2192

IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 62, NO. 4, APRIL 2014

Site-Specific Models of the Received Power for

Radio Communication in Urban Street Canyons
Jonathan S. Lu, Student Member, IEEE, Henry L. Bertoni, Life Fellow, IEEE, Kate A. Remley, Fellow, IEEE,
William F. Young, Member, IEEE, and John Ladbury, Member, IEEE

AbstractThis paper presents site-specific models for the

real-time prediction of the received power from waves propagating through urban street canyons (i.e., in the horizontal
plane containing the transmitter and receiver) for radio communication in urban environments. The line-of-sight (LOS) and
non-line-of-sight (NLOS) models presented here are based on the
two-ray model and are used to predict the small-area average
received power (i.e., long-term/shadow fading and distance-dependence). These models have two adjustable parameters that
account for clutter such as road traffic and pedestrians, and
for scattering from objects and buildings at street intersections,
respectively. Validation of these models is performed with mobile-to-mobile measurements recorded in the high-rise sections of
Denver and New York City for frequencies ranging from 430 MHz
to 4.86 GHz.
Index TermsChannel model, diffraction, DSA, LOS, MANET,
NLOS, path loss, scattering, urban propagation, VANET.

I. INTRODUCTION

HIS work presents computationally efficient site-specific

models of the received power for peer-to-peer (mobile-to-mobile), backhaul and access radio links. These links
comprise various narrowband (e.g., police and emergency
responder systems) and wideband (e.g., MANET, VANET, and
WSN) networks operating in urban street canyons. For such
radio links, these models can be used for many applications,
including 1) system planning via Monte Carlo simulations
of performance [1]; 2) interference/connectivity calculations
for dynamic spectrum access (DSA) [2]; and 3) real-time
routing [3].
Our goal for accurate real-time predictions requires computation time less than a second. This fast computation time goal
can be achieved using non-site-specific models of distance dependence, such as empirical statistical models (e.g., [4] and [5]).
Manuscript received July 14, 2013; revised October 17, 2013; accepted
November 29, 2013. Date of publication January 02, 2014; date of current
version April 03, 2014. This work was funded by the Wireless Internet Center
for Advanced Technology (WICAT), NSF I/UCRC, New York University
Polytechnic School of Engineering. The Denver measurements used in this
work were funded by the NIST Public Safety Communications Research
Laboratory.
J. S. Lu and H. L. Bertoni are with the NYU Wireless Center of the New
York University Polytechnic School of Engineering, Brooklyn, NY 11201 USA
(e-mail: lushiaoen@gmail.com; hbertoni@poly.edu).
K. A. Remley, W. F. Young, and J. Ladbury are with the Electromagnetics
Division, National Institute of Standards and Technology (NIST), Boulder,
CO 80305 USA (e-mail: kate.remley@nist.gov; william.young@nist.gov;
john.ladbury@nist.gov).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TAP.2013.2297164

Fig. 1. LOS and NLOS one-turn and two-turn radio links.

Because these models cannot account for the site-specific layout

of a radio link relative to its environment, they are prone to large
prediction errors with standard deviation
10 dB [6]. Alternatively, site-specific tools such as ray-tracing codes [7][11]
can be used for more accurate predictions of path loss with stan3 dB [7]. Unfortunately,
dard deviation of error as low as
the greater accuracy comes with data and computational costs.
Computation time for a single radio link may take many minutes, or even hours.
To reduce running time, some ray-tracing codes limit the
types of rays traced to those in the vertical plane (VP) and
horizontal plane (HP) containing the transmitting and receiving
antennas [8][11]. However, even the limited 2D ray-tracing in
the HP cannot satisfy our computation time goal. Note that the
contributions from waves in the HP, those propagating through
the street canyons, as suggested in Fig. 1, are significant for
radio links operating well below the rooftops. These contributions will even dominate in high-rise urban environments
where diffraction over the rooftops significantly reduces the
VP contribution. Our HP models account for the shadowing by
buildings without tracing many rays to find the few reaching
the receiver.
Various measurements of received power in line-ofsight (LOS) links have been made in open regions and
urban canyons using frequencies from 200 MHz to 3 GHz [5],
[12][16]. It was found that the distance dependence of the
received power follows that of the two-ray model, consisting
of the coherent addition of direct and ground-reflected rays.
For non-line-of-sight (NLOS) links, propagation takes place
via rays reflected and scattered by building surfaces [16][19],
diffracted at building corners [20][23], and scattered by lampposts, etc., near the street intersections [24], [25].
The signal arriving at a receiver via these multiple rays are
found to fade over a scale length of about half a wavelength
(referred to as fast, short-term or Rayleigh fading). These fades

