institute of Standards and Technology, 2.4 GHz and 4.95 GHz.
Boulder, CO 80305 USA. He is now with the Institute for Telecommunication In this paper, we focus on additional parameters relevant to
Sciences, Boulder, CO 80305 USA (e-mail: bjohnk@its.bldrdoc.gov).
Color versions of one or more of the figures in this paper are available online
successful transmission of modulated signals including wide-
at http://ieeexplore.ieee.org. band channel frequency response, excess path loss, time delay
Digital Object Identifier 10.1109/TAP.2010.2041143 spread, channel power and error vector magnitude. We focus on
0018-926X/$26.00 © 2010 IEEE

Authorized licensed use limited to: NIST Virtual Library (NVL). Downloaded on July 15,2010 at 17:30:04 UTC from IEEE Xplore. Restrictions apply.
REMLEY et al.: RADIO-WAVE PROPAGATION INTO LARGE BUILDING STRUCTURES—PART 2: CHARACTERIZATION 1291

carrier frequencies over 1 GHz, focusing on frequency bands

around 2.4 GHz and 4.95 GHz to investigate the differences
in transmission between existing wireless systems in the unli-
censed 2.4 GHz industrial/scientific/medical (ISM) frequency
band (which is sometimes used by public-safety organizations)
and systems proposed for use in the licensed public-safety fre-
quency band covering 4.94 GHz to 4.99 GHz [17], [18].
In NIST Technical Note 1546 [6], we conducted tests in four
representative environments that are notoriously difficult in
terms of radio reception for emergency responders. These are a
multi-story apartment building, an oil refinery, a long corridor
in an office building typical of many commercial facilities, and
a subterranean tunnel. Here we provide representative results
of the characterization of multipath in the propagation channel Fig. 1. Wideband measurement system based on a vector network analyzer.
Frequency-domain measurements, synchronized by the optical fiber link, are
and discuss key findings from [6]. transformed to the time domain in post-processing. This enables determination
We extend the work of [6] to demonstrate how compact “sum- of excess path loss, time-delay spread, and other figures of merit important in
mary” metrics that quantify the channel characteristics with a characterizing broadband modulated-signal transmissions.
single number, may be used to classify propagation channels for
responder applications. Summary metrics are often easier and
more inexpensive to acquire than are extensive frequency-do- with broadband digitally modulated signals at 2.4 GHz and
main measurements. By classifying propagation channel types, 4.95 GHz. From these, we found the error vector magnitude
responders may better know how to deploy wireless systems, (EVM) in each environment. EVM is a figure of merit that de-
and standards development organizations may develop better scribes the level of distortion in received, demodulated symbols
performance metrics and verification tests for wireless tech- of a digitally modulated signal. In the following sections, we
nology. These data may also be used to support general building describe these measurements.
or scenario classifications such as those presented in [9], [10].
B. Wideband Vector Network Analyzer Measurements
In Section II, we discuss the instrumentation, calibrations,
and post-processing methods we used to collect the various data. 1) Measurement Set-Up: We measured the wideband fre-
In Section III, we show representative measured data collected quency response and time-delay characteristics of the propaga-
at two of the four large structures reported in [5], [6]: an 11-story tion channel using a measurement system based on a vector net-
apartment building and an oil refinery. We summarize the prop- work analyzer, shown in Fig. 1. This instrument collects data
agation effects that we observed and draw some conclusions on over a very wide frequency range, from 25 MHz to 18 GHz
the characteristics of the propagation channels seen in the struc- for the system we used. This system, also described in [19],
tures. lets us measure the complex transfer function of the channel,
Occasionally product names are specified solely for com- including frequency-selective characteristics. By taking the in-
pleteness of description, but such identification constitutes no verse Fourier transform of the measured transfer function, the
endorsement by the National Institute of Standards and Tech- power delay profile and RMS delay spread of the channel are
nology. Other products may work as well or better. found in post processing.
The VNA acts as both transmitter and receiver in this system.
II. MEASUREMENTS FOR MODULATED-SIGNAL The transmitting section of the VNA steps through the frequen-
CHANNEL CHARACTERIZATION cies a single frequency at a time. The signal is amplified and
fed to a transmitting antenna, as shown in Fig. 1. The received
A. Overview signal is returned to the VNA via a fiber-optic cable. Transmit-
In the study of [6], we conducted three types of measure- ting the received signal along the fiber optic cable back to the
ments: single-frequency received-power using a communi- VNA eliminates the loss and phase changes that would be asso-
cations receiver; frequency response data over a very broad ciated with RF coaxial cables between the receive antenna and
frequency band at fixed points in the propagation environment the transmit antenna, allowing characterization of the complex
using a vector network analyzer (VNA); and modulated-signal radio channel. One advantage of this system is that it provides
measurements using a vector signal analyzer (VSA). The a high dynamic range relative to true time-domain-based mea-
single-frequency data are similar to those reported in [5] and surement instruments.
are not discussed here. From the VNA data, we determined In Fig. 1, the system is configured for a line-of-sight reference
the wideband frequency response of the propagation channel measurement. In practice, the transmitting and receiving an-
and, from this, the excess path loss (the loss that exceeds tennas may be separated by significant distances, although they
that measured in a free-space environment [19], [20]). After must remain tethered together by the fiber-optic link. While di-
transforming these data to the time domain, we calculated the rectional horn antennas are shown in Fig. 1, omnidirectional an-
root-mean-square (RMS) time-delay spread. The RMS delay tennas were also used in our measurements. Omnidirectional an-
spread is a figure of merit that quantifies the time it takes for tennas are most often used in public-safety applications. How-
reflections in a received signal to die out. Using the VSA, ever, high-frequency measurements often benefit from the use
we also collected modulated-signal measurements associated of directional antennas to maximize gain in a specific direction.

