IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 15, NO.

1, FIRST QUARTER 2013 255

A Survey of Wireless Path Loss Prediction and

Coverage Mapping Methods
Caleb Phillips, Student Member, IEEE, Douglas Sicker, Member, IEEE, and Dirk Grunwald, Member, IEEE

Abstract—In this paper we provide a thorough and up to date 1) Theoretical/Foundational Models (§III-A)
survey of path loss prediction methods, spanning more than 2) Basic Models (§III-B)
60 years of fairly continuous research. These methods take a 3) Terrain Models (§III-C)
variety of approaches to modeling the signal attenuation between
wireless transceivers: purely theoretical models, empirically fitted 4) Supplementary Models (§III-D)
(often statistical) models, deterministic ray-optical models, and 5) Stochastic Fading Models (§III-E)
measurement-directed methods. Our work here extends and 6) Many-Ray Models (§III-F)
updates excellent, but now dated prior surveys of this important 7) Active Measurement Models (§IV)
field. We provide a new taxonomy for reasoning about the
similarities and differences of the many approaches and provide Our work here is exhaustive, including more than 50 proposed
a brief but complete overview of the various methods as well models from the last 60 years, 30 of which we describe in
as describing insights into future directions for research in this detail. We describe models at a high level and focus on briefly
area. identifying their chief differences from other models. Figure
Index Terms—Path loss, prediction, wireless, attenuation, 1 provides a family tree of the majority of path loss models
model, networking, signal, ray-tracing, measurement discussed in the following subsections and may prove useful
for understanding the lineage of various proposals as well as
I. I NTRODUCTION their functional relationship to one another.
In the next section, we will give a brief tutorial on radio

T ODAY, wireless networks are absolutely ubiquitous and

the importance of their role in our daily lives cannot be
underestimated. To a large extent, our ability to build and un-
propagation. In section III, we will discuss the bulk of models,
which make their predictions a priori, without insight from
measurements. In section IV, we discuss models and methods
derstand these networks hinges on understanding how wireless that do use (possibly directed) measurements to inform their
signals are attenuated over distance in realistic environments. predictions. In section V we we will survey comparative
By predicting the attenuation of a radio signal, we can evaluations of the performance of these models, and in section
better plan and diagnose networks as well as build futuristic VI, we will provide a summary and conclusions with respect
networks that adapt to the spatiotemporal radio environment. to areas of future work and promising next steps.
For instance, today’s network engineers need methods for
accurately mapping the extent of coverage of existing and
II. R ADIO P ROPAGATION BASICS
planned networks, yet the efficacy of those approaches is
determined by the predictive power of the underlying path This section introduces the basic concepts of radio prop-
loss model (or interpolation regime). Similarly, researchers agation. For a more thorough treatment, the intrepid reader
that investigate dynamic spectrum access networks require can refer to any number of textbooks, including the excellent
accurate radio environment maps to make appropriate and surveys by Rappaport [71] and Seybold [79].
timely frequency allocation decisions, yet the performance
of these systems is tied intimately to their ability to make A. Signal Propagation
meaningful predictions about the current and future occupancy
When asked to describe radio, Albert Einstein famously
of the radio channel.
responded:
Since the 1940’s, researchers and engineers have pondered
this problem and have developed myriad schemes that purport You see, wire telegraph is a kind of a very, very
to predict the value or distribution of signal attenuation (path long cat. You pull his tail in New York and his
loss) in many different environments and at different frequen- head is meowing in Los Angeles. Do you understand
cies. In this work, we attempt to give a complete treatment of this? And radio operates exactly the same way: you
work to date, updating and extending a series of excellent but send signals here, they receive them there. The only
dated surveys from the last 15 years (e.g., [48], [7], [71], [79], difference is that there is no cat.
[64]). We propose a new taxonomy for path loss models that The study of radio propagation is largely concerned with what
groups all proposals into seven major categories and fourteen happens in between the head and the tail of the, well, no cat. At
subcategories. The seven major categories are: each end of the radio link, there is a transceiver that is attached
to an antenna of some geometry. The transmitter produces a
Manuscript received 31 October 2011; revised 8 February 2012. signal (an electromagnetic plane wave) that is modulated onto
The authors are with the Department of Computer Science, University of
Colorado, Boulder. (e-mail: [email protected]). the carrier frequency. On its way to the receiver (at roughly the
Digital Object Identifier 10.1109/SURV.2012.022412.00172 speed of light), the signal reacts with any number of obstacles
1553-877X/13/$31.00 
c 2013 IEEE
Authorized licensed use limited to: Telecom ParisTech. Downloaded on April 15,2024 at 16:38:27 UTC from IEEE Xplore. Restrictions apply.
256 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 15, NO. 1, FIRST QUARTER 2013

Fig. 1. Path loss model family tree. Individual models are shown as circles and categories as are shown as rectangles. Major categories are green. Minor
categories are blue.

and then is induced on the receiver’s antenna and demodulated. and when it is caused by small transient obstacles, and varies
Obstacles in the environment cause the signal to be reflected, with time, it is called scattering, fast fading, or small scale
refracted, or diffracted, which attenuate the power of the fading.
signal (through absorption) and cause scattering and secondary When the signals interact with the environment, they can
waves. Obstacles that are near the line of sight (LOS) path are be delayed by reflections, or frequency-shifted by diffractions.
said to obstruct the Fresnel zone (technically, the first Fresnel Mobile transceivers also incur frequency shift due to Doppler
zone’s circular aperture) and are most problematic. spreading. Frequency shifts and delay spread both contribute
Actually, things are a bit more complicated than this. to small scale fading.
Because an antenna radiates its signal simultaneously in all
directions, the signal can take many paths to the receiver. B. Path Loss
Each path may interact with the environment in a chaotically The geometry of the antennas that the transmitter and
different way and arrive at the receiver delayed by some receiver use emphasize signals arriving from some directions
amount. If these delayed signals are in phase with one another, over others. An omnidirectional antenna emphasizes signals
then they produce constructive interference. If they are out of in the azimuthal plane and de-emphasizes signals arriving
phase with one another, they produce destructive interference. from above or below. As a result, the gain pattern tends to
The spread of this delay is called the delay spread and be shaped like a donut, as can been seen in figure 3. A
the resulting attenuation is called multipath fading. When directional antenna, such as a patch panel, parabolic dish,
this attenuation is caused by large unmoving obstacles, it is or sector, typically emphasizes signals arriving from a single
referred to as shadowing, slow-fading, or large-scale fading direction (lobe) within some beamwidth. The gain pattern of
Authorized licensed use limited to: Telecom ParisTech. Downloaded on April 15,2024 at 16:38:27 UTC from IEEE Xplore. Restrictions apply.
PHILLIPS et al.: A SURVEY OF WIRELESS PATH LOSS PREDICTION AND COVERAGE MAPPING METHODS 257

Fig. 2. Horizontal and vertical radiation patterns for a (highly directional) 24 dBi parabolic dish antenna. Image taken from L-COM antenna specifications
[54].

Fig. 3. Horizontal and vertical radiation patterns for 7 dBi colinear omnidirectional antenna. Image taken from L-COM antenna specifications [54].

these antennas more closely resembles a baseball bat, as can set of interferers, the Signal to Interference and Noise Ratio
be seen in figure 2. However, perfect isolation is impossible (SINR) is defined as:
and geometries that emphasize a single direction also have ⎛ ⎞
substantial gain in other directions (side lobes and back lobes) n

as a result. Antenna gain is typically measured in dBi, which SIN R = Prx − ⎝N + Ij ⎠ (2)
j
is decibels relative to an isotropic transmitter (an isotropic
transmitter’s gain pattern is a sphere). For a given receiver design and modulation scheme, there is
If the transmitter’s radio has a transmit power of Ptx Watts a known relationship between SNR and bit error rate. Using
(W) and an antenna gain of Gt dBi, then the total Effective this relationship, we can determine the minimum detectable
Isotropic Radiated Power (EIRP) is Ptx ∗ Gtx . In the log signal for a given radio as a function of the acceptable error
domain, Ptx is given in dBm, which is decibels relative to rate: M DS(Pe ), where Pe is the probability of bit error. Then,
a mW, and the EIRP is simply Ptx + Gtx . The entire radio determining the points that are covered is simply the set of
link can then be summarized by the common log-domain link receiver locations that satisfy the inequality:
budget equation:
Ptx + Gtx + Grx − P L ≥ M DS(Pe ) (3)
Prx = Ptx + Gtx + Grx − P L (1) Because the P and G terms are known for a given link, the
with Prx and Grx being the power received at the receiver and difficulty becomes predicting the quantity P L given what we
the receiver’s antenna gain in the direction of the transmitter. know about the environment and the radio link. Or, for the
Here, the P L term includes all attenuation due to path loss. case of measurement-based approaches, interpolating the P L
This formula describes the aggregate gain and attenuation of value for the points we have not measured.
many competing signals. It also assumes that our radio link is As it is defined here, a model’s task is to predict the value
isolated from any sources of external noise in the environment of Lt + Ls in this log-domain equation:
(i.e., thermal noise and interference from other transmitters).
Commonly, the signal quality at a given point is written as P L = Lt + Ls + Lf (t) (4)
the ratio between Signal and Noise: SN R = Prx − N (in the where Lt is the trivial free-space path loss, Ls is the loss due
log domain). Alternately, including interference from a known to shadowing (slow fading) from large unmoving obstacles
Authorized licensed use limited to: Telecom ParisTech. Downloaded on April 15,2024 at 16:38:27 UTC from IEEE Xplore. Restrictions apply.
258 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 15, NO. 1, FIRST QUARTER 2013