0018-926X 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

LU et al.: SITE-SPECIFIC MODELS OF THE RECEIVED POWER FOR RADIO COMMUNICATION IN URBAN STREET CANYONS

can be 20 dB or more. Measurement procedures often remove

this fading by averaging the received power over a small area
[12]. The resulting small-area
or along a distance of about
average signal strength exhibits the range dependence and local
long-term (shadow) fading.
This paper proposes and validates simple models of the smallarea average power due to propagation in the HP for LOS links
and two types of NLOS links. LOS links arise when the mobiles
are on the same straight section of a street, while NLOS links
result from building blockage when the mobiles are 1) on intersecting streets and 2) on parallel streets, as shown in Fig. 1.
These models combine the two-ray distance dependence with
all of the corner-turning effects into simple expressions with adjustable parameters. Since minimal information about the environment is required, the models are computationally efficient.
Running times less than one second were achieved using the
implementation outlined in [26].
The organization of this paper is as follows. The LOS and
NLOS models are developed in Section II. An overview of our
430, 750,
Denver and New York City measurements for
905, 1834, 2400, and 4860 MHz, is presented in Section III.
The Denver measurements are used to validate the LOS and
NLOS models in Section IV. In Section V, an empirical model
for the scattering/diffraction parameter is developed and factors
affecting this parameter are discussed.
II. MODELS FOR LOS AND NLOS LINKS
For radios operating in an urban environment with a locally
rectangular street grid, as in Fig. 1, there are three types of radio
links in the HP. LOS links arise when the mobiles are on the
same straight section of a street. NLOS links arise when the mobiles are located on two streets that intersect, or are parallel to
each other. For mobiles on intersecting streets, the radio signal
propagating in the HP can reach the receiver after turning one
corner. We call this radio link a one-turn link. When the mobiles are on parallel streets, the radio signal propagating from
one to the other must turn two corners. We call this type of link
a two-turn link. Other paths in the HP connect the mobiles in
these three cases but involve turning more corners. Since turning
a corner involves substantial loss in amplitude [16], these other
paths are neglected. For one-turn and two-turn links, the two-ray
model used for LOS links is modified to account for diffraction and scattering by vertical structures, such as lampposts and
building corners.
A. LOS Links
In [5] and [13], the distance dependence of the received
power for LOS transmitter (TX) and receiver (RX) was found
to follow the two-ray model. This model coherently adds the
electric fields of the direct unobstructed ray and ground-reflected ray. For isotropic, matched antennas, the received power
in watts is given by [12]

(1)

2193

Fig. 2. LOS received power predictions for TM- and TE- polarized 900 MHz
1 W, antenna heights
1.6 m, and ground
links with
.
relative permittivity of

Here, is the horizontal separation between the RX and TX

antennas, is the wavelength,
is the TX antenna height,
and
is the RX antenna height, all in meters.
is the
is the reflection coefficient
transmitted power in watts and
of the ground.
An example of the received power at 900 MHz between two
1.6 m with
1 W,
antennas of height
is plotted as a function of in Fig. 2 for TM (red dotted line)
and TE (black solid line) polarizations. The ground is assumed
. As
smooth and flat and has a relative permittivity of
seen in Fig. 2, for smaller values of ,
has a variation
about an amplitude decrease of
. For larger values of ,
has an amplitude decrease of
[12].
To account for the vehicular and pedestrian clutter, Oda and
Tsunekawa [14], and Masui, et al. [15], [16] introduced an effrom which to measure the antenna
fective ground height
height. They observed that the clutter on urban LOS links at
midday caused the ground to appear to be between 1 m to 2 m
higher than it was. They also observed that around midnight,
when there was less traffic, the effective ground appeared to
be only 0.5 m higher. The effective ground height serves as an
adjustable parameter in our LOS and NLOS models and can
be estimated from measurements or from results presented in
[14][16].
B. One-Turn NLOS Links
Radio signals can turn a corner from the street where the TX
is located, to a cross street by means of reflection from building
surfaces, and diffraction and/or scattering from building corners, lampposts, and other tall vertical structures in the street
intersection. Reflection is important only near the intersection
since reflected rays enter the side street at angles near normal
to the buildings on the side street. Because of the near-normal
incidence, reflected rays lose amplitude after a few reflections
before traveling very far down the side street. In contrast, the
diffracted and scattered rays experience significant loss from
turning the corner but have a weak dependence on distance
along the side street [27].
Fig. 3 shows direct and ground-reflected rays diffracted at a
vertical edge, such as the corner of a building. If the distance

2194

IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 62, NO. 4, APRIL 2014

Fig. 3. Two-ray diffraction at a vertical edge.

Fig. 4. Two-ray diffraction at two vertical edges.

traveled by the rays along the ground from the TX antenna to RX antenna is large compared with the antenna heights
and
, then the diffraction loss at the corner will be
nearly the same for both the direct and ground-reflected rays.
Thus, for this pair of rays, diffraction will introduce an additional loss [28] to the received power, such that

The parameter
is the second adjustable parameter in
our models and is selected to describe the diffraction/scattering mechanisms at all intersections of a particular urban
environment.