Authorized licensed use limited to: NIST Virtual Library (NVL). Downloaded on July 15,2010 at 17:30:04 UTC from IEEE Xplore. Restrictions apply.
1292 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 58, NO. 4, APRIL 2010

Emergency response agencies may use directional antennas in

these situations as well.
We made measurements in two frequency bands: a low
band that ranged from 100 MHz to 1.2 GHz, and a high band,
that ranged from 1 GHz to 18 GHz. We used omnidirectional
transmit and receive antennas for the lower frequency band
and, for the measurements reported on here, a dual-ridge-guide
(DRG) directional transmit antenna and omnidirectional receive
antenna for the higher frequency band (1 GHz to 18 GHz). The
beamwidth of the omnidirectional antennas is approximately
40 to 50 in the vertical direction. The beamwidth of the DRG
varies with frequency, from almost at the low end at
about 1 GHz to at 18 GHz, necessitating reorientation of
the antenna to track the receiver’s position in certain structures
such as the apartment building.
It is well known that in propagation environments where the
transmitting antenna is located in the midst of highly reflecting Fig. 2. Reference measurement at three meters for a dual-ridge guide horn an-
objects, the use of a directional horn antenna can reduce the tenna, transformed to the time domain. The waveform shows the antenna re-
apparent multipath by eliminating reflections from behind. Our sponse, the ground-bounce response and the spurious environmental effects.
studies in an automobile manufacturing plant [21] demonstrated
this. For the measurements presented here, strong reflectors (the
building structures) were located in front of the transmitting an- antenna and the reference location by dividing by
tenna with very few scatterers behind. Thus, we anticipate the .
use of directional antennas introduced only small deviations to To find the excess path loss, we additionally reduce the total
the measured multipath effects. path loss by the expected free-space path loss over the overall
To make a measurement, the vector network analyzer is separation distance between the transmit and receive antennas.
first calibrated by use of standard techniques where known To do this, we divide the measurement of by .
impedance standards are measured. The calibration enables us Equivalently, we can multiply by .
to correct for the response of the fiber-optic system, amplifiers, The distance may be measured or estimated from maps, de-
and any other passive elements and electronics used in the pending on the environment. As stated earlier, this provides the
measurement. We also high-pass filter our measurements in loss in excess of that which would be measured at the same dis-
post processing to suppress the large, low-frequency oscillation tance in free space. We note that communication engineers typ-
that occurs in the optical fiber link. ically think of excess path loss as a single frequency or narrow-
For the measurements reported here, the VNA-based mea- band measurement. However, the VNA measurements provide
surement system was set up with the following parameters: the a much richer data set because they include both magnitude and
initial output power was set between 15 dBm to 13 dBm. phase information over a broad frequency range.
The gain of the amplifier and the optical link and the system The reference measurement is made at a specified distance
losses resulted in a received power level no more than 0 dBm. and may be acquired either during field tests or from a labora-
An intermediate-frequency (IF) averaging bandwidth of around tory measurement. In the field, the measurement includes envi-
1 kHz was used to average the received signal. We typically ronmental effects, and we use time-domain gating to minimize
recorded 6401 points per frequency band and chose the number reflections on the free-space reference. If we are not able to gate
of bands recorded in each measurement to avoid aliasing of the out the reflections satisfactorily, the reference measurement is
signal. We report here on our high-band measurements ranging made in a laboratory facility such as an anechoic chamber or an
from 1 GHz to 18 GHz, which were taken by measuring 48003 open-area test site. In this case, we use the same antennas and
points for a total of three bands. Low-frequency measurements measurement system as were used in the field. For the measure-
from 100 MHz to 1.2 GHz are reported in [6]. The dwell time ments shown below, we used a two-meter reference made in the
was approximately 25 per point and the frequency spacing field. We chose this distance to balance the need to be in the an-
was approximately 379 kHz. tenna far field of our lowest frequency of interest (1 GHz for the
2) Wideband Frequency Response and Path Loss: Our wide- results reported here), while keeping the reference measurement
band measurements provide a channel transfer function , as free from environmental reflections as possible.
where typically is derived from the measured transmis- As an example, Fig. 2 shows the time-domain response for
sion parameter . To find the frequency-dependent path a reference measurement using a pair of dual-ridged-guide an-
loss between the transmit and receive antennas, we first com- tennas separated by 3 m. In Fig. 2, the reference measurement,
pute , where is a free-space reference transformed to the time domain, is shown with all environmental
made at a known distance from the transmit antenna. The effects. The reference measurement is gated (windowed) from
use of a ratio to find the path loss enables us to calibrate 20 ns to 32 ns to isolate the antenna response, which was de-
out the antenna response of the system. We correct the mea- termined previously in a separate measurement. The frequency-
surements for the free-space path loss between the transmit domain responses for the reference measurement would show

Authorized licensed use limited to: NIST Virtual Library (NVL). Downloaded on July 15,2010 at 17:30:04 UTC from IEEE Xplore. Restrictions apply.
REMLEY et al.: RADIO-WAVE PROPAGATION INTO LARGE BUILDING STRUCTURES—PART 2: CHARACTERIZATION 1293

C. Modulated-Signal Measurements

Along with the single-frequency measurements described in

[5], [6] and the wideband VNA measurements described above,
a third set of tests involved measuring the waveforms of digitally
modulated signals. From these, we plotted the spectra of the
signals, and calculated the received power and the error vector
magnitude associated with a given propagation channel. EVM
gives an indication of the distortion introduced into a digitally
modulated signal as it passes through a propagation channel.
Mathematically, this can be expressed as

(2)