like mountains and buildings, and Lf (t) is the small-scale fast Or, more commonly, we solve for the power at the receiver in
fading due to destructive interference from multipath effects terms of the power from the transmitter and the path loss:
and small scatterers (which varies with time t). Small-scale  2
fading is often both time and frequency selective, meaning λ
Prx = Ptx (7)
that it varies with time and frequency. Models cannot, without 4πd
perfect knowledge of the environment, be expected to predict Converting equation 7 to take distance in km instead of m,
the quantity Lf (t). In most applications, this additional error frequency in MHz instead of wavelength in m, and converting
is computed “stochastically” using a probability distribution the linear domain power units (W) to log domain units (dBm),
(often Raleigh, although Ricean and m-Nakagami are pop- gives the commonly seen reference equation for path loss as
ular). In this way, frequency and time selective fades can a function of carrier frequency and distance:
be simulated, if not predicted exactly, which allows for the
analysis of their effect on modulation schemes (e.g., [38],
[84]). In the following sections, we will discuss the many Prx = Ptx − (20log10 (d) + 20log10 (f ) + 32.45) (8)
methods proposed for predicting the value of Lt + Ls and the
Where power in decibels relative to a milliwatt (dBm) can be
distribution of Lf (t).
obtained from power in Watts (W) using this conversion:

III. M ODELING PATH L OSS A Priori PdBm = 10log10 (PmW ) (9)

[t]
2) Flexible Path Loss Exponent: Friis’ equation assumes
The models discussed in this section are a priori, meaning
that signal degrades as a function of d2 , a common extension
they make predictions using only available prior knowledge
to non-line-of-sight (NLOS) environments is to use a larger
and do not use explicit measurements in their predictions.
exponent. To allow for this, we simply substitute in α, which
Hence, these models are most appropriate for making predic-
can be set to any value greater than zero, but is most
tions in situations where it is impossible or difficult to obtain
commonly set to 2:
measurements. We subdivide these models into six categories:
1) Theoretical/Foundational Models (§III-A)
2) Basic Models (§III-B) Prx = Ptx − (10αlog10 (d) + 20log10 (f ) + 32.45) (10)
3) Terrain Models (§III-C) Often, this model will be given relative to some reference
4) Supplementary Models (§III-D) distance d0 (commonly 100m), where the assumption is that
5) Stochastic Fading Models (§III-E) several measurements are made at this distance, and those
6) Many-Ray Models (§III-F) values are used to fit a slope:
Each category and its respective subcategories are discussed
in turn in the following subsections. Table I provides a
chronological list of the models discussed here and provides Prx = Ptx − (10αlog10 (d/d0 ) + 20log10 (f ) + 32.45) (11)
their major category, coverage, and initial publication. 3) Ground Reflection: As a modest extension to the free-
space path loss model, the Two-Ray Ground Reflection model
A. Theoretical/Foundational Models considers a second path that reflects from the ground between
the transmitter and receiver [71], [79], [68]. First, we calculate
The first models worth considering are purely analytical
the break distance:
models derived from the theory of idealized electromagnetic
propagation. Although these models are questionably accurate,
dc = (4πhtx hrx )/λ (12)
they are simple to understand and implement and as a result
they have been widely adopted into network simulators and where htx and hrx are the heights of the transmitter and
other applications and often serve to compute a minimum loss receiver antennas respectively (in m). For distances shorter
for other, more complex, models. than this break distance, we simply use Friis’ equation as
1) Freespace Between Isotropic Antennas: In [32], Friis the receiver is not far enough away to receive a substantial
proposed a basic formula for free-space transmission loss: ground reflected path loss. For distances longer than the break
distance, we use the modified path loss formula:
Prx Arx Atx
= (5)
Ptx d2 λ2 Ptx h2tx h2rx
Pr = (13)
This formula describes the ratio between received power (Prx ) d4
and transmitted power (Ptx ) in terms of the effective area of In [66], Oda et al. propose a minor extension to this model
the transmitting antenna (Atx ), receiving antenna (Arx ), the where the plane of reflection is considered to be above the
distance between (d) in meters, and the wavelength of the nominal ground clutter, and a factor for probability of collision
carrier (λ) in meters. For ideal isotropic antennas, this formula per unit distance is considered. By adjusting this height
can be simplified to: offset (h0 ), the reflectivity coefficient (R) and this negative
 2 exponentially weighted factor, one can coax the simple Two-
Prx λ Ray model into more closely fitting some types of measured
= (6)
Ptx 4πd data.
Authorized licensed use limited to: Telecom ParisTech. Downloaded on April 15,2024 at 16:38:27 UTC from IEEE Xplore. Restrictions apply.
PHILLIPS et al.: A SURVEY OF WIRELESS PATH LOSS PREDICTION AND COVERAGE MAPPING METHODS 259

TABLE I
A priori MODELS STUDIED ALONG WITH THEIR CATEGORIZATION , REQUIRED INPUT, COVERAGE REMARKS , RELEVANT CITATIONS , AND YEAR OF
( INITIAL ) PUBLICATION .

Name Short Name Category Coverage Notes Citations Year

Friis’ Freespace friis Foundational d > 2a2 /λ [32] 1946
Egli egli Basic 30M Hz < f < 3GHz [27], [79] 1957
Hata-Okumura hata Basic 1km < d < 10km; 150 <= f <= 1500M Hz [67] 1968
30 <= h1 <= 200m; 1 <= h1 <= 20
Edwards-Durkin edwards Basic/Terrain [26], [21] 1969
Allsebrook-Parsons allsebrook Basic/Terrain f ∈ 85, 167, 441M Hz; Urban [3], [21] 1977
Blomquist-Ladell blomquist Basic/Terrain [8], [21] 1977
Longley-Rice Irregular itm Terrain 1km < d < 2000km [42], [43] 1982
Terrain Model (ITM) 20M Hz < f < 20GHz
Walfisch-Bertoni bertoni Basic [90] 1988
Flat-Edge flatedge Basic [77] 1991
TM90 tm90 Basic d <= 10miles; h1 <= 300f eet [17] 1991
COST-231 cost231 Basic 1km < d < 20km; [12] 1993
Walfisch-Ikegami walfish Basic 200m < d < 5km; 800M Hz < f < 2GHz; [12], [65], [7] 1993
4m < hb < 50m; 1m < hm < 3m
Two-Ray (Ground Reflection) two.ray Foundational [71], [79], [68] 1994
Hata-Davidson davidson Basic 1km < d < 300km; 150M Hz < f < 1.5GHz; [9], [65] 1997
30m < hb < 1500m; 1m < hm < 20m
Oda oda Basic [66] 1997
Erceg-Greenstein erceg Basic f ≈ 1.9GHz; Suburban [28] 1998
Directional Gain Reduction grf Supplementary Dir. Recv. Ant.,f ≈ 1.9GHz [36] 1999
Factor (GRF)
Rural Hata rural.hata Basic f ∈ 160, 450, 900M Hz; Rural (Lithuania) [62] 2000
ITU Terrain itu Terrain [79], [48] 2001
Stanford University sui Basic 2.5 < f < 2.7GHz [29], [2] 2001
Interium (SUI)
Green-Obaidat green Basic [35] 2002
ITU-R itur Basic 1km < d < 10km; 1.5GHz < f < 2GHz; [48], [65] 2002
30m < hb < 200m; 1m < hm < 10m
ECC-33 ecc33 Basic 1km < d < 10km; 700 <= f <= 3000M Hz [24], [2] 2003
20 <= h1 <= 200m; 5 <= h1 <= 10
Riback-Medbo fc Supplementary 460M Hz < f < 5.1GHz [73] 2006
ITU-R 452 itur452 Terrain [50] 2007
IMT-2000 imt2000 Basic Urban [33] 2007
deSouza desouza Basic f ≈ 2.4GHz; d < 120 m [20] 2008
Effective Directivity edam Supplementary Directional Antennas; f ≈ 2.4GHz [5] 2009
Antenna Model (EDAM)
Herring Air-to-Ground herring.atg Basic f ≈ 2.4GHz [39] 2010
Herring Ground-to-Ground herring.gtg Basic f ≈ 2.4GHz [39] 2010

models use one or more random variables to account for

channel variation (and hence, can predict a distribution instead
of a median value).
1) Egli: The Egli Model [27] is an early empirical model
based on measurements made in New York City and parts
of New Jersey by John Egli. The simplified version, based
on extracting a model from numerous graphs and nomograms
[21] and [7] is:

Prx = Ptx −20log10(f )+40log10 (d)−20log10 (htx )+k (14)

Fig. 4. Schematic of link geometry used by basic models.
with
76.3 − 10log10 (hrx ) hrx ≤ 10
k= (15)
85.9 − 20log10 (hrx ) hrx > 10
B. Basic Models
2) Green-Obaidat: The Green-Obaidat model suggested in
Basic models are the most numerous of the model types. [35] is a small modification to free-space path loss that adjusts
They compute path loss along a single path and often use for the relative heights of the transmitter and receiver and uses
corrections based on measurements made in one or more envi- a path loss exponent of α = 4:
ronments. In general, they use the distance, carrier frequency,
and transmitter and receiver heights as input. Figure 4 gives
a schematic of the basic link geometry. Some models also Prx = Ptx − (40log10 (d) + 20log10 (f ) − 20log10(htx hrx ))
have their own esoteric parameters to select between different (16)
modes of computation or fine tuning. Here we subdivide In this model, d is given in km, htx and hrx in m, and f in
these models into deterministic and stochastic. The stochastic MHz. The authors perform minimal validation using 802.11
Authorized licensed use limited to: Telecom ParisTech. Downloaded on April 15,2024 at 16:38:27 UTC from IEEE Xplore. Restrictions apply.
260 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 15, NO. 1, FIRST QUARTER 2013

TABLE II
C OMMONLY USED MATHEMATICAL SYMBOLS .

d distance between transmitter and receiver along line of sight path in km

dm distance between transmitter and receiver along line of sight path in m (1000d)
htx /hrx height of transmitter/receiver above ground in m
Ptx /Prx power produce by transmitter/received by receiver
f carrier frequency in MHz
λ carrier wavelength in m
gtx /grx gain of the transmitters/receiver’s antenna in the azimuthal direction of the transmitter
θ angle from transmitter to receiver in azimuthal plane relative to true north
θ angle from receiver to transmitter ...
φ angle of elevation between transmitter and receiver relative to horizontal (inclination)
φ angle from receiver to transmitter ...
U (a, b) a uniformly distributed random variable between a and b (inclusive)
N (μ, σ) a normally distributed random variable of mean μ and standard deviation σ
LN (μ, σ) a lognormally distributed random variable of mean μ and standard deviation σ
R the radius of the earth in m (≈ 6.371 ∗ 106 )
C the speed of light in m/s (≈ 299.792 ∗ 106 )
r relative permittivity (of obstructing material)

devices operating at 2.4 GHz. This model is one of several that The authors claim very impressive performance at the sites
extends basic models to include the relative height of nodes (2 indoor, 2 outdoor) they study (from which the fitted β
in their calculations—in this case, the heights are multiplied. parameters are derived). The short distances studied (< 120 m)
3) Edwards-Durkin: The Edwards-Durkin model [26] sim- suggest that this model may be inappropriate for modeling
ply sums classical free-space loss (lf ) with an additional lengthier links.
correction due to plane earth propagation loss from Bullington 7) TM90: In [17], the authors propose a propagation model
[10]: intended for suburban areas and for propagation distances
less than 10 miles. This model is very simple, using a
flexible path loss exponent model with α = 4, accounting for
lp = 118.7 − 20log10(hrx) − (17)
antenna heights as in the Hata-Okumura model, and adding
20log10(htx) + 40log10(d) an additional loss for average building penetration (outdoor-
P L = lf + lp (18) indoor interface loss). This model is the Federal Communi-
cations Commission (FCC) recommended model for shorter
The constants in this formula are fitted from empirical
propagation distances (as opposed to the Irregular Terrain
measurements made in the United Kingdom by Durkin [25].
Model (ITM), which is most appropriate for long links).
In [21], Delisle updates this model with a statistical terrain
8) Hata-Okumura: The Hata-Okumura model is an em-
diffraction loss estimate (ld(Δh), described in section III-D2b)
pirical model where measurements made by Okumura in
and leaving out the freespace term:
and around Tokyo, Japan are approximated with equations
proposed by Hata [7], [67]. The model is considered valid for
P L = lp + ld(Δh) (19)
frequencies from 150 MHz to 1500 MHz and for transmitter
4) Blomquist-Ladell: The Blomquist-Ladell model [8] is heights between 30 m and 200 m and receiver heights between
similar in construction to the Edwards-Durkin model. It com- 1 m and 10 m and distances greater than 1 km. The model
putes an excess plane earth loss, with a correction factor, and takes an additional environment parameter that can be one of
sums it with classical free-space loss. As with the Edwards- “open”, “suburban”, “urban medium”, or “urban large”, which
Durkin model, it can be extended with a statistical terrain selects among different modes of computation for differing
diffraction loss estimate. The fitted constants in this model levels of environment complexity (as related to population
were derived from measurements in the VHF and UHF bands density).
over rolling terrian in Sweden. Due to the popularity of the Hata-Okumura model, there
5) Allsebrook-Parsons: The Allsebrook-Parsons model [3] have been numerous extensions and corrections:
is an extension to the Blomquist-Laddell model that adds • COST-Hata/Extended Hata: an extension to cover fre-
an additional loss due to buildings. The authors based the quencies up to 2000 MHz. It was proposed as part of the
empirical adjustment on measurements taken in British cities. COST-231 [7], [33], [12].
The model also suggests a constant additional loss (named γ • Hata-Davidson: an extension to provide corrections for
here) of 13 dB for frequencies above 200 MHz1 . long links (up to 300 km) and high transmitters (up
6) deSouza-Lins: In [20], de Souza and Lins present an to 2500 m). It was proposed in Telecommunications
entirely empirical model explicitly fitted to data collected at Industry Association (TIA) recommendation TSB-88-B
2.4 GHz. This model is a function of distance (in meters) and [85], [65].
relative humidity percentage (h): • ECC-33: an extension to cover frequencies up to 3500
MHz, proposed by the Electronic Communication Com-
Prx = Ptx − (β0 + β1 log10 (d) + β2 d + β3 log10 (h)) (20) mittee (ECC) within the European Conference of Postal
and Telecommunications Administrations (CEPT) [24].
1 The validity of this correction is questioned in [21]. • ITU-R/CCIR: a modification which takes a real-valued
Authorized licensed use limited to: Telecom ParisTech. Downloaded on April 15,2024 at 16:38:27 UTC from IEEE Xplore. Restrictions apply.
PHILLIPS et al.: A SURVEY OF WIRELESS PATH LOSS PREDICTION AND COVERAGE MAPPING METHODS 261