(2)
is the received power given by (1) for
Here,
distance
, and
is the diffraction coefficient of the
and
are sufficiently large relative to
edge. Note that when
the street widths, can be approximated as a constant, because
the diffraction angle approaches 90 .
For a one-turn link in an urban environment, there are usually four corners at an intersection, as seen in Fig. 1. In addition
to corners, there may be tall vertical scatterers such as lampposts near the intersection. Each scatterer will contribute two
rays (direct and ground reflected) whose fields must be added
coherently. The power from each ray pair is then expressed as
replaced by a scattering coefficient [29].
in (2) with
Ray pairs arriving from distinct directions produce a standing
wave pattern whose small-area average power is the incoherent
sum of the power carried by the individual ray pairs [12]. Accordingly, for the average power of a one-turn link, the power
from the ray pair for the th corner/scatterer is first found,
and then summed incoherently over all . Note that direct and
ground-reflected rays of a pair are not separated because they
arrive almost parallel from the same direction.
Under the assumption that the TX and RX distances to and
from the intersection are larger than the street width, the disfrom the TX antenna to the th corner/scatterer can
tance
then be approximated as the distance
from the TX antenna
to the center of the intersection, as shown in Fig. 1. Similarly,
from the th corner/scatterer to the RX antenna
the distance
can be approximated by the distance
from the center of the
intersection to the RX antenna. This assumption is equivalent to
placing the vertical structures at the center of the intersection.
Furthermore, instead of separately computing different diffraction and scattering coefficients for the individual corners and
vertical scatterers, we employ a diffraction/scattering parameter
, which is the incoherent sum of the magnitude squared of the
individual diffraction and scattering coefficients. The total avcan then be expressed as
erage power
(3)

C. Two-Turn NLOS Links

When the TX and RX are located on parallel streets, the signal
components propagating through the street canyons must turn
at two intersections. When the TX and RX are located away
from the intersections, the dominant contributions to the received signal come from rays either diffracted by the building
corners or scattered by objects near the intersections. For the
case of diffraction around two corners, the two-ray model is
and
are much smaller
shown in Fig. 4. Assuming that
than the horizontal distance traveled
, the received
power is found in a similar fashion as diffraction from a single
corner. Accounting for the additional loss caused by the diffractions at the two corners, the small-area average received power
is
(4)
where
and
are the diffraction coefficients of the first and
second corners.
An expression for a two-turn path must account for the four
corners in the first and second intersections and the vertical
structures located at each intersection. To arrive at a simplified expression for the two-turn path, we use a similar approach
to the one-turn expression (3), where the th and th corners/
scatterers of the first and second intersections, respectively, are
placed at the centers of their respective intersections. The effects
of the corners and scatterers at each intersection are then comand , rebined into the diffraction/scattering parameters
spectively. Thus, the small-area average received power is given
by

(5)
In a particular urban environment, and without further information, it can be assumed that
, where is the same
value found for one-turn links.
Unlike LOS and one-turn links on a rectangular street grid in
an urban environment, there are many two-turn paths by which
a signal may travel to the RX of a two-turn link, as seen in

LU et al.: SITE-SPECIFIC MODELS OF THE RECEIVED POWER FOR RADIO COMMUNICATION IN URBAN STREET CANYONS

2195

TABLE I
MEASUREMENT SETUP PARAMETERS

Fig. 5. RX locations and TX walking routes in Denver.

Fig. 6. RX locations and TX walking routes in New York.

Fig. 1. An expression for the total small-area spatially averaged

that considers all two-turn paths is
received power
(6)
where the superscript denotes the th two-turn path. From
simulation results [26] not shown, only those two-turn paths
segments lay inside or immediately outside the rewhose
gion between the TX and RX are important. Therefore, the summation in (6) can usually be truncated to include only a few
two-turn paths.
III. MEASUREMENT OVERVIEW
Measurements were separately carried out in high-rise sections of Denver [4] and New York [30] to simulate public-safety
personnel walking through a high-rise urban environment while
communicating with a mobile command center. As such, an operator carried single-frequency (continuous wave (CW)) transmitters (TXs) along the paths shown in Figs. 5 and 6, while receivers (RXs) at multiple locations recorded the received signal
level.
A. Transmitter and Receiver Setup and Equipment
The transmitter part of the measurement system consisted
430, 750, 905, 1834,
of handheld CW TXs operating at
2400, and 4860 MHz. The TXs had either linear dipole arrays
or monopole antennas that radiated either 1- or 2-W nominal
. The
and antenna gains
for the different
power
frequencies in each city are given in Table I. The TX antennas

were always held at a height of

1.5 m and approximately
0.33 m away from the torso of the operator carrying them.
The receiver part of the measurement system consisted of an
antenna, spectrum analyzer, laptop, and controlling software.
Linear dipole arrays, discone, and conical antennas were used.
for each frequency by city is given in
The RX antenna gain
Table I. In Denver, the RX antennas were held by tripods
2 m high at four sites (RX1RX4) denoted by the crosses in
Fig. 5. For the New York City measurements, the RX antenna
site is denoted by the cross in Fig. 6. There, the RX antennas
were mounted on top of an emergency vehicle at a height of
4 m.
The cable loss
values in the New York measurements
values were not
were measured to be 4 dB. In Denver,
were estimated as
measured. Instead, effective values
the difference between LOS measurements and received power
predictions found using (1) with effective ground height
1.2 m and the
,
, and
values given in Table I. The
resulting values ranged from 0.8 to 18.8 dB and are listed in
Table I. Note that the values of
are directly affected by
so that a low
results in greater predicted rethe choice of
ceived powers, which, in turn, results in smaller
values.
B. Description of TX Route and RX Locations
The measurement collections in Denver and New York City
were performed during the daytime when there was significant
vehicular traffic on the streets, as well as pedestrian traffic on
the sidewalks. The TXs were carried on the sidewalks along the
walking routes shown in Figs. 5 and 6. The RX antennas were
positioned at four sites in Denver and one site in New York City.
The numbering of the Denver RX sites used here differs from
that used in [4]. The measurements reported in [4] also include
an additional RX site and TX walking route located in a narrow
pedestrian mall. Since the mall structure does not match that of
a street, the measurements from those locations were not used
in this study. The RX1 site was located on a sidewalk near a
street intersection, while RX2 was located on a street 44 m
from the center of the nearest street intersection. RX3 and RX4
were located in parking lots.
The blocks in the measurement areas had both rectangular
and trapezoidal shapes so that the routes included corners that
were not 90 , as seen for the corner near position 5 in Fig. 5