Fig. 3. Power-delay profile for a building propagation measurement. Important

parameters for a measured signal are the peak level, the maximum dynamic where is the normalized symbol in a stream of
range, the mean delay, and the RMS delay spread.
measured symbols, is the ideal normalized constella-
tion point for the symbol, and is the number of unique
symbols in the constellation. The fractional form of EVM given
a noisier trace when environmental effects are included, com- in (2) is often represented as a percentage. The algorithm used
pared with a smoother trace for isolated antennas. The gated to find EVM was built into the receiver.
response is what we would see if the antenna were measured in The modulated signal used as excitation was based on or-
a free-space environment, free from environmental reflections. thogonal frequency-division multiplexing (OFDM) 802.11a/g,
In the following, we present graphs of the inverse excess path as specified by the IEEE 802.11a-1999 standard [25]–[27].
loss in decibels. Graphs of path loss would have positive ordi- OFDM is used in wireless local-area networks (WLANs) and
nates and increase with distance. Thus, we refer to the curves in the public-safety band at 4.95 GHz. In the latter, OFDM
in our graphs as “penetration.” This terminology is commonly signals may be transmitted in a 10 MHz wide channel using the
used in studies of electromagnetic shielding. 802.11j standard, instead of the 20 MHz wide channel utilized
3) RMS Delay Spread: The time-domain representation of in 802.11a. Our demodulator was able to measure only the
802.11a standard. As a consequence, this is what we report on
the signal was calculated from the excess path loss data in post
in the following sections.
processing. From this we found the RMS delay spread. RMS
The OFDM multiplexing scheme was developed to provide
delay spread is calculated as the square root of the second cen- immunity to interference. Data are encoded onto 52 narrow-
tral moment of the power-delay profile of a measured signal band, frequency-multiplexed subcarriers. For strong received
[22]–[24]. Fig. 3 shows the power-delay profile for a representa- signals, data are transmitted up to a maximum of 54 megabits
tive building propagation measurement. The peak level usually per second (Mbps). As the received signal strength decreases
occurs when the signal first arrives at the receiving antenna, al- or the level of multipath increases, the data rate decreases to
though in high multipath environments we sometimes see the compensate for the decrease in signal-to-noise ratio. In the tests
signal build up over time to a peak value and then fall off. reported here, we force the signal generator to transmit either a
A common rule of thumb is to calculate the RMS delay spread slow-data-rate quadrature-phase-shift-keyed (QPSK) or a high-
from signals at least 10 dB above the noise floor of the mea- data-rate 64-quadrature-amplitude-modulated (64QAM) signal.
surement [24]. For the measurements described in the following This lets us assess the impact of multipath for a given channel
sections, we defined the minimum dynamic range to be approx- when different modulation schemes are used.
imately 40 dB below the peak value, although this value was re- For the measurements described in the following sections, a
duced for lower signal levels. For the illustrative measurement vector signal generator was used to create the digitally modu-
shown in Fig. 3, we extended the window down to 70 dB below lated signals. The signal generator was mounted on a rolling cart
the peak value. Whether we use a 40 dB or a 70 dB threshold, and moved through the various propagation environments. We
the RMS delay spread does not change appreciably due to the used omnidirectional antennas where possible to mimic those
almost constant slope of the power decay curve. used by emergency responders. In some environments reported
here, we were limited by the dynamic range of the receiver and
The RMS delay spread can be defined as
so we used directional, dual-ridge-guide horn antennas.
We used a vector signal analyzer to acquire the signals. The
VSA maintains the phase of the measured frequency compo-
(1) nents relative to one another and enables measurements of com-
plex distortion in digitally modulated signals, including EVM.
In (1), is defined as the average value of the power-delay pro- The VSA used had a 36 MHz measurement bandwidth. No cor-
file in the defined dynamic range window, and is the variance rection for system effects such as antenna gain, system elec-
of the power-delay profile within this window. tronics, or system impedance mismatch was conducted. Thus,

Authorized licensed use limited to: NIST Virtual Library (NVL). Downloaded on July 15,2010 at 17:30:04 UTC from IEEE Xplore. Restrictions apply.
1294 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 58, NO. 4, APRIL 2010

these measurements are not absolute, but are relative to the mea-
surement configuration we used. The VSA was adjusted for
minimum received-signal distortion before each set of measure-
ments. This entailed performing an internal calibration followed
by a range adjustment under line-of-sight conditions.

III. MEASUREMENT RESULTS AND IDENTIFICATION OF

CHANNEL CLASSIFICATIONS

A. Overview
As stated in the introduction, in [6] we collected measure-
ments in four large public structures, two of which are reported
on here. By observing key features from graphs of the measure-
ments, we are able to classify distinct propagation effects de-
pending on the distance and type of structural obstruction be-
tween the transmitter and receiver. The propagation effects il-
lustrated here are specifically relevant to point-to-point commu-
nications used by most emergency response organizations.
The primary features we observed to make our classifications
include (i) the excess path loss, (ii) the amount and structure of Fig. 4. The 11-story, concrete, steel, and brick apartment building where the
NIST measurements were made.
frequency-selective distortion over the modulation bandwidth
(in this case, the spectrum covering 2.4 GHz and 4.95 GHz
10 MHz), (iii) the RMS delay spread, (iv) the received channel
power, and (v) the error vector magnitude. We illustrate how the
summary metrics (iii)–(v) can be used to gain a similar insight
into the channel as the more complete frequency-domain mea-
surements (i) and (ii). We illustrate these concepts using mea-
sured results from an apartment building and an oil refinery.
The latter was discussed in detail in [5] with respect to the use
of single-frequency statistics to classify channels. The oil re-
finery was also discussed in [28] and [29] as part of overviews on
the difficulties faced by public-safety practitioners and wireless
sensor networks, respectively, in high-multipath environments.
Here, we discuss how a combination of the metrics above can
help to identify classes of propagation environments that may
be important to the emergency response community. Again, the
reader is referred to [6] for a more complete discussion.
Fig. 5. Layout of a typical floor of the 11-story apartment building. Circles
show the test positions where measurements were made. The receiver site shown
B. Apartment Building is approximately 60 m east of the building.
We carried out propagation measurements at an 11-story
apartment building located in Boulder, Colorado in October
2007. The building, shown in Fig. 4, is constructed of rein- ments [1], including concrete construction, stairwells at the ends
forced concrete, steel, and brick. It contains standard interior of the hallways, apartments off a main corridor with outside-
finish materials. The building was fully furnished and occupied facing windows, and the need for single- or two-wall radio-wave
during the experiments. Measurements were performed during penetration. The Apartment Fire Scenario of [1] deals with a fire
daytime hours, so people were moving throughout the building response on the second floor of such an apartment building.
during the experiments. The apartment building propagation environment consisted
The layout of each floor of the apartment building was entirely of non-line-of-sight (NLOS) propagation paths because
T-shaped, with two elevators near the junction of the T. This is all of the received signals penetrated through the outside walls
illustrated in Fig. 5. The hallway along top of the T was approx- of the structure and at least one interior wall. The penetration
imately 20 m long and the body of the T was approximately measurement examples in Fig. 6 show a predominantly mono-
50 meters long. Our receiver site was located approximately tonic roll-off with frequency. These graphs show excess path
60 m from the building in a parking lot, also shown in Fig. 5. loss from 1 GHz to 18 GHz made at test position 2 (Fig. 6(a)),
Both VNA and VSA measurements were made every 5 meters, where the only obstructions between transmitter and receiver are
on Floors 2 and 7, as illustrated in Fig. 5. the building walls and windows, and test position 5 (Fig. 6(b)),
This apartment building was chosen because it has several where a metallic elevator obstructs the signal path.
features in common with the building described in the Apart- The building penetration decreases with frequency indicating
ment Fire Scenario of the SAFECOM Statement of Require- a channel that includes attenuation due to signal penetration

Authorized licensed use limited to: NIST Virtual Library (NVL). Downloaded on July 15,2010 at 17:30:04 UTC from IEEE Xplore. Restrictions apply.
REMLEY et al.: RADIO-WAVE PROPAGATION INTO LARGE BUILDING STRUCTURES—PART 2: CHARACTERIZATION 1295

Fig. 7. Measured spectra across a 30 MHz bandwidth in the apartment building.