If, however, this is not the case, then a complicated series

of Fresnel calculations are required to compute ln. Those
equations are well summarized in [77] Appendix B.3. The
additional loss due to diffraction over the final obstruction is
calculated using the method of Ikegami [44].
10) Walfisch-Bertoni: The Walfisch-Bertoni model is the
limiting case of the Flat-Edge model when the number of
buildings is large enough for the field to settle [7]. Hence, this
model takes as parameters the distance between obstructions
and their nominal size, but not the number of them, which is
implicit to the calculation.
Fig. 5. Schematic of link geometry used by the Flat-Edge family of basic 11) Walfisch-Ikegami: The COST231/Walfisch-Ikegami is
models.
a compromise proposal by the COST-231 that combines the
Walfisch-Bertoni model with an additional reflection down
parameter, the percentage of area covered by buildings to the receiver using the Ikegami model [44] along with
(bp), instead of a discrete environment class. This model some empirical corrections from measurements [7]. The model
is an attempt at correcting systematic underestimations distinguishes between LOS propagation and non line-of-sight
observed in the Hata-Okumura model and is in essence (NLOS) and uses different calculations for each. In addition to
the Hata-Okumura model for “urban-medium” environ- the expected parameters describing the geometry of the line-
ments with an additional correction factor related to the of-sight path, this model requires specification of the constant
new parameter. It was proposed by the International Ra- building height, street width (w), distance between buildings
dio Consultive Committee (CCIR) (now the International (b, such that b − w is the nominal building width), the angle
Telecommunications Union Radio communication Sector of the incident wave to the street (π radians for vertically
(ITU-R)) [65]. polarized antennas, 0 for horizontal), and the building size
• Rural Hata: a correction for the classic Hata-Okumura (either “medium” or “large”).
model as defined in ITU-R 529 to correct for systematic For Non Line-of-Sight (NLOS) links, the model includes
overestimations of path loss in rural terrain. The model calculations for excess loss above free-space loss due to
proposes new fitted values for path loss exponent and roof-to-street diffraction loss and multiscreen diffraction
fixed offset to replace those that are defined in the default loss. After calculating this excess loss, if it is positive, it is
model. These fits are obtained from data collected using summed with the free-space loss and used. Otherwise, vanilla
a simple random sampling scheme in rural Lithuania at free-space loss is returned. For LOS links, the returned value
three frequencies below 900 MHz. In addition to this, the is free-space loss with a fudge factor to attempt to avoid
authors propose a method to do site-specific fitting in a underestimates: 6 ∗ log10 (50 ∗ d). Because this model is
similar way so that their approach can be used in other reasonably complicated, we refer the interested reader to
environments and at other frequencies [62]. the excellent slides maintained by the National Institute of
9) Flat-Edge: The Flat-Edge model, proposed by Saunders Standards and Technology (NIST) [65] for further details.
and Bonar [77] takes a very different approach as compared
to the Hata family of models. Saunders and Bonar propose a The remaining basic models include a random variate
model that computes approximate knife-edge diffraction losses (stochastic) term that attempts to capture the time-varying
due to multiple obstructions (buildings) that are regularly nature of the wireless channel due to small scale fading.
spaced. Figure 5 provides a schematic of this setup. The model 12) Herring: The Herring model is a recent proposal by
takes as parameters the number of obstructions between the Herring et al. [39]. The model proposes two distinct models,
transmitter and receiver (n), the constant distance between one for air-to-ground (ATG) communications and one for
them (w), and their constant height (h0 ). The assumption is ground-to-ground communications (GTG), both of which are
that there is a transmitter either above or below a series of based on fits to data collected by the authors at 2.4 GHz in
obstacles of the constant size and spacing and the receiver is Cambridge, Massachusetts. The ATG model is a simple error
below the top of the buildings. The model works by summing term on top of the free-space path loss model:
the loss due to Fresnel obstruction by the obstacles, the basic
free-space propagation loss (lf ), computed using equation 8, Prx = Ptx − (lf + N (30, 8.3)) (23)
and the loss due to diffraction over the final obstruction. where lf is calculated as in equation 8 and N (30, 8.3) is a
First, the value of t is calculated: random Gaussian with mean 30 and standard deviation of 8.3.
The GTG model is slightly more complex. It first computes
πw
t=φ (21) a random Gaussian path loss exponent with uniform random
λ offset:
If 1 ≤ n ≤ 100 and −1 ≤ t < 0, then the approximate fit
due to Barclay [7] is used: α = U (2, 5) + N (0, 0.22) (24)
This path loss exponent is then used along with a larger excess
ln = −(3.29 + 9.90log10(t) − (0.77 + 0.26log10(n))) (22) loss value:
Authorized licensed use limited to: Telecom ParisTech. Downloaded on April 15,2024 at 16:38:27 UTC from IEEE Xplore. Restrictions apply.
262 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 15, NO. 1, FIRST QUARTER 2013

Prx = Ptx − (lf (α) + N (40, 5.5)) (25)

where lf (α) is computed as in equation 10.
13) Erceg-Greenstein: In [28], Erceg et al. present a mea-
surement based model for path loss around 1.9 GHz using a
large data set collected by AT&T in suburban locations around
New Jersey. The model is a fitted model that combines a fit for
median path loss at some distance d and a randomly distributed
variation:

Fig. 6. Schematic of link geometry used by terrain models.

P L = A + 10(a − b ∗ htx + (26)
c d d
( )log10( ) + x10log10( ) +
hrx d0 d0 1) ITU Terrain: The ITU terrain model is a simple model
yμσ + yzσσ that combines free-space path loss with a single diffraction
due to terrain [79], [48]. A Digital Elevation Model (DEM)
where the values of a, b, c, σγ , μσ , and σσ are fitted parameters
is used to compute the loss due to the most significant path
for each of the three terrain categories: hilly with moderate to
heavy tree density (A), hilly with light tree density or flat obstruction in terms of Fresnel zone blockage. In the event
that the transmission path has no blockage, then free-space
with moderate to heavy tree density (B), or flat with light
path loss (with an exponent of 2) is used. The radius of the
tree density (C). The value A is the trivial free space path
loss (from equation 8, for instance) at some reference distance first Fresnel zone is computed at the site of blockage:
(d0 , usually 100 m). And, x, y, and z are normally distributed
d1 d2
random variables between -2 and 2 (x is between -1.5 and f1 = 17.3 (30)
1.5). fd
14) IMT-2000: Pedestrian Environment: Three path loss where d1 is the distance from the transmitter to the obstruction,
models for IMT-2000/3G are provided in [33], one for the d2 is the distance from the receiver to the obstruction, d is the
indoor office environment, one for the outdoor to indoor and total distance, and f is the carrier frequency. The constant,
pedestrian environment, and one for the vehicular environ- 17.3, is derived from equations governing the physical optics
ment. It is the pedestrian model which we describe here, which of Fresnel lenses. Then the additional path loss (outside of
is simply equation 10 with α = 4, a constant (optional) offset free-space) is used for this blockage:
for building penetration loss (k1 ) and a lognormally distributed
offset to account for shadowing loss (k2 ): al = −20.0h/f1 + 10.0 (31)
The model suggests that a negative loss due to the blockage
Prx = Ptx − (40log10(d) + 30log10(f ) + k1 + k2 + 21) (27) (which is actually a gain, i.e., negative loss) or any loss less
than 6 dB should be discarded. The maximum additional loss
with
is then used to “correct” the free-space loss assumption.
18 indoors 2) ITU-R 452: The clear-air interference prediction algo-
k1 = (28) rithm described in ITU-R 452 [50] serves a fine example of
0 o.w.
the state of the art in terrain path loss models. This model
and makes a prediction of median path loss based on the sum of
free-space path loss with several corrections:
k2 = LN (0, 10) = e0+10N (0,1) (29) 1) losses from knife-edge diffractions over terrain obstacles
where LN (0, 10) is a lognormally distributed random variable 2) losses from absorption due to atmospheric gases (water
with zero mean and a standard deviation of 10. vapor)
3) losses from tropospheric scatter, ducting, coupling, and
layer reflection in the atmosphere
C. Terrain Models 4) losses due to obstruction from the curvature of the earth
Terrain models are similar to the basic models, but also 5) additional clutter losses derived from land cover classi-
attempt to compute diffraction losses along the line-of-sight fication [47] near the transmitter and receiver
path due to obstructions (terrain or buildings, for instance). The model computes the path loss in terms of a confidence
See figure 6 for a schematic. They are an order of magnitude value p, which is the not-to-exceed probability. Using p = 50
more complex, but are immensely popular especially for long computes a median value, p = 100 computes a worst-case
propagation distances at high power in the VHF band (i.e., value and p = 0 computes a best-case value.
television transmitters). Because of the relative complexity of In addition to p, the model takes a handful of other
these models, the reader will need to refer to the citations for parameters: Δn, which is the radio refractivity of the earth
details of the implementations. Here, we have summarized between the transmitter and receiver (values between 35 and
their functionality at a high level. 60 are typical for the environments we are concerned with),
Authorized licensed use limited to: Telecom ParisTech. Downloaded on April 15,2024 at 16:38:27 UTC from IEEE Xplore. Restrictions apply.
PHILLIPS et al.: A SURVEY OF WIRELESS PATH LOSS PREDICTION AND COVERAGE MAPPING METHODS 263