2196

IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 62, NO. 4, APRIL 2014

and corners 3 and 4 in Fig. 6. Also, in Denver, the paths between some of the TX locations, and RX3 and RX4, which
were in parking lots, turn corners of less than 90 . As will be
shown in Section IV, our models provide good predictions even
in the presence of such nonideal street configurations and corner
angles.
C. Measurement Collection and Processing
The laptop and spectrum analyzer at the RX recorded
a narrow frequency band (called the capture bandwidth)
around the nominal center frequency of the transmitted CW
signal. The capture bandwidth was typically less than 20 kHz.
The peak signal within the capture bandwidth was recorded as
. As with similar urban
the measured CW signal power
RF propagation measurements, instrument measurement uncertainty is expected to be negligible relative to the RF channel
variability [31].
The time resolution of the received signals was determined by
the sampling rate of the complete measurement process. Sampling rates in Denver were 1 sample/s for all frequencies. Sampling rates in New York were 3.3 and 5 samples/s for 905 and
1834 MHz, respectively. Accounting for a 1-m/s walking speed,
the spatial separation of our Denver measurements was approximately 1 m. In New York, it was 0.3 and 0.2 m for 905 and
1834 MHz, respectively
D. Determining the Equivalent Isotropic, Small-Area Spatially
Averaged Power
To compare with our models for radio links with lossless
and
and cable
isotropic antennas, the antenna gains
were removed from the measured data in postproloss
cessing. First, small-area spatial averaging was carried out for
each TX antenna location. To do this, the average power
was computed by averaging, in watts, all
samples recorded within
of the TX antennas location. Then,
,
, and
values given in Table I, the
using the
values in decibel watts were converted using the
expression
(7)

IV. COMPARISON WITH DENVER MEASUREMENTS

In this section, the effective ground height
and diffraction/scattering parameter are estimated from the isotopic lossless measurements recorded at RX1 in Denver. To validate the
received power models presented in Section II, predictions are
and , and
computed from the models using the estimated
then compared with measurements recorded at RX2, RX3, and
RX4.
A. Extracting the Effective Ground Height
To extract
from the measurements, a 0-dB mean-error
constraint was applied to LOS predictions from (1) for measurements recorded at RX1 when the TX antenna was within 60 m
of RX1. Here, error is defined as the measured received power
in decibel watts minus the predicted received power in decibel

Fig. 7. Denver
tions for RX1.

750 MHz LOS and one-turn measurements and predic-

watts. Considering measurements at all frequencies listed in

Table I,
was found to be 1.2 m, which is consistent with the
1.2 m was used in the
findings in [14][16]. The value
750-MHz LOS predictions, plotted as the black solid curves in
Fig. 7. The LOS predictions and measurements compare well,
even outside of our fitting range, that is, for distances greater
than 60 m from the RX antenna. The mean LOS prediction error
is 1.1 dB, and the standard deviation of the prediction error is
4.3 dB.
B. Extracting the Diffraction/Scattering Parameter S
To extract a value of for each frequency listed in Table I, a
0-dB mean-error constraint was again applied when comparing
predictions computed with the one-turn model of (3) to the
NLOS Denver measurements recorded at location RX1 in
Fig. 5. Note that all NLOS measurements recorded at RX1 are
one-turn, as seen in Fig. 5. The values found for are [4.1,
2.9, 7.1, 6.4, 4.4, 2.8] at
430, 750, 905, 1834, 2400, and
4860 MHz, respectively. As an example, the predictions from
the one-turn model of (3) are plotted as the black dashed line
750 MHz. The mean
in Fig. 7 using the zero-mean for
deviation of the measurements from the model is 0 dB with
5.2-dB standard deviation.
C. Link Classification
In order to generate predictions for the various TX and RX antenna locations, the radio links must be classified as either LOS,
one-turn or two-turn. This classification was performed visually
for all TX antenna locations on the route shown in Fig. 5 and all
RX antenna sites. For NLOS links to RX2, the turn locations
were taken as the centers of the street intersections. To determine the turn locations for NLOS links to RX3 and RX4, sections of the parking lots were treated as streets. As an example,
the turn locations for RX3 are shown in Fig. 8. Four TX antenna
locations are also shown in this figure. Each TX has a one-turn
radio link with RX3. In the case of TX3 and TX4, the turn location is at the center of a street intersection, as denoted by the red
and
were measured from this point.
dot. The distances
For TX1 and TX2, sections of the propagation paths were taken
along the inner boundaries of the parking lot, as shown in Fig. 8,

LU et al.: SITE-SPECIFIC MODELS OF THE RECEIVED POWER FOR RADIO COMMUNICATION IN URBAN STREET CANYONS

2197

Fig. 8. One-turn radio link examples for RX3.