(a) Test point 2, Floor 2; (b) test point 5, Floor 2. Top graphs for each location
represent measurements for a carrier frequency of 2.4 GHz and bottom graphs
for a carrier frequency of 4.95 GHz. Note that transmissions on other channels
within the 2.4 GHz band are visible on the graphs.

Fig. 6. Measured building penetration 1/(excess path loss), in dB, on Floor 2

(top, dark curves) and Floor 7 (middle, lighter curves) of the apartment building that have structure (at higher frequencies) occur when the signal
for frequencies from 1 GHz to 18 GHz at (a) position 2 and (b) position 5. The level approaches noise floor of the measurement system and do
lower curves represent the noise floor of the VNA-based measurement system. not give us information on the propagation channel. We neglect
Channel parameters such as loss and delay spread are calculated only where the
signal is significantly in excess of the noise floor. A 200 MHz moving average data whose level is not significantly in excess of the noise floor
filter has been applied to the data. in our calculations of excess path loss and RMS delay spread.
Given that this is a multipath channel, it is useful to deter-
mine whether the multipath distortion is narrowband—where
through some type of building material. Building materials all frequency components in the modulation bandwidth are
typically exhibit a monotonic increase in attenuation with fre- similarly affected by reflections, resulting in “flat” fading—or
quency, changing on the order of decibels over several decades wideband—where peaks and nulls occur within the modulation
of frequency. The fact that the penetration curves (where the bandwidth of the excitation, resulting in “frequency-selective”
free-space path loss has been removed) in Fig. 6(a) measured fading. Typically, wideband distortion is harder to overcome
on Floors 2 and 7 practically overlay indicates that the atten- for radio receivers because different signal components are
uation due to building penetration was similar on both floors. affected by the channel differently. Note that the definition
The difference in penetration between Floors 2 and 7 when of wideband distortion for a particular environment will be
the receiver is located behind the elevator indicates that more dependent on the excitation signal and the receiver used.
complicated, angle-dependent multipath scattering is involved The VSA measurements exemplified in Figs. 7(a) and (b) in-
in this propagation path. dicate whether wideband distortion is present in the frequency
Fig. 6 shows peaks and nulls that vary rapidly with frequency. band of interest. Here, we see the spectrum of the signal plotted
These indicate multipath in the environment in addition to the versus frequency for the OFDM digitally modulated signal (dark
building penetration effects. Because all of the received signal curve) and for a multisine signal designed to simulate the RF sta-
components arrive on non-line-of-sight paths, the phase rela- tistics of the digitally modulated signal (lighter, dashed curve).
tionship between them is random, indicating a Rayleigh dis- The multisine is easier to generate and characterize using stan-
tribution for the received signal. Note that the peaks and nulls dard RF measurement instrumentation [30].

Authorized licensed use limited to: NIST Virtual Library (NVL). Downloaded on July 15,2010 at 17:30:04 UTC from IEEE Xplore. Restrictions apply.
1296 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 58, NO. 4, APRIL 2010

Fig. 8. Received power averaged across a 30 MHz bandwidth at 13 positions Fig. 9. RMS delay spread (ns) at 13 positions on two different floors of the
on two different floors of the 11-story apartment building for frequencies from 11-story apartment building made using a directional transmit antenna for fre-
1 GHz to 18 GHz. Measurements were not recorded at all locations, as shown quencies between 1 GHz to 18 GHz.
by the missing data points.

The graphs in Fig. 7 show the received spectrum at test po-

sitions 2 and 5 on Floor 2. In each figure, the measurements in
the top graph had a center frequency of 2.4 GHz and the bottom
graph had a center frequency of 4.95 GHz. Note the additional
signals received in the 2.4 GHz frequency band, both above and
below the OFDM signal. These are from other 2.4 GHz wire-
less devices operating nearby. Emergency response organiza-
tions using unlicensed frequency bands may be susceptible to
interference from other sources [17].
The VSA measurements show that the channel is generally
frequency flat when simple building penetration through win-
dows and concrete walls occurs; for example, at test position 2.
However, frequency-selective distortion can be seen when the Fig. 10. EVM on Floor 2 of the apartment building for a QPSK-modulated
transmitter is behind the elevators or stairwell doors; for ex- OFDM signal for carrier frequencies of 2.4 GHz (circles) and 4.95 GHz (trian-
gles). The thick black line represents an EVM of 15%. Measurements on Floor
ample, at test position 5. 7 were weak enough that it was difficult to determine the EVM.
A simple classification scheme for the propagation channels
encountered in this type of building structure may be derived
from the two frequency-domain measurements illustrated in the entire high-frequency band, at all test positions on Floors 2
Figs. 6 and 7. For the majority of the locations we tested, and 7. We see an increase in RMS delay spread when the re-
the channel could be classified as “simple building penetra- ceiver is blocked by the metallic elevator shaft, as expected. We
tion,” consisting of flat fading combined with decade-scale, see only a moderate increase in the RMS delay spread between
frequency- and elevation-dependent path loss. The frequency the different floors, which is consistent with our classification
and elevation dependence are indicated by the reduced overall of the majority of test positions as simple building penetration
signal level for the 4.95 GHz measurements, and for higher (attenuation as opposed to multipath).
floors, respectively. In the location behind the elevator, we The error vector magnitude plots shown in Fig. 10 illustrate
may classify the channel as “building penetration combined a significantly lower EVM for the 2.4 GHz signal than for the
with frequency-selective fading.” The latter, illustrated by the 4.95 GHz signal. We see an increase in EVM at 2.4 GHz when
spectral variations shown in Fig. 7(b), is not the dominant the transmitter is shadowed by the elevator. Received signals
channel classification in this type of building. whose EVM is high will have difficulty maintaining a link. An
We next consider three summary metrics that can be used to EVM of 15% is marked by the thick black line on the graph
derive the same classifications for the propagation channels as to denote channels that may be unusable. Because the dynamic
were derived from the frequency-domain representations above. range of the VSA is not as high that of the VNA-based mea-
Fig. 8 shows the received power of the OFDM signals aver- surement system, we were unable to obtain meaningful EVM
aged across a 30 MHz bandwidth. These graphs illustrate the measurements on Floor 7. However, we expect that the channel
frequency- and elevation-dependent path loss. When compared characteristics on Floor 7 are similar to those on Floor 2, based
to the 2.4 GHz signal on Floor 2, the received power for the on the RMS delay spread measurements shown in Fig. 9.
4.95 GHz carrier frequency was on the order of 15 dB to 20 dB The received power measurements again show a frequency
lower, and the received power for the 2.4 GHz carrier on Floor and elevation dependence. RMS delay spread and EVM indi-
7 was around 10 dB lower. cate that the multipath environment introduces both narrow-
To classify the fading characteristics, we consider the RMS band- and wideband-distortion. Whether we use the frequency
delay spread. Fig. 9 shows the RMS delay spread, computed for measurements directly or a combination of summary metrics,