n0 which is the surface level refractivity and ω, which is the a significant amount of data in a limited number of (suburban)
fraction of the path over water (i.e., for intercontinental links. environments, from which the empirical constants are derived.
ω = 0 for all our environments). This model is leaps and 2) Obstructions: Obstruction models account for losses due
bounds more complex than those presented above, requiring a to obstructions along the main (or some secondary) path.
tremendous number of calculations often based on numerical They are the most numerous and varying of the supplementary
approximations (i.e., knife-edge diffraction). models:
ITU-R 452 suggests additional extensions for modeling the a) Atmospheric Gases: The effects due to absorption by
interference due to hydrometers such as rain and weather atmospheric gases are minimal at UHF frequencies and totally
cells. This adds substantial complexity to the algorithm with negligible at higher frequencies. However, it is worth noting
negligible benefit for many communications applications op- that such corrective models are available for water vapor and
erating in the upper end of the VHF band. Indeed, many of to a lesser extent for other gases (e.g., [51]).
the parameters computed by the ITU-452 model are negligible b) Statistical Terrain Diffraction Estimate: Because ter-
for commonly used communications frequencies (for instance, rain information is not always available and computing indi-
absorption due to atmospheric gases). In [92], Whitteker vidual diffractions over terrain can be computationally costly,
suggests a similar model which shares many of the attributes [21] proposes a method for computing an estimate of addi-
of ITU-R 452, with slightly less complexity. tional losses due to terrain. In addition to the geometry of
3) Longley-Rice Irregular Terrain Model: The ITM [42], the line-of-sight path, this approach makes use of a single
[43] may be the most widely known general purpose path parameter, Δh, which describes the “roughness” of the terrain.
loss model and is used in a number of popular network A value of ≈ 15 is considered minimal, ≈ 200 is used for hilly
planning tools (e.g., [60], [15]). This model was developed by terrain, and ≈ 400 for very rugged terrain. In [21], Delisle et
the NTIA in order to plan deployments of VHF transmitters al. propose the use of this estimate in combination with other
(i.e., broadcast television in the US). Hence, much like the models, such as Allsebrook-Parsons, Blomquist-Ladell, and
ITU-R model it is designed for very long transmission at Edwards-Durkin. In this way, it can be used to retrofit any
high power from well-positioned transmitters. For this reason, basic model with corrections for losses from terrain obstacles
its applicability to modeling propagation in, e.g., urban mi- and clutter.
crocells, is questionable at best. Much like ITU-R 452, the c) Building-Transmission: The Building-Transmission
ITM computes excess loss from free-space by considering model proposed by de Jong et al. in [18] attempts to model
knife-edge diffractions at terrain obstacles, losses due to the the loss due to transmission through a building in an urban
curvature of the earth, and tropospheric scatter. The principle environment. The authors attempt to isolate this effect from
difference is that ITU-R 452 includes some calculation for fades along other paths and instead present a statistical model
local clutter losses based on land cover classification data, for just the loss encountered by transmission through a number
otherwise the models can be thought to be quite similar. of representative buildings at 1.9 GHz. They find that on
average there is a loss of approximately 2.1 dB/m at this
frequency and use this to develop an algorithm to compute
D. Supplementary Models
total transmission loss, including refraction at the exterior
The next category of models are supplementary models, walls. For this model to be of use in practice, one must know
which cannot stand on their own, but are instead used to make the positions and shape of buildings along with the permittivity
corrections to existing (complete) models. Here we subdivide and conductivity of the buildings’ outer surfaces.
the models by the phenomenon they wish to correct for. d) Durgin-Rapaport: In [22], Durgin et al. make nu-
1) Frequency Coverage: The Riback-Medbo model [73] merous measurements around residential homes and trees at
attempts to correct for the (ill) effects of using a model 5.85 GHz. They use the collected data to come up with
intended for one frequency at a different frequency. The constant fitted values for losses associated with outdoor-indoor
algorithm the authors propose provides a fitted correction interface loss, loss due to single trees and stands of trees, as
when given the computed path loss, the assumed frequency, well as interior walls. These values are then used to form
the target frequency based on measurements they make a three the basis of a “partition” path loss model that computes the
different frequencies: final signal strength by computing the free-space loss and
then reducing it by the summed loss associated with each
a = 0.09 (32) obstruction. A model of the same flavor and by the same
6 authors is also proposed in [59], but for 2.4 GHz.
b = 256 ∗ 10 (33) e) Vegetation: There have been a number of works that
c = 1.8 (34) attempt to, in one way or another, model the losses due to
k = a(atan(f0 /b − c) − atan(f0 /b − c)) (35) vegetation obstructions. [49] proposes a very complex formu-
P Lf c = P L0 + 20log10(f /f0) − k(P L0 − A) (36) lation that attempts to model the diffraction above and around
a stand of trees. Parameters are provided for several species
where k is the correction factor which is used to correct the of trees, both in leaf and out of leaf. In [86], Torico et al.
path loss value (P L0 ) at a given frequency (f0 ) so that it is present an interesting but largely impractical theoretical model
better able to predict the loss at the desired frequency (f ). The for loss due to trees. In this work, trees are modeled as a screen
value A is the trivial free space loss (from equation 8) at the containing randomly placed cylindrical defractors. Although
original frequency (f0 ). The authors validate this model using not useful for general prediction, this model demonstrates
Authorized licensed use limited to: Telecom ParisTech. Downloaded on April 15,2024 at 16:38:27 UTC from IEEE Xplore. Restrictions apply.
264 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 15, NO. 1, FIRST QUARTER 2013

that vegetation can cause substantial losses (and that complex This model is commonly referred to as the “lognormal shad-
theoretical models can say mostly anything you want them owing” model and can be used as an empirically corrected
to if you tweak the parameters just so). In [11], Chee et al. model where values of α and σ are determined from mea-
present a similar analytical model. The lack of availability surements. This is the most coarse stochastic fading model
of vector data describing the location, shape, and type of and is usually considered to be appropriate only for modeling
vegetation prohibits use in most applications. A more practical large scale effects [71].
proposal is described in [58], where rain forest vegetation is Small scale (time varying) stochastic fading models typ-
modeled using four layers (ground, trees, foilage, sky) with ically look to either Rayleigh, Ricean, or Nakagami dis-
different propagation characteristics and interlayer ducting. tributions. The inquisitive reader can refer to the excellent
3) Directivity: Directivity models attempt to account for treatment by Skylar of Rayleigh and Ricean fading in [82]
multipath (scattering) losses that are unique to situations where or [97], [63] for discussions of the Nakagami distribution.
the transmitter, or more importantly the receiver, is using Some low-level applications may choose to explicitly model
a directional antenna. The problem here is that directional inter-symbol interference by determining the delay spread of
antennas “emphasize” some azimuthal directions more than arriving signals, as observed at the reciever, from a representa-
others, which leads to nontrivial multipath effects at the tive distribution. In [37], for instance, Greenstein et al., show
receiver. If the goal is to model a link involving directional that both delay-spread and path gain appear to be lognormally
antennas and the antenna is assumed isotropic (perhaps with distributed in their measurements at 900 MHz.
the gain assumed to be equal to the maximum gain of the c) Barclay-Okumura: The Barclay-Okumura model is a
main lobe), a substantial deviation from reality can occur. simple model for stochastic fading proposed by Barclay in
a) Gain Reduction Factor: In [36], Greenstein and Erceg [7] based on data collected by Okumura. It can operate in
find that there can be substantial gain reduction at the receiver. either “urban” or “suburban” mode, and computes a zero-mean
The authors make measurements in suburban New Jersey at Gaussian distributed fade with standard deviation σ:
1.9 Ghz and fit a model to the effects. The model is fitted to
the beamwidth of the receiving antenna and whether or not the 5.2 urban
a= (38)
measurements are made in winter (i.e. with or without leaves 6.6 suburban
on trees).
b) EDAM: In [5], Anderson et al. describe the Effective σ = 0.65log10(f )2 − 1.3log10 (f ) + a (39)
Directivity Antenna Model (EDAM). This is a bin-fitted model
derived from a large number of measurements made in several
F. Many-Ray Models
representative environments (multiple indoor and multiple
outdoor environments) both with commodity hardware and Many-ray models are typically referred to as ray-tracing
with special purpose hardware. The result is a model that, or ray-launching models in the literature. In our taxonomy
when given an environment class, will provide a correction as we call them “many-ray” models to highlight the way they
a function of the gain pattern at the receiver in the direction differ from all of the aforementioned models—they attempt
of the transmitter and vice versa. The model is also able to be to calculate the path loss by summing the loss along many
used in a stochastic fashion for a repeated measures approach distinct paths instead of only the line-of-sight (LOS) path.
and with or without a Gaussian distributed fading correction. These models require substantial and precise knowledge about
In [6], the authors showed this model was better suited to the environment. Two and three dimensional vector models of
making path loss predictions in simulation-based evaluations buildings and interfering structures are the most commonly
involving directionality than standard models. used data. These models trace the interaction of many indi-
vidual paths and these obstacles, computing reflection, refrac-
tion, and diffraction using the Uniform Theory of Diffraction
E. Stochastic Fading Models (UTD), or an equivalent numerical approximation. As a result,
Stochastic fading models add a random variable to a path they are able to compute not only the median path loss
loss model to account for additional fading in the wireless predicted at the receiver, but also the delay spread (which can
channel. This includes fades due to scattering and multipath be used to computer inter-symbol interference) and frequency
effects that are uncorrelated in measurements over small dis- shift (which can be used to model frequency-selective fading
tances (i.e., less than a wavelength). These fades are selective effects) of arriving signals.
in both time and frequency, meaning that attentuation may Early papers in this area include the work of Ikegami et
vary as a function of either (or both). Stochastic fading models al. in [44] and Vogler in [88], where it is proposed that
are especially useful in the design of physical layer/data-link mean field strength be calculated by computing diffractions
layer of wireless networks. and reflections from building vector data. Some work has
A number of measurement studies, find that residual error been done to increase the accuracy and speed of calculating
in an explicit fit to measurements follows a lognormal dis- diffractions (e.g., [77], [95] and the comparative discussion in
tribution. This is equivalent to adding a zero mean normally [23]).
distributed error term Xσ to equation 10: The early applications of these ideas were applied in two-
dimensional ray-tracing models. In [98], Zhang and Moayeri
propose a purely theoretical model that assumes a regular city
Prx = Ptx − (10αlog10 (d) + 20log10 (f ) + 32.45 + Xσ ) (37) grid and predicts a single reflected path (around corners) and
Authorized licensed use limited to: Telecom ParisTech. Downloaded on April 15,2024 at 16:38:27 UTC from IEEE Xplore. Restrictions apply.
PHILLIPS et al.: A SURVEY OF WIRELESS PATH LOSS PREDICTION AND COVERAGE MAPPING METHODS 265