TABLE II
LINK CLASSIFICATION PERCENTAGES

with the turn locations in the middle of the streets away from the
street intersection. i h
Link classifications are indicated by color in Fig. 9(a)(c) for
the 750-MHz measurements recorded at RX2, RX3, and RX4.
For the predictions presented later, the classifications are indicated by line type. Blue denotes LOS measurements, while
red and green denote one-turn and two-turn measurements, respectively. The yellow color denotes measurements recorded
on three-turn links in which the wave propagating through the
street canyons must turn at three intersections to reach the RX.
This classification arises for a limited percentage of links because the Denver block grid is not uniform and perfectly rectangular. It should be noted that in three-turn links, the waves
propagating over the top of the buildings (i.e., in the VP) may
be dominant relative to those traveling in the HP. The percentages of LOS, one-turn, two-turn, and three-turn measurements
are listed in Table II. These percentages are nearly identical for
measurements recorded for all frequencies at the same RX antenna location due to the roughly constant sampling rate and
walking speed.
D. Comparison of Model Predictions to RX2, RX3, and RX4
Measurements
To show that only a limited number of measurements are
needed for accurate characterization of LOS and NLOS propand exagation in urban street canyons, the values of
tracted from the RX1 measurements were used to predict the
received power recorded at RX2, RX3, and RX4. Examples of
the 750-MHz predictions are plotted over the measurements in
Fig. 9(a)(c). Predictions for LOS links are shown as thick solid
black curves, while those for one-turn links and two-turn links
are shown as dashed and dotted black colored curves, respectively. The predictions and measurements generally compare
well in the prediction of the distance dependence, even near to
the intersections. In addition, predictions using the value of
and extracted from one-turn measurements recorded at RX1,

Fig. 9. Denver
750 MHz measurements and predictions for (a) RX2,
(b) RX3, and (c) RX4.

give good comparison with the measurements recorded at RX2,

RX3, and RX4, even for the prediction of two-turn links.
To quantitatively characterize the accuracy of our models,
prediction error statistics are computed. A focus is placed on
the NLOS model results. The LOS model accuracy has been
well characterized in [5], [13][16] and is seen to compare well
with our 750-MHz measurements in Figs. 7 and 9(a)(c) with
1.9-dB mean error and 4-dB standard deviation of error calculated over all RX sites. The NLOS mean and standard deviation
of errors for all Denver RX antenna locations and frequencies
are plotted in Fig. 10(a) and (b), respectively. In most cases,

2198

IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 62, NO. 4, APRIL 2014

Fig. 11. New York city

1834 MHz measurements, and LOS (1), one-turn
(3), and two-turn (6) models predictions.

Fig. 10. Denver NLOS measurements: (a) mean errors and (b) standard deviation of errors.

TABLE III
DIFFRACTION/SCATTERING PARAMETER

For measurements recorded at the same frequency in the same

city, the values of are consistent, as shown in Table III. This
further supports our approach in which a single value for is
used for the NLOS power predictions in a city. On the other
hand, for different cities, the values at a specific frequency
may vary. This may be due to the street furniture [24], [25],
building architecture [32], and/or block layout unique to each
environment.
To develop a model for , the values in Table III were scatter. A least-squares error fit to the
plotted in decibels versus
scatter plot was then computed to obtain the power law dependence on in hertz. The resulting dependence in linear terms is
(8)

the mean-error magnitude is less than 2.5 dB, and the standard
deviation is less than 6 dB. The standard deviation of the error
increases with frequency, as seen in Fig. 10(b). This increase
is partly due to the limited spatial resolution of our measurements. Because the physical window for small-area spatial averaging decreases as frequency increases, fewer samples were
included in the average (e.g., one or two samples at 4.86 GHz).
Thus, the small-scale (short-term) fading was not completely
averaged out, and the resulting variation is reflected in the standard deviation of the error.

Note that if diffraction at building corners is the only propagation mechanism by which a signal turns a corner, then the
theory of diffraction by a wedge suggests that the frequency de, rather than the weak dependence in
pendence should be
(8). However, as discussed previously, the signal may also turn
the corner as a result of scattering at lampposts and other vertical objects [24], [25]. In [24], results show that the contribution of scattering from a single lamppost can be equal to that
due to diffraction from four corners. For UHF frequencies, the
scattering cross section of a cylindrical metal lamppost is approximately frequency independent [29], which may explain the
weak frequency dependence in (8).

V. CHARACTERIZING THE DIFFRACTION/SCATTERING

PARAMETER

VI. CONCLUSION

In the previous section, measurements recorded at RX2, RX3,

and RX4 were compared with predictions made using the values
and extracted from measurements at RX1. In this secof
tion, measurements recorded at RX2, RX3, and RX4 were reprocessed to extract values of using the 0-dB mean-error criterion and
1.2 m, as in Section IV-B. The resulting values
of are listed in Table III. Measurements recorded in New York
905 MHz and 1834 MHz were similarly processed
City at
to obtain values of using
1.33 m. The resulting values
of are also listed in Table III. An example of the New York
1834 MHz are given
measurements and predictions for the
in Fig. 11. Again, the LOS and NLOS models are seen to compare well with the small-area-averaged measurements.