Authorized licensed use limited to: NIST Virtual Library (NVL). Downloaded on July 15,2010 at 17:30:04 UTC from IEEE Xplore. Restrictions apply.
REMLEY et al.: RADIO-WAVE PROPAGATION INTO LARGE BUILDING STRUCTURES—PART 2: CHARACTERIZATION 1297

Fig. 12. Layout of the test positions for the excess path loss and modulated-
signal measurements in the oil refinery complex. The test positions were located
under dense overhead piping and metallic structures, in most cases several sto-
ries high.

Fig. 11. Modulated-signal measurements made at an oil refinery: (a) transmit-

ting unit consisting of a vector signal generator and omnidirectional antenna on
a mobile cart inside dense piping in the facility. Photograph (b) shows the om-
nidirectional receiving antenna, indicated by an arrow, located on top of the ITS
van.

analysis tells us we are dealing with a combination of simple

building penetration in a multipath environment. This combina-
tion is representative of many similar structures where signals
do not penetrate too deeply within a building.

C. Oil Refinery
A second example of the use of multiple metrics for prop-
agation channel classification is illustrated by measurements
conducted at a large oil refinery in Commerce City, Colorado
in March 2007. As with the measurements at the apartment
building, we chose this facility to simulate a response scenario
(the Chemical Plant Explosion) in the SAFECOM Statement of
Requirements [1].
The refinery is an outdoor facility covering many hectares
in area, with intricate piping systems. We carried out tests
primarily in an area of dense piping that forms a tunnel-like
structure, shown in Fig. 11(a). We studied the propagation
Fig. 13. Penetration 1/(excess path loss), in dB, measured at two locations in
from a location outside the piping tunnel, shown in Fig. 11(b), the oil refinery covering frequencies from 1 GHz to 18 GHz at (a) test position
to within the tunnel. Even though the site was outdoors, the 6 (LOS) and (b) test position 8 (NLOS).
dense piping was a significant barrier to radio communications
and the propagation channel may be thought of as involving
structure penetration. antenna for the VSA measurements) was located outside the
Measurements were made at locations in the oil refinery in- piping complex on top of a mobile test van owned by the In-
dicated in Fig. 12 for both VNA and VSA measurements. One stitute for Telecommunication Sciences (ITS), a sister Depart-
antenna (transmit antenna for the VNA measurements, receive ment of Commerce organization at the Boulder Labs Site. This

Authorized licensed use limited to: NIST Virtual Library (NVL). Downloaded on July 15,2010 at 17:30:04 UTC from IEEE Xplore. Restrictions apply.
1298 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 58, NO. 4, APRIL 2010

Fig. 15. Total received power averaged across a 30 MHz bandwidth for OFDM
signals measured in the oil refinery for test positions 1 to 12.

Fig. 14. Examples of the signal spectra derived from the VSA measurements Fig. 16. RMS delay spread in the oil refinery at test positions 1 to 12. The curve
of the modulated signal at (a) test position 6 and (b) test position 8 in the oil with circles represents frequencies from 25 MHz to 1.2 GHz and the curve with
refinery. The top graphs represent measurements made at a carrier frequency of inverted triangles represents frequencies from 1 GHz to 18 GHz.
2.4 GHz and the bottom graphs are for a carrier frequency of 4.95 GHz. These
examples show 64QAM modulated signals.

multipath that does introduce frequency-selective distortion, as

antenna was vertically polarized. The photo in Fig. 11(b) shows shown in Fig. 14(b).
the omnidirectional antenna mounted on the mast. Based on the frequency-domain measurements above, the
As may be expected, the many metallic surfaces in this en- propagation channel in the oil refinery can be classified as “low
vironment introduced significant multipath into the signals we attenuation, frequency flat” for the LOS conditions and “high
measured. Fig. 13 shows representative measurements of the attenuation, frequency-selective” for the NLOS conditions. We
penetration for frequencies from 1 GHz to 18 GHz. next study the use of the summary metrics total received power,
Between test positions 2 and 6, the overhead piping created RMS delay spread, and EVM to assess the same channel.
a tunnel-like environment where the multipath has significant Fig. 15 shows the received power averaged across the modu-
structure. We see distinct frequency resonances at these posi- lation bandwidth of the OFDM signals. The signals with a car-
tions, even though the receiving antenna had a line-of-sight con- rier frequency of 4.95 GHz were approximately 5 dB to 10 dB
dition with the transmitting antenna. Once the receiving antenna lower than those with a carrier frequency of 2.4 GHz. When the
turned the corner, for test positions 7 and higher, no line-of-sight receiver turned the corner into a NLOS condition, the received
condition existed. The received signal took on a rapidly varying, power levels dropped by around 15 dB for the 2.4 GHz signal
Rayleigh, appearance. and around 20 dB for the 4.95 GHz signal. Results were similar
We again consider the spectrum of the OFDM signal to for both QPSK and 64QAM modulation formats, because the
assess the frequency-selective distortion in this environment. output power was the same for both transmissions.
Representative VSA measurements are shown in Fig. 14. Our A graph of the RMS delay spread in Fig. 16 shows short-dura-
data showed only moderate frequency-selective distortion at tion values in the line of sight condition, and longer values after
test positions 2 to 6 (test position 6 is shown in Fig. 14(a)). the receiver has turned the corner. The increase in delay spread
This indicates that the extremely deep, frequency-resonant is more significant for the low frequency band (100 MHz to
nulls in the penetration measurements of Fig. 13 are spaced 1.4 GHz) than for the high frequency band (1 GHz to 18 GHz).
widely enough that the modulated signal experiences multipath The EVM graphs in Fig. 17 show that once the transmitter
distortion that is flat across the bandwidth. Once the transmitter turns the corner, the EVM increases significantly. Before the
turns the corner, however, the signal experiences additional turn, the EVM remains below 10% for both modulation formats.