a constant adjustment for other multipath effects. Different IV. M ODELING W ITH M EASUREMENTS
calculations are used based on whether the receiver is on a All of the preceding models discussed are a priori. They
neighboring street or a side (perpendicular) street. In [76], make predictions about a given network and a given environ-
Rustako et al. suggest that only 6 rays are necessary for model- ment either using analytical expectations about propagation
ing line-of-sight links in urban street-canyons. In [52], Kanatas or empirical models collected from a different (but hopefully
et al. suggest a simple two-dimensional ray tracing model that similar) environment, or some combination thereof. The final
assumes a uniform rectilinear building/street layout and makes category of models are those whose design is based on the
a minimal validation against measurements. In [74], Rizk et assumption that there is no single set of a priori constants,
al. propose a two-dimensional ray-tracing approach that can functions, or data that allow for sufficient description of a new
deal with arbitrary building layouts and go to some effort to environment with sufficient accuracy. These models assume
validate their approach. In [70], Piazzi et al. evaluate a 2-D ray that the burden of making some number of measurements
tracing approach in a residental environment and find decent is unavoidable. In a sense, these are more than models—
results when the transmitter is positioned above the rooftops. they define a method for collecting measurements (sampling
In [34], the authors extend the Walfisch-Ikegami model to strategy) and a means of predicting (interpolating) the values
include corrections from ray-tracing and static adjustments for at locations that have not been measured.
the presence of trees. The seminal work in this area is by W.C. Lee in [56]. In this
More recently, authors have proposed three-dimensional work, Lee proposes a theoretically justified methodology for
models that require substantially more computation. In [94], averaging signal strength. He suggests that a mobile receiver
Wölfle et al. propose a ray-optical three-dimensional model should make measurements in arcs at varying distances from
that utilizes substantial preprocessing to improve performance, the transmitter. He argues that measurements within 20 to 40
as well as using the COST-231 model for LOS links. In wavelengths of one another should be averaged to obtain a
[89] the same authors propose heuristics to simplify the central tendency and that an appropriate sample size is at
computational complexity of prediction by only calculating minimum, 36 measurements. For 2.4 GHz, this works out
the most important (“dominant”) paths. In [83], Sridhara et to between 0.625 and 1.25 m, which is in agreement with a
al. propose a ray tracing approach, but only claim that its ac- study made by Shin, 25 years after Lee’s original publication
curacy is sufficient for simulation (and not prediction). Finally, [81]. In this work, Shin does a measurement study of IEEE
[46] provides a survey of various ray tracing approaches. In 802.11b/g networks, attempting to model signal strength vari-
addition to those papers published in the academic literature, ation over small distances. He finds that the wideband modula-
there are also a number of commercial planning systems tion schemes used in 802.11g result in some immunity to fast
that provide similar prediction tools (e.g., [72], [93], [14]). fading effects, and that small scale variations are “averaged
The Remcom Wireless Insight software [72], for instance, out” within a radius of approximately 1 wavelength (3.1 mm
packages a number of popular path loss prediction models for 2.4 GHz). He discovers that measurements have a strong
discussed above with their own three-dimensional ray-tracing spatial correlation within ≈ 1 m and become uncorrelated at
system. larger distances. In [55] and [57], Lee expands his original
The majority of recent work in this area is concerned measurement based work into a general purpose fitted model
with optimization and preprocessing to make feasible the that is still commonly used in planning cellular networks.
intractable number of calculations required for this approach. In [30], Evans et al. put Lee’s proposals to work to model
Although in some ways, these models are the most advanced the propagation of a transmitter at 1.9 GHz and find that they
of all the models on the table, they are not useful in practice are able to achieve approximately 9 dB Root Mean Square
for accuracy-sensitive coverage mapping because of their large Error (RMSE). A similar approach was also taken in [62],
computation and data requirements. Computing the many path where Medeisis and Kajackas fit measurements to the Hata
loss estimates required to generate a coverage map for a model and do some investigation of the number of measure-
large urban area is simply outside the abilities of the current ments needed to sufficiently correct the model and appropriate
models in a reasonable amount of time. Those models that can measurement methods. They find that in their environment 15-
compute results quickly, do so by selecting a relatively small 20 measurements are needed to tune the model sufficiently,
subset of rays to model, which may or may not be the most and that measurements are most useful when taken in clusters
important. Precise two and three dimensional environmental along a path. In [19] the authors explicitly fit measurements
vector data is seldom available, becomes stale quickly, and in their environment but fail to show significant improvement
is often costly even when it is available. When this data is over a priori predictions (achieving, on average, 9 dB RMSE
available, it is not clear which attributes are most important— no matter the approach).
in many scenarios, building materials (and their conductivity
and permittivity properties) must also be known to make
accurate predictions. In short, while these models offer a great A. Explicit Mapping
deal of promise, there is still much work needed to understand Hills carried out some of the early high level work on
their accuracy, and reduce the cost associated with their use formalizing wireless planning in his attempts to design a
(both in terms of time and data acquisition). In particular, network for Carnegie Mellon University [40]. Based on his
developing an understanding of the relationship between the experiences, he would go on to develop a measurement
performance of these models to the fidelity of their input data apparatus for doing on the fly mapping of indoor propagation
is essential area for further validation. to aid in network planning [41]. Hills’ “Rollabout” cart counts
Authorized licensed use limited to: Telecom ParisTech. Downloaded on April 15,2024 at 16:38:27 UTC from IEEE Xplore. Restrictions apply.
266 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 15, NO. 1, FIRST QUARTER 2013