In this paper, we presented computationally efficient urban

propagation models to compute the horizontal-plane contribution to the total power for LOS and NLOS radio links in urban
environments. Because the horizontal-plane contribution is
dominant in high-rise urban environments, we have validated
these models against measurements conducted in the high-rise
portions of Denver and New York City over a wide range of
frequencies.
The LOS and NLOS models presented here require minimal
information about the environment: 1) link classification and 2)
pertinent distances for the models, i.e., , , , and . Our
models use two site-specific empirical parameters to account
for the clutter in the streets such as pedestrians and vehicles,

LU et al.: SITE-SPECIFIC MODELS OF THE RECEIVED POWER FOR RADIO COMMUNICATION IN URBAN STREET CANYONS

and diffraction and scattering near street intersections, respectively. The effective ground height
accounts for the clutter
and is frequency independent, with values of 1.2 m in Denver
and 1.33 m in New York City. The diffraction/scattering parameter accounts for the site-specific vertical features in and near
the street intersections, such as corner composition, lampposts,
and acute corner angles, and has values ranging from 1.7 to 7.1.
Comparisons show that at the same frequency and city, values
of are fairly consistent.
For cases in which NLOS measurements are not readily available, we have developed a simple frequency dependent model
for . It remains to develop a more complete model of that
accounts for other site-specific features, such as street width and
corner construction.
As discussed, the models presented here only consider waves
propagating in the horizontal plane. These models have been
shown to be sufficient for power prediction in high-rise urban
environments. To perform predictions in an urban environment
with lower buildings, it would be necessary to complement the
NLOS models given here with models of propagation over the
buildings in the vertical plane, such as those discussed in [26]
and [33].
ACKNOWLEDGMENT
The authors express their appreciation to C. Chrysantos and
Dr. J. Boksiner of CERDEC, U.S. Army, for many stimulating
discussions and helpful comments. NIST acknowledges The
U.S. Department of Homeland Security (DHS) Science and
Technology Directorates Standards Branch, P. Mattson, Program Manager, which provided funding for the New York City
measurements. For their assistance with the Denver measurements, the authors thank C. Holloway, G. Koepke, J. Coder,
and D. Camell of NIST and D. Matolak of the University of
South Carolina.
REFERENCES
[1] I. K. Eltahir, The impact of different radio propagation models for mobile ad hoc networks (MANET) in urban area environment, presented
at the Wireless Broadband and UWM Comm. Conf., Sydney, Australia,
Sep. 2007.
[2] ECC, Technical and operational requirements for the possible operation of cognitive radio systems in the white spaces of the frequency
band 470790 MHz, ECC meeting, Cardiff, Wales, 2011, ECO WG
SE43, ECC Rep. 159.
[3] J. Punnoose, P. V. Nikitin, J. Broch, and D. D. Stancil, Optimizing
wireless network protocols using real-time predictive propagation
modeling, in Proc. Radio Wireless Conf. (RAWCON), Aug. 1999, pp.
3944.
[4] D. W. Matolak, Q. Zhang, and Q. Wu, Path loss in an urban peer-topeer channel for six public-safety frequency bands, IEEE Wireless
Commun. Lett., vol. 2, no. 3, pp. 263266, 2013.
[5] J. R. Hampton, N. M. Merheb, W. L. Lain, D. E. Paunil, R. M. Shuford, and W. T. Kasch, Urban propagation measurements for ground
based communication in the military UHF band, IEEE Trans. Antennas Propag., vol. 54, no. 2, pp. 644654, Feb. 2006.
[6] C. Phillips, D. Sicker, and D. Grunwald, Bounding the error of path
loss models, in Proc. IEEE Symp. New Frontiers Dynam. Spectrum
Access Netw. (DySPAN), May 2011, pp. 7182.
[7] G. E. Athanasiadou, A. R. Nix, and J. P. McGeehan, A microcellular
ray-tracing propagation model and evaluation of its narrow-band and
wide-band predictions, IEEE J. Sel. Areas Commun., vol. 18, no. 3,
pp. 322335, Mar. 2000.
[8] K. Rizk, J. F. Wagen, and F. Gardiol, Two-dimensional ray-tracing
modeling for propagation prediction in microcellular environments,
IEEE Trans. Veh. Technol., vol. 46, no. 2, pp. 508518, May 1997.
[9] Wavecall website [Online]. Available: http://www.wavecall.com/