Authorized licensed use limited to: NIST Virtual Library (NVL). Downloaded on July 15,2010 at 17:30:04 UTC from IEEE Xplore. Restrictions apply.
REMLEY et al.: RADIO-WAVE PROPAGATION INTO LARGE BUILDING STRUCTURES—PART 2: CHARACTERIZATION 1299

shown here, response organizations can expect to encounter

simple building penetration with narrowband and, occasionally,
wideband fading. These effects are expected to be common
to environments that superficially appear quite different from
each other. For certain propagation mechanisms such as simple
building penetration, single-frequency signal-strength or
channel power measurements may be sufficient to characterize
the environment. In cases where multipath conditions exist,
measurements that cover multiple frequencies and metrics such
as RMS delay spread or EVM can offer additional insight.
It is clear that one critical factor for successful radio commu-
nications is knowledge of both the basic environmental charac-
Fig. 17. EVM at the oil refinery for QPSK- and 64QAM-modulated OFDM
teristics (highly reflective, as in the industrial facility, or lossy
signals for carrier frequencies of 2.4 GHz (circles and squares) and 4.95 GHz building materials, such as windows and brick in the apartment
(triangles and inverted triangles.) building). A second factor is an understanding of the trans-
mitted signal itself. If the signal is broadband, it is important
to understand the sensitivity of the modulation format to flat-
Even though significant multipath exists, this channel is prob-
ably usable. and frequency-selective multipath and attenuation. Finally, it is
Thus, using the frequency-domain metrics or the sum- necessary to select the appropriate measurement metrics to be
mary metrics, the oil refinery environment can be described as a able to assess the environment for a given application. These
channel that is low attenuation and flat fading under line-of-sight procedures, as we have tried to illustrate here, are not difficult,
conditions, and high-attenuation with frequency-selective but they do take some fore knowledge of the communication
fading when in non-line-of-sight conditions. Additionally (not scenario. Consideration of these concepts will help to enable
shown here), at the lower frequencies we saw attenuation due to reliable communications for the public-safety community. Fi-
lossy waveguide effects, where frequencies below the “cut-off” nally, we direct the reader to [5], [6] for data on other large
frequency of the tunnel were significantly attenuated [6], [31]. structures.

IV. CONCLUSION REFERENCES

[1] “SAFECOM Statement of Requirements,” vol. 1, ver. 1.2 [Online].
We studied radio-wave propagation in representative envi- Available:http://www.safecomprogram.gov/NR/rdonlyres/8930E37C-
ronments of interest to the public-safety community. We used C672-48BA-8C1B-83784D855C1E/0/SoR1_v12_10182006.pdf, Sec-
tion 3.3 and Section 3.5
measurement configurations that replicate response scenarios, [2] C. L. Holloway, G. Koepke, D. Camell, K. A. Remley, D. F. Williams,
including point-to-point communications at human height, and S. A. Schima, S. Canales, and D. T. Tamura, “Propagation and detec-
receive sites located where response vehicles may be placed. tion of radio signals before, during, and after the implosion of a 13-story
apartment building,” Nat. Inst. Stand. Tech. Note 1540, May 2005.
Our study focused on measurements of quantities of interest [3] C. L. Holloway, G. Koepke, D. Camell, K. A. Remley, D. F. Williams,
in design, testing, and standards development for broadband, S. A. Schima, S. Canales, and D. T. Tamura, “Propagation and detec-
modulated-signal transmissions. These measurements included tion of radio signals before, during, and after the implosion of a large
sports stadium (Veterans’ Stadium in Philadelphia),” Nat. Inst. Stand.
wideband penetration 1/(excess path-loss), in dB, measure- Tech. Note 1541, Oct. 2005.
ments at specific locations in each structure and measurements [4] C. L. Holloway, G. Koepke, D. Camell, K. A. Remley, S. A. Schima,
of digitally modulated signals under the OFDM protocol at car- M. McKinley, and R. T. Johnk, “Propagation and detection of radio
signals before, during, and after the implosion of a large convention
rier frequencies of 2.4 GHz and 4.95 GHz. From the wideband center,” Nat. Inst. Stand. Tech. Note 1542, Jun. 2006.
path-loss measurements, we calculated channel power and [5] C. L. Holloway, W. F. Young, G. Koepke, K. A. Remley, D. Camell,
RMS delay spread. From the modulated-signal measurements, and Y. Becquet, “Attenuation of radio wave signals coupled into
twelve large building structures,” Nat. Inst. Stand. Tech. Note 1545,
we calculated error vector magnitude. Apr. 2008.
Here we reported on the results for an 11-story apartment [6] K. A. Remley, G. Koepke, C. L. Holloway, C. Grosvenor, D. Camell,
building and an oil refinery. These tests show that a number J. Ladbury, R. T. Johnk, D. Novotny, W. F. Young, G. Hough, M.
D. McKinley, Y. Becquet, and J. Korsnes, “Measurements to support
of factors are needed to reliably define the propagation envi- modulated-signal radio transmissions for the public-safety sector,” Nat.
ronment. Use of a combination of summary metrics such as re- Inst. Stand. Tech. Note 1546, Aug. 2008.
ceived signal strength, RMS delay spread, and error vector mag- [7] [Online]. Available: http://www.nist.gov/eeel/electromagnetics/rf_
fields/wireless.cfm
nitude can provide a fairly complete picture of the propagation [8] [Online]. Available: http://www.antd.nist.gov/seamlessandsecure.
environment, necessary for reliable deployment of radio sys- shtml
tems, especially those employing wideband digital modulation. [9] D. Molkdar, “Review on radio propagation into and within buildings,”
Inst. Elect. Eng. Proc.-H, vol. 38, no. 1, pp. 61–73, Feb. 1991.
Because these summary measurements may be less expensive [10] S. R. Sounders, K. Kelly, S. M. R. Jones, M. Dell-Anna, and T. J. Har-
and complex to perform, use of summary metrics can save time rold, “The indoor-outdoor radio environment,” Inst. Elect. Eng. Elect.
and money. Comm. Eng. J., pp. 249–261, Dec. 2000.
[11] P. Papazian, “Basic transmission loss and delay spread measurements
In buildings where only one or two walls separate the for frequencies between 430 and 5750 MHz,” IEEE Trans. Antennas
transmit and receive antennas, such as the apartment building Propag., vol. 53, no. 2, pp. 694–701, Feb. 2005.