wheel rotations to determine position and orientation within a vector data terrain map that describes the types of buildings
a building. The network engineer must place a temporary between an AP and each possible receiver site (pixel). A
transmitter and roll the cart around collecting measurements. training phase determines the path loss per unit distance for
The software on the cart plots signal strength measurements each building type, which then informs the offset function:
and will even make suggestions about channel assignment 
to minimize interference with neighboring networks. In [31], β(n, p) = Cf × w(n, p, f ) (41)
Fretzagias and Papadopouli suggest a method for mapping f ∈F
indoor environments where the total area is divided into grid
where f ∈ F are the terrain “features” on the LOS path
cells. A large number of nodes are used to sound the channel
between the node n and point p, Cf is the fitted weight (i.e.,
and make measurements. Then the measurements from each
path loss per unit distance) of the feature type f and w(n, p, f )
node are used in tournament/voting fashion to determine the
is the length of intersection between this feature and the line-
average signal at each grid cell.
of-sight path between n and p.
In Robinson’s proposal, sufficient “pilot” measurements are
B. Partition Models made to determine the Cf values for all f and the environment
The next group of models worth mentioning are “partition wide α is determined. Then, this model is used to predict the
based” models, where measurements are taken in an environ- signal strength of each AP to a large number of equally spaced
ment where the key obstructions are identified (i.e., walls, points around the node. A coverage metric must be defined
trees, buildings, etc.). In this approach, measurements are (e.g., SN R > 20), which says where a point is “covered”
taken and static path loss values are fitted for each obstruction. or not. By applying this metric to the predictions around the
Once the model is bootstrapped with these fits, it can be radius of a node we obtain the range of the node as a function
used (in theory) in other environments. An early example of the azimuth angle. Robinson fits a step function to this curve
of this approach is in the very nice work by Durgin et al. and uses the number of segments in the fitted step function
in [22], where the authors study path loss in a suburban to create a “segmented” coverage prediction of each node
environment at 5.8 GHz. Naturally, this approach extends with a relatively small number of segments. The remainder
easily to indoor environments where there are a large number of modeling involves iterative refinement. A measurement is
of explicit obstacles (walls). This approach has been investi- made as close to each coverage boundary as possible and then
gated much more thoroughly by Rappaport and colleagues at the boundary is pushed or pulled by a constant amount. This
various frequencies [78], [4]. In [99] Zvanovec et al. propose process is repeated until the push/pull amount is less than some
a similar model. However, due to the lack of substantive threshold (Robinson suggests 3 dB, which seems reasonable
quantitative analysis in this paper it is difficult to draw strong based prior studies of expected repeated measures variance,
conclusions from the results. In [96], Xiang et al. propose e.g., [74]).
another partition based model that also gives some attention
to sampling. They propose a “lazy sampling” algorithm that
D. Active Learning and Geostatistics
greedily selects transmitter locations. A receiver is then used to
make measurements on a regular grid and the measurements As a generalization of the iterative refinement approach
are used to train a partition model. The authors show that described above, the machine learning literature offers an
this approach can produce an interpolated coverage map with approach called “active learning”. In active learning systems,
approximately 6 dB residual error. an algorithm is able to choose its training data, as opposed to
passively trained systems that must learn a behavior from a
set of “random” observations.
C. Iterative Heuristic Refinement
In [13], Cohn et al. provide a summary of this area, deriving
The most recent active measurement model is that of active learning approaches to three kinds of learning systems:
Robinson et al. in [75]. In this work, the authors attempt neural networks, mixed Gaussians, and locally weighted re-
to identify coverage holes in large wireless networks. They gression. Additional training data (samples) are chosen to
study the Technology for All (TFA) network operated by Rice minimize model variance. Cohn shows that active learning
and the Google WiFi network in Mountain View, California. approaches far outperform randomly selected training data for
Robinson’s approach combines an a priori model with a fitted training a model to solve the arm kinematics problem2.
partition model and then uses a push-pull heuristic to make Active learning has an analogous problem in the realm of
corrections from measurements. For a given Access Point (AP) geostatistics (and typically applied in ecological soil sampling)
node (n), and a given point (p), the SNR is predicted by: termed “optimized sampling” [87], [61]. In this version of
  the problem, additional data for a trained model is selected
d(n, p) by minimizing some metric of variance (Kriging variance
PdB (p, n) = P0 − 10αlog + β(n, p) (40)
d0 is generally used in geostatistical treatments). Regardless of
the domain from which it is drawn, the task is fundamental:
where P0 is the transmitter EIRP, d(n, p) is distance from
given some existing model, can we chose the next set of
the point to the node, α is the path loss exponent, d0 is
measurements that most improves the accuracy of the model
the reference distance, and β(n, p) is a fitted offset function.
Omitting the offset function, this equation is identical to 2 In the arm kinematics problem, a trained model attempts to predict the
equation 11 in section III-A2. The offset function makes use of tip position of a robotic arm given a set of joint angles.
Authorized licensed use limited to: Telecom ParisTech. Downloaded on April 15,2024 at 16:38:27 UTC from IEEE Xplore. Restrictions apply.
PHILLIPS et al.: A SURVEY OF WIRELESS PATH LOSS PREDICTION AND COVERAGE MAPPING METHODS 267

itself. Although there has been some preliminary work in of this quality can be obtained after only 20-40 measurements.
terms of applying geostatistical modeling to radio environment Hence, we can say with some confidence that whether a
prediction (e.g., [53]), there are still a great deal of open network operator does a small random sampling and basic fit,
questions in this direction. or carefully tunes an a priori model to their environment, they
can still expect predictions that are only accurate to within 3
V. C OMPARATIVE S TUDIES to 5 orders of magnitude.
The vast majority of existing work analyzing the efficacy of
path loss models has been carried out by those authors who VI. S UMMARY AND C ONCLUSION
are proposing their own improved algorithm. In such cases, Making sense of the vast and varied landscape of path
the authors often collect data in an environment of interest and loss models can be a precarious task for the uninitiated
then show that their model is better able to describe this data researcher. In this work, we have developed a new taxonomy
than one or two competing models. Unfortunately, this data is for reasoning about commonalities between these models. In
rarely published to the community, which makes comparative terms of functionality and intent, we can further categorize
evaluations impossible. One noteworthy exception is the work models into classes based on those that are appropriate for (a)
of the COST-231 group in the early 1990’s, which published coverage and radio environment mapping, (b) rough planning,
a benchmark data set (900 MHz measurements taken in and (c) simulation. Applications that require accurate maps of
European cities) and produced a number of competing models the radio environment are probably best suited for an active
that were well performing with respect to this reference [12]. measurement method that can resolve predictions with directed
This effort produced a number of well validated models that measurements. When it is not possible to make measurements
are tuned for 900 MHz transmitters in urban environments. of the environment directly, an experimenter must accept some
Similarly, there was substantial work done in the US, Japan, (possibly substantial) error. Many-ray methods are promising,
and several other countries in the 1960s and 1970s to come up but their accuracy is intimately tied to the accuracy of data
with accurate models for predicting the propagation of analog describing the environment and obstacles, which is seldom
TV signals (e.g., [16]). This flurry of work produced many of available at a useful resolution and can be very costly to
the models that are still used today in network simulators and collect and update. These models are also famously slow,
wireless planning tools: the ITM [42], the Egli Model [27], requiring a substantial amount of computation for even a
and the Hata-Okumura model [67], to name a few. However, few predictions. Those looking to path loss models for rough
it is unclear what the implications are of using these models, planning are able to choose amongst dozens of seemingly
which were created for use in a specific domain, to make similar proposals, accepting the caveat that it is impossible
predictions about another domain. to verify accuracy. For this reason, we recommend the most-
There are several works that compare a number of models heavily used standard models for the sake of comparability
with respect to some data. In [21], the authors compare five (i.e., Okumura-Hata, Longley-Rice ITM, etc.). Simulations
models with respect to data collected in rural and suburban have similar needs to rough planning applications, except
environments with a mobile receiver at 910 MHz. They they also require the prediction of a distribution of reasonable
discuss the abilities of each model, but abstain from picking values around the median for repeated-measured/Monte Carlo
a winner. In [2], the authors compare three popular models techniques. Hence, stochastic basic models (or deterministic
to measurements collected at 3.5 GHz by comparing a least models with a stochastic fading parameter) are likely the
squares fit of measurements to model predictions. The authors most suitable and there are several to chose from. Again,
highlight the best of the three, which turns out to be the there is value in choosing amongst the most well known,
ECC-33 model proposed in [29]. In [80], Sharma et al. do standard models (e.g., Hata with lognormal fading, or the
a very similar analysis, but instead focus on measurements recent Herring model).
made in India at 900 and 1800 MHz. In contrast to [2], Although there are many possible directions for future work
they find that the Stanford University Interim (SUI) mo In in this area, we believe that measurement-based methods and
[69], we performed the first in-depth and rigorous analysis rigorous (comparative) validation are most needed. Applica-
of a large number of diverse propagation models using a tions that make use of these models require an understanding
large and realistic data set from a production network. This of their real-world accuracy, and researchers need guidance
is also the first such comparative study looking at results in choosing amongst the many existing proposals. Work such
for the widely used 2.4 and 5.8 GHz bands. For the models as [69] and [80], are an important first step in this direction,
implemented in that study and the data sets analyzed, a priori but more work is needed to resolve the imbalance between
path loss modeling achieves, at least, 8-9 dB RMSE in urban the quantity of models proposed and the extent to which they
environments and ≈ 15 dB RMSE in rural environments. This have been validated in practice.
is true almost regardless of the model selected,how complex Of all the models discussed so far, we see two extremes
it is, or how well it is tuned. This bound seems to agree with in terms of information requirements. On one end of the
prior work at other frequencies in similar environments that spectrum are basic models, like the Hata model, that require
have also produced results with RMSE in the neighborhood of very little information about the environment—simply the link
9 dB (e.g., [30], [19]). Direct approaches to data fitting, such geometry and some notion of the general environmental cate-
as a straight line fit to the log/log relationship between path gory. At the other end are many-ray models which make use
loss and distance produce a similar level of error: 8-9 dB for of vector data for obstacles to calculate specific interactions,
urban environments and ≈ 15 dB for rural environments. Fits requiring knowledge of the exact position and shape of all
Authorized licensed use limited to: Telecom ParisTech. Downloaded on April 15,2024 at 16:38:27 UTC from IEEE Xplore. Restrictions apply.
268 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 15, NO. 1, FIRST QUARTER 2013