2199

[10] T. Krner, D. J. Cichon, and W. Wiesbeck, Concepts and results for

3D digital terrain-based wave propagation models: An overview,
IEEE J. Sel. Areas Commun., vol. 11, no. 7, pp. 10021012, Sep. 1993.
[11] K. Rizk, R. Valenzuela, S. Fortune, D. Chizhik, and F. Gardiol, Lateral, full-3D and vertical plane propagation in microcells and small
cells, in Proc. IEEE Veh. Technol. Conf., 1998, pp. 9981003.
[12] H. L. Bertoni, Radio Propagation for Modern Wireless Applications.
Upper Saddle River, NJ, USA: Prentice-Hall, PTR, 2000.
[13] H. H. Xia, H. L. Bertoni, L. R. Maciel, A. Lindsy-Stewart, and R.
Rowe, Radio propagation characteristics for line-of-sight microcellular and personal communications, IEEE Trans. Antennas Propag.,
vol. 41, no. 10, pp. 14391447, 1994.
[14] Oda and K. Tsunekawa, Advanced LOS path loss model in microwave mobile communications, in Proc. IEEE Ant. Propag. Conf.,
Apr. 1997, vol. 2, pp. 170173.
[15] H. Masui, M. Ishii, K. Sakawa, H. Shimizu, T. Kobayashi, and M.
Akaike, Microwave propagation characteristics in an urban LOS
environment in different traffic conditions, in Proc. IEEE Antennas
Propag. Symp., 2000, pp. 11501153.
[16] H. Masui, M. Ishii, K. Sakawa, H. Shimizu, T. Kobayashi, and M.
Akaike, Microwave path-loss characteristics in urban LOS and
NLOS, in Proc. IEEE Veh. Technol. Conf., 2001, pp. 395398.
[17] J. H. Tarng and K. M. Ju, A novel 3-D scattering model of 1.8-GHz
radio propagation in microcellular urban environment, IEEE Trans.
Electromagn. Compat., vol. 41, no. 2, pp. 100106, 1999.
[18] M. R. J. A. E. Kwakkernaat and M. H. A. J. Herven, Analysis of
scattering in mobile radio channels based on clustered multipath estimates, Int. J. Wireless Inf. Netw., Mar. 24, 2009, 10,1007/s10776009-0096-y.
[19] V. Erceg, S. Ghassemzadeh, M. Taylor, D. Li, and D. L. Schilling,
Urban, suburban out-of-sight propagation modeling, IEEE Commun.
Mag., vol. 30, no. 6, pp. 5661, Jun. 1992.
[20] H. M. El-Sallabi, Fast path loss prediction by using virtual source
technique for urban microcells, in Proc. IEEE Veh. Technol. Conf.,
2000, vol. 3, pp. 21832187.
[21] S. Y. Tan and H. S. Tan, Propagation model in an urban street scene for
microcellular communications, IEEE Trans. Electromagn. Compat.,
vol. 35, no. 4, pp. 423428, Nov. 1993.
[22] V. Erceg, A. J. Rustako, and R. S. Roman, Diffraction around corners
and its effects on the microcell coverage area in urban and suburban environments at 900 MHz, 2 GHz and 6 GHz, IEEE Trans. Veh. Technol.,
vol. 43, no. 3, pp. 762766, Aug. 1994.
[23] J. Lee and H. L. Bertoni, Coupling at cross, T and L junctions in tunnels, IEEE Trans. Antennas Propag., vol. 51, no. 5, pp. 926935, May
2003.
[24] K. Rizk, J. F. Wagen, J. Li, and F. Gardiol, Lamppost and panel scattering compared to building reflection and diffraction, in Proc. COST
259, May 1996.
[25] M. Ghoraishi, J. Takada, and T. Imai, Identification of scattering objects in microcell urban mobile propagation channel, IEEE Trans. Antennas Propag., vol. 54, no. 11, pp. 34733480, Nov. 2006.
[26] J. S. Lu, Simplified urban propagation models for rapid computation
of site-specific path gain, Masters thesis, Polytechnic Institute, New
York Univ., New York, NY, USA, May 2010, partial fulfillment of
MS/EE.
[27] F. Niu and H. L. Bertoni, Path loss and cell coverage of urban microcells in high-rise building environments, in Proc. IEEE Global
Telecommun. Conf. (GLOBECOM), Nov. 1993, vol. 1, pp. 266270.
[28] D. A. McNamara, C. W. I. Pistorius, and J. A. G. Malherbe, Introduction to the Uniform Geometrical Theory of Diffraction. Boston, MA,
USA: Artech House, 1990.
[29] C. A. Balanis, Advanced Engineering Electromagnetics, 2nd
ed. Hoboken, NJ, USA: Wiley, 2012.
[30] W. F. Young, K. A. Remley, C. Koepke, D. Cameil, and J. Healy,
Performance analysis of RF-Based electronic safety equipment on a
subway station and the Empire State Building, NIST, Boulder, CO,
USA, NIST Tech. Note 1792, Mar. 2013.
[31] W. F. Young, K. A. Remley, D. W. Matolak, Q. Zhang, C. L. Holloway,
C. Grosvenor, C. Gentile, G. Koepke, and Q. Wu, Measurements and
models for the wireless channel in a ground-based urban setting in two
public safety frequency bands, NIST, Boulder, CO, USA, NIST Tech.
Note 1557, Jan. 2011.
[32] H. M. El-Sallabi, G. Liang, H. L. Bertoni, I. T. Rekanos, and P.
Vainikainen, Influence of diffraction coefficient and corner shape on
ray prediction of power and delay spread in urban microcells, IEEE
Trans. Antennas Propag., vol. 50, no. 5, pp. 703712, May 2002.
[33] C. Chrysanthou, J. K. Breakall, K. L. Labowshki, S. G. Bilen, and W.
J. Glessner, A simplified analytical Urban Propagation Model (UPM)
for use in CJSMPT, in Proc. IEEE Military Commun. Conf., Oct.
2007.