Authorized licensed use limited to: NIST Virtual Library (NVL). Downloaded on July 15,2010 at 17:30:04 UTC from IEEE Xplore. Restrictions apply.
1300 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 58, NO. 4, APRIL 2010

[12] E. S. Sousa, V. M. Jovanovic, and C. Daigneault, “Delay spread mea- Kate A. Remley (S’92–M’99–SM’06) was born in
surements for the digital cellular channel in Toronto,” IEEE Trans. Veh. Ann Arbor, MI. She received the Ph.D. degree in
Tech., vol. 43, no. 4, pp. 837–847, Nov. 1994. electrical and computer engineering from Oregon
[13] G. Calcev et al., “A wideband spatial channel model for system-wide State University, Corvallis, in 1999.
simulations,” IEEE Trans. Veh. Tech., vol. 56, no. 2, pp. 389–403, Mar. From 1983 to 1992, she was a Broadcast Engineer
2007. in Eugene, OR, serving as Chief Engineer of an
[14] J. R. Hampton, N. M. Merheb, W. L. Lain, D. E. Paunil, R. M. Shurfor, AM/FM broadcast station from 1989–1991. In
1999, she joined the Electromagnetics Division,
and W. T. Kasch, “Urban propagation measurements for ground based
National Institute of Standards and Technology
communication in the military UHF band,” IEEE Trans. Antennas
(NIST), Boulder, CO, as an Electronics Engineer.
Propag., vol. 54, no. 2, pp. 644–654, Feb. 2006. Her research activities include metrology for wire-
[15] R. J. C. Bultitude, Y. L. C. de Jong, J. A. Pugh, S. Salous, and less systems, characterizing the link between nonlinear circuits and system
K. Khokhar, “Measurement and modeling of emergency vehicle-to- performance, and developing methods for improved radio communications for
indoor radio channels and prediction of IEEE 802.16 performance for the public-safety community.
public safety applications,” IET Comm., vol. 2, no. 7, pp. 878–885, Dr. Remley was the recipient of the Department of Commerce Bronze and
Aug. 2008. Silver Medals and an ARFTG Best Paper Award. She is currently the Editor-in-
[16] W. F. Young, C. L. Holloway, G. Koepke, D. Camell, Y. Becquet, Chief of IEEE Microwave Magazine and Chair of the MTT-11 Technical Com-
and K. A. Remley, “Radio wave signal propagation into large building mittee on Microwave Measurements.
structures—Part 1: CW signal attenuation and variability,” IEEE Trans.
Antennas Propag., vol. 58, no. 4, p. , Apr. 2010.
[17] T. L. Duomi, “Spectrum considerations for public safety in the United
States,” IEEE Comm. Mag., vol. 44, no. 1, pp. 30–37, Jan. 2006. Galen Koepke (M’94) received the B.S.E.E. degree
[18] K. Balachandran, K. C. Budka, T. P. Chu, T. L. Duomi, and J. H. Kang, from the University of Nebraska, Lincoln, in 1973
“Mobile responder communication networks for public safety,” IEEE and the M.S.E.E. degree from the University of Col-
Comm. Mag., vol. 44, no. 1, pp. 56–64, Jan. 2006. orado at Boulder, in 1981.
[19] B. Davis, C. Grosvenor, R. T. Johnk, D. Novotny, J. Baker-Jarvis, and He is an NARTE Certified EMC Engineer. He
M. Janezic, “Complex permittivity of planar building materials mea- has contributed, over the years, to a wide range of
sured with an ultra-wideband free-field antenna measurement system,” electromagnetic issues. These include measurements
and research looking at emissions, immunity, elec-
Nat. Inst. Stand. Technol. J. Res., vol. 112, no. 1, pp. 67–73, Jan.–Feb.
tromagnetic shielding, probe development, antenna
2007.
and probe calibrations, and generating standard
[20] M. Riback, J. Medbo, J. Berg, F. Harryson, and H. Asplund, “Carrier electric and magnetic fields. Much of this work
frequency effects on path loss,” in Proc. 63rd IEEE Vehic. Technol. has focused on TEM cell, anechoic chamber, open-area-test-site (OATS),
Conf., 2006, vol. 6, pp. 2717–2721. and reverberation chamber measurement techniques along with a portion
[21] K. A. Remley, G. Koepke, C. Grosvenor, R. T. Johnk, J. Ladbury, D. devoted to instrumentation software and probe development. He now serves
Camell, and J. Coder, NIST Tests of the wireless environment in auto- as Project Leader for the Field Parameters and EMC Applications program in
mobile manufacturing facilities Nat. Inst. Stand. Tech. Note 1550, Oct. the Radio-Frequency Fields Group. The goals of this program are to develop
2008. standards and measurement techniques for radiated electromagnetic fields and
[22] J. C.-I. Chuang, “The effects of time delay spread on portable radio to apply statistical techniques to complex electromagnetic environments and
communications channels with digital modulation,” IEEE J. Sel. Areas measurement situations. A cornerstone of this program has been National
Comm., vol. SAC-5, no. 5, pp. 879–889, Jun. 1987. Institute of Standards and Technology (NIST), work in complex cavities such
[23] Y. Oda, R. Tsuchihashi, K. Tsuenekawa, and M. Hata, “Measured path as the reverberation chamber, aircraft compartments, etc.
loss and multipath propagation characteristics in UHF and microwave
frequency bands for urban mobile communications,” in Proc. 53rd
IEEE Vehic. Technol. Conf., May 2001, vol. 1, pp. 337–341.
[24] J. A. Wepman, J. R. Hoffman, and L. H. Loew, Impulse response mea- Christopher L. Holloway (S’86–M’92–SM’04–
surements in the 1850–1990 MHz band in large outdoor cells NTIA F’10) was born in Chattanooga, TN, on March
Rep. 94-309, Jun. 1994. 26, 1962. He received the B.S. degree from the
[25] IEEE Standard for Wireless LAN Medium Access Control (MAC) and University of Tennessee at Chattanooga in 1986, and
the M.S. and Ph.D. degrees from the University of
Physical Layer (PHY) Specifications: High-Speed Physical Layer in
Colorado at Boulder, in 1988 and 1992, respectively,
the 5 GHz Band, IEEE Standard 802.11a-1999.
both in electrical engineering.
[26] IEEE Standard for Wireless LAN Medium Access Control (MAC) and During 1992, he was a Research Scientist with
Physical Layer (PHY) Specifications: Higher-Speed Physical Layer Electro Magnetic Applications, Inc., Lakewood, CO.
Extension in the 2.4 GHz Band, IEEE Standard 802.11b-1999. His responsibilities included theoretical analysis
[27] M. D. McKinley, K. A. Remley, M. Myslinski, J. S. Kenney, D. and finite-difference time-domain modeling of
Schreurs, and B. Nauwelaers, “EVM calculation for broadband mod- various electromagnetic problems. From fall 1992 to 1994, he was with the
ulated signals,” in 64th ARFTG Conf. Dig., Orlando, FL, Dec. 2004, National Center for Atmospheric Research (NCAR), Boulder. While at NCAR
pp. 45–52. his duties included wave propagation modeling, signal processing studies,
[28] K. A. Remley, G. Koepke, C. L. Holloway, C. Grosvenor, D. G. Camell, and radar systems design. From 1994 to 2000, he was with the Institute for
and R. T. Johnk, “Radio communications for emergency responders in Telecommunication Sciences (ITS), U.S. Department of Commerce in Boulder,
high-multipath outdoor environments,” in Proc. Int. Symp. Advanced where he was involved in wave propagation studies. Since 2000, he has been
Radio Tech., Boulder, CO, Jun. 2008, pp. 106–111. with the National Institute of Standards and Technology (NIST), Boulder, CO,
[29] K. A. Remley, G. Koepke, C. L. Holloway, D. Camell, and C. where he works on electromagnetic theory. He is also on the Graduate Faculty
Grosvenor, “Measurements in harsh RF propagation environments to at the University of Colorado at Boulder.
Dr. Holloway was awarded the 2008 IEEE EMC Society Richard R. Stoddart
support performance evaluation of wireless sensor networks,” Sensor
Award, the 2006 Department of Commerce Bronze Medal for his work on radio
Rev., vol. 29, no. 3, 2009.
wave propagation, the 1999 Department of Commerce Silver Medal for his
[30] N. B. Carvalho, K. A. Remley, D. Schreurs, and K. G. Gard, “Multisine work in electromagnetic theory, and the 1998 Department of Commerce Bronze
signals for wireless system test and design,” IEEE Microw. Mag., pp. Medal for his work on printed circuit boards. His research interests include
122–138, Jun. 2008. electromagnetic field theory, wave propagation, guided wave structures, remote
[31] K. A. Remley, G. Hough, G. Koepke, D. Camell, C. Grosvenor, and R. sensing, numerical methods, and EMC/EMI issues. He is currently serving as
T. Johnk, “Wireless communications in tunnels for urban search and Co-Chair for Commission A of the International Union of Radio Science and
rescue robots,” in Proc. Performance Metrics for Intelligent Systems is an Associate Editor for the IEEE TRANSACTIONS ON ELECTROMAGNETIC
Workshop (PerMIS), Aug. 2008, pp. 236–243, NIST Special Publica- COMPATIBILITY. He was the Chairman for the Technical Committee on Compu-
tion 1090. tational Electromagnetics (TC-9) of the IEEE Electromagnetic Compatibility