2200

IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 62, NO. 4, APRIL 2014

Jonathan S. Lu (S10) received the B.S. and M.S.

degrees in electrical engineering from New York
University Polytechnic School of Engineering
(NYU), New York, NY, USA, in 2010. He is currently working towards the Ph.D. degree in electrical
engineering at the same university.
Since 2008, he has been a Research Fellow at
NYU Polytechnic. During this period, he has also
interned with CERDEC, U.S. Army from 2009 to
2011 and InterDigital Inc., from 2012 to 2013. He
has developed real-time site-specific performance
and planning tools for various wireless systems operating in rural and urban
environments. His research activities and interests are in millimeter wave
(5764 GHz) and UHF band propagation measurement and modeling, and
spectrum sensing for cognitive radio.
Mr. Lu received the Myron M. Rosenthal Award for Best M.S. Academic
Achievement in Electrical and Computer Engineering in 2010.

Henry L. Bertoni (M67SM79F87LF04)

received the B.S. degree in electrical engineering
from Northwestern University, Evanston, IL, USA,
in 1960 and the M.S. degree in electrical engineering
and the Ph.D. degree in electrophysics, both from the
Polytechnic Institute of Brooklyn (now New York
University Polytechnic School of Engineering),
Brooklyn, NY, USA, in 1962 and 1967, respectively.
He joined the faculty of the then Polytechnic
Institute of Brooklyn in 1966, and at various times
has been Head of the Electrical and Computer
Engineering Department from 1990 to 1995 and from 2001 to 2004, and
Vice Provost of Graduate Studies from 1995 to 1996. After retiring from
the regular faculty in 2005, he has continued to be active at Polytechnic as
an adjunct teacher and as Director of the NSF-sponsored Wireless Internet
Center for Advance Technology (WICAT). He has conducted research on
wave phenomena in electromagnetics, ultrasonics, acoustics, and optics. He
has authored or coauthored 90 journal papers and 10 book chapters on these
topics as well as the book Radio Propagation for Modern Wireless Systems
(Englewood Cliffs, NJ, USA: Prentice Hall PTR, 2000).
Dr. Bertoni received Best Paper Awards for five of his journal articles . His
later research, which dealt with prediction of UHF propagation characteristics
in urban environments for wireless application has been incorporated into commercially used software packages for designing cellular systems and earned him
the 2003 James R. Evans Avant Garde Award from the IEEE Vehicular Technology Society. He was the first Chairman of the Technical Committee on Personal Communications of the IEEE Communications Society and was IEEE
representative to, and Chairman of, the Hoover Medal Board of Award. From
1998 to 2001, he was a Distinguished Lecturer for the IEEE Antennas and Propagation Society.

Kate A. Remley (S92M99SM06F13) was

born in Ann Arbor, MI, USA. She received the Ph.D.
degree in electrical and computer engineering from
Oregon State University, Corvallis, OR, USA, in
1999.
From 1983 to 1992, she was a Broadcast Engineer
in Eugene, OR, serving as Chief Engineer of an
AM/FM broadcast station from 1989 to 1991. In
1999, she joined the Electromagnetics Division of
the National Institute of Standards and Technology
(NIST), Boulder, CO, USA, as an Electronics Engineer. Her research activities at NIST include metrology for wireless systems,
characterizing the link between nonlinear circuits and system performance, and
developing methods for improved radio communications for the public-safety
community.
Dr. Remley was the recipient of the Department of Commerce Bronze and
Silver Medals, an ARFTG Best Paper Award, and is a member of the Oregon
State University Academy of Distinguished Engineers. She was the Editor-inChief of IEEE MICROWAVE MAGAZINE from 2009 to 2011 and was the Chair of
the MTT-11 Technical Committee on Microwave Measurements from 2008 to
2010.

William F. Young (M06) received the M.S. degree

from Washington State University, Pullman, WA,
USA, in 1998 and the Ph.D. degree from the University of Colorado, Boulder, CO, USA, in 2006, both
in electrical engineering.
He worked at Sandia National Laboratories from
1998 to 2010, and has collaborated with the National
Institute of Standards and Technology (NIST) on
wireless systems and measurements since 2003. In
2010, he joined the Electromagnetics Division at
NIST, Boulder, CO, USA. His experience in wireless
communication systems includes diversity antenna design, radio frequency
propagation measurements, MIMO system applications, electromagnetic
interference testing, and wireless network security analysis. He is currently
developing radio frequency laboratory measurement techniques that support
standards specification efforts by the National Fire Protection Association
(NFPA) for Personal Alert Safety Systems (PASS), and by the CTIA-The
Wireless Association for Long Term Evolution (LTE) MIMO technology
evaluation.

John Ladbury (M09) received the B.S.E.E. and

M.S.E.E. degrees (specializing in signal processing)
from the University of Colorado, Boulder, CO, USA,
in 1987 and 1992, respectively.
Since 1987, he has worked on EMC metrology
and facilities with the Electromagnetics Division,
RF Fields Group of NIST, Boulder, CO, USA. His
principal focus has been on reverberation chambers,
with some investigations into other EMC related
topics such as time-domain measurements and probe
calibrations.
Mr. Ladbury is a member of the International Electrotechnical Commission
(IEC) joint task force on reverberation chambers. He has received five Best
Paper Awards at various conferences and international symposia over the last
six years.