Authorized licensed use limited to: NIST Virtual Library (NVL). Downloaded on July 15,2010 at 17:30:04 UTC from IEEE Xplore. Restrictions apply.
REMLEY et al.: RADIO-WAVE PROPAGATION INTO LARGE BUILDING STRUCTURES—PART 2: CHARACTERIZATION 1301

Society from 2000–2005, served as an IEEE Distinguished lecturer for the John Ladbury (M’92) was born Denver, CO, 1965.
EMC Society from 2004–2006, and is currently serving as Co-Chair for the He received the B.S.E.E. and M.S.E.E. degrees in
Technical Committee on Nano-Technology and Advanced Materials (TC-11) signal processing from the University of Colorado,
of the IEEE EMC Society. Boulder, in 1987 and 1992, respectively.
Since 1987 he has worked on EMC metrology
and facilities with the Radio Frequency Technology
Division, National Institute of Standards and Tech-
Chriss A. Grosvenor (M’91) was born in Denver, CO. She received the B.A. nology (NIST), Boulder, CO. His principal focus
degree in physics and M.S. degree in electrical engineering from the University has been on reverberation chambers, with some
of Colorado, Boulder, in 1989 and 1991, respectively. investigations into other EMC-related topics such as
In 1990, she joined the Electronics and Electrical Engineering Laboratory, time-domain measurements and probe calibrations.
National Institute of Standards and Technology (NIST), Boulder, CO. Her work He was involved with the revision of RTCA DO160D and is a member of the
at NIST includes design and analysis of large stripline cavities for materials IEC joint task force on reverberation chambers.
measurements as well as a large 77 mm diameter coaxial system and a 60 GHz Mr. Ladbury has received three “best paper” awards at IEEE International
Fabry-Perot resonator. She has worked in the noise temperature project and as- EMC symposia over the last six years.
sembled the 1 to 12 GHz automated systems and repackaged the 30 and 60 MHz
noise temperature measurement systems. She joined the time-domain project in
2002 and has worked on measurements of shielding effectiveness of aircraft in-
cluding the orbiter Endeavour. She has authored papers in all of these technical Robert T. Johnk, photograph and biography not available at the time of
areas. publication.
Ms. Grosvenor was awarded the Bronze Medal for her work with the orbiter
Endeavour.

William F. Young (M’06) received the B.S. degree

in electronic engineering technology from Central
Dennis Camell (M’94) received the B.S. and M.E. Washington University, Ellensburg, in 1992, the M.S.
degrees in electrical engineering from the University degree in electrical engineering from Washington
of Colorado, Boulder, in 1982 and 1994, respectively. State University at Pullman, in 1998, and the Ph.D.
From 1982 to 1984, he worked for the Instrumen- degree from the University of Colorado at Boulder,
tation Directorate, White Sands Missile Range, NM. in 2006.
Since 1984, he has worked on probe calibrations and Since 1998, he has worked for Sandia National
EMI/EMC measurements with the Electromagnetics Laboratories in Albuquerque, NM, where he is
Division, National Institute of Standards and Tech- currently a Principal Member of the Technical Staff.
nology (NIST), Boulder, CO. His current interests His work at Sandia includes the analysis and design
are measurement analysis (including uncertainties) of cyber security mechanisms for both wired and wireless communication
in various environments, such as OATS and ane- systems used in the National Infrastructure and the Department of Defense.
choic chamber, and development of time domain techniques for use in EMC He has also been a Guest Researcher at the National Institute of Standards and
measurements and EMC standards. He is involved with several EMC working Technology in Boulder, CO, from 2003 to 2009, and is working on improving
standards committees and is chair of ANSI ASC C63 SC1. wireless communication systems for emergency responders. His current
research interests are in electromagnetic propagation for wireless systems,
and the impacts of the wireless channel on overall communication network
behavior.

Authorized licensed use limited to: NIST Virtual Library (NVL). Downloaded on July 15,2010 at 17:30:04 UTC from IEEE Xplore. Restrictions apply.