obstacles. In between these two extremes, there are very few [4] C.R. Anderson and T.S. Rappaport. In-building wideband partition loss
models. Possible example include The ITM and ITU-R 452 measurements at 2.5 and 60 ghz. IEEE Trans. Wireless Commun.,
3(3):922 – 928, May 2004.
models, which make use of some additional information from [5] Eric Anderson, Caleb Phillips, Douglas Sicker, and Dirk Grunwald.
public geographic datasets. A natural question then, is whether Modeling environmental effects on directionality in wireless networks.
there is some other source of data available that could be In 5th International Workshop on Wireless Network Measurements
(WiNMee), 2009.
used to inform better predictions, but is not as costly or [6] Eric Anderson, Gary Yee, Caleb Phillips, Douglas Sicker, and Dirk
difficult to obtain as detailed vector data. For instance: models Grunwald. The impact of directional antenna models on simulation
that make use of high resolution satellite orthoimagery and accuracy. In 7th International Symposium on Modeling and Optimization
in Mobile, Ad Hoc, and Wireless Networks (WiOpt), 2009.
machine vision techniques, a high resolution Digital Surface [7] Les Barclay. Propagation of radiowaves. Institution of Electrical
Model (DSM) (where surface clutter is not “smoothed away” Engineers, 2003.
as it is in digital elevation/terrain models, e.g., [45]), “crowd- [8] A. Blomquist and L. Ladell. Prediction and calculation of transmission
loss in different types of terrain. NATA AGARD Conf. Pub. CP. 144,
sourced” building vector data vis a vis Google Sketchup [1], Res. Inst. Nat. Defense, Dept. 3, S-10450:32/1–32/17, 1977.
or topographic and zoning maps (e.g., [75]). So far,this data- [9] David Brown and Gregory M. Stone. A report on technology in-
mining approach to prediction, although promising, has seen dependent methodology for the modeling, simulation, and empirical
verification of wireless communication system performance in noise and
little rigorous investigation. interference limited systems operation on frequencies between 30 and
There is simply no better way to generate truthful predic- 1500 mhz. Technical report, TIA, May 1997.
tions than to start with ground-truth itself. For this reason, [10] K. Bullington. Radio propagation for vehicular communication. IEEE
Trans. Antennas Propag., VT-26:295–308, 1977.
we believe that the future of wireless path loss prediction [11] Kin Lien Chee, S.A. Torrico, and T. Kurner. Foliage attenuation over
methods will be active measurement designs that attempt to mixed terrains in rural areas for broadband wireless access at 3.5 ghz.
extract information from directed measurements. In particular, IEEE Trans. Antennas Propag., 59(7):2698 –2706, july 2011.
[12] Dieter J. Cichon and Thomas Kürner. Digital mobile radio towards
geostatistical approaches that favor robust sampling designs future generation systems: Cost 231 final report. Technical report, COST
and explicitly model the spatial structure of measurements European Cooperation in the Field of Scientific and Technical Research
are promising (e.g., [53], [91]). General machine learning - Action 231, 1993.
[13] David A. Cohn, Zoubin Ghahramani, and Michael I. Jordan. Active
approaches, and active learning strategies may also be fruitful, learning with statistical models. Journal of Artificial Intelligence
but applying those methods to the domain of path loss mod- Research, 4:129–145, 1996.
eling and coverage mapping is currently unexplored. Future [14] AWE Communications. Wave propagation and radio network planning.
http://www.awe-communications.com/, January 2012.
work in this area is likely to focus on refining sampling and [15] Roger Coudé. Radio Mobile. http://www.cplus.org/rmw/english1.html,
learning strategies using measurement based methods, as well July 2010.
as extracting as much information as possible from existing [16] Jack Damelin, William A. Daniel, Harry Fine, and George V. Waldo.
Development of vhf and uhf propagation curves for tv and fm broad-
sources using data mining. Methods for parallelizing com- casting. Technical Report R-6602, Federal Communications Comission
putation and preprocessing datasets are also needed to make (FCC), September 1966.
predictions quickly (this is especially true when these models [17] William Daniel and Harry Wong. Propagation in suburban areas at
distances less than ten miles. Technical report, Federal Communications
are used in real time applications). And, once predictions are Commission, Office of Engineering and Technology, 1991.
made, efficient storage and querying of these spatial databases [18] Y.L.C. de Jong, M.H.J.L. Koelen, and M.H.A.J. Herben. A building-
is an opportune area for further work. transmission model for improved propagation prediction in urban mi-
crocells. IEEE Trans. Antennas Propag., 53(2):490 – 502, march 2004.
As the the prevalence and importance of wireless networks [19] J.N.C. de Oliveira, M.S. Alencar, and V.C. da Rocha. A new propagation
continues to grow, so too will the need for better methods model for cellular planning. In IEEE International Telecommunications
of modeling and measuring wireless signal propagation. In Symposium, 2006.
[20] Rafael S. de Souza and Rafael D. Lins. A new propagation model for
this paper we have given a broad overview of approaches 2.4 ghz wireless lan. In APCC, 2008.
to solving this problem proposed in the last 60 years. Most [21] Gilles Y. Delisle, Jean-Paul Lefévre, Michel Lecours, and Jean-Yves
of this work has been dominated by models that extend Chouinard. Propagation loss prediction: A comparative study with
application to the mobile radio channel. IEEE Trans. Antennas Propag.,
on the basic electromagnetic principles of attenuation with VT-34:86–96, 1985.
theoretical and empirical corrections. More recently, work [22] G. Durgin, T.S. Rappaport, and Hao Xu. Measurements and models for
has focused on developing complex theoretical deterministic radio path loss and penetration loss in and around homes and trees at
5.85 ghz. IEEE Trans. Commun., 46(11):1484 –1496, nov 1998.
models. We believe the next generation of models will be [23] G.D. Durgin. The practical behavior of various edge-diffraction formu-
data-centric, deriving insight from directed measurements and las. IEEE Antennas Propagat. Mag., 51(3):24 –35, 2009.
possibly using hybridized prediction techniques. Regardless of [24] ECC. The analysis of the coexistence of FWA cells in the 3.4 - 3.8
the approach that is taken, there is substantial possibility for GHz bands. Technical Report 33, European Conference of Postal and
Telecommunications Administrations, 2003.
future work in this area, with the promise of great impact in [25] R. Edwards and J. Durkin. Computer prediction of service areas for
many crucial applications. v.h.f. mobile radio networks. Electrical Engineers, Proc. Institution of,
116(9):1493 –1500, september 1969.
[26] R.E. Edwards and J. Durkin. Computer prediction of field strength in
R EFERENCES the planning of radio systems. In Proc. Inst. Elec. Eng., 1969.
[27] John J. Egli. Radio propagation above 40 mc over irregular terrain. In
[1] Google SketchUp. http://sketchup.google.com/, January 2011. Proc. IRE (IEEE), volume 45, pages 1383–1391, October 1957.
[2] V.S. Abhayawardhana, I.J. Wassell, D. Crosby, M.P. Sellars, and M.G. [28] V. Erceg, L.J. Greenstein, S. Tjandra, S.R. Parkoff, A. Gupta, B. Kulic,
Brown. Comparison of empirical propagation path loss models for fixed A. Julius, and R. Jastrzab. An empirically-based path loss model
wireless access systems. In VTC 2005-Spring, volume 1, pages 73–77, for wireless channels in suburban environments. In GLOBECOM’98,
May 2005. volume 2, pages 922–927, 1998.
[3] K. Allsebrook and J.D. Parsons. Mobile radio propagation in brittish [29] V. Erceg, K.V.S. Hari, and et al. Channel models for fixed wireless
cities at frequencies in the vhf and uhf bands. IEEE Trans. Antennas applications. Technical report, IEEE 802.16 Broadband Wireless Access
Propag., VT-26:313–323, 1977. Working Group, 2001.
Authorized licensed use limited to: Telecom ParisTech. Downloaded on April 15,2024 at 16:38:27 UTC from IEEE Xplore. Restrictions apply.
PHILLIPS et al.: A SURVEY OF WIRELESS PATH LOSS PREDICTION AND COVERAGE MAPPING METHODS 269

[30] G. Evans, B. Joslin, L. Vinson, and B. Foose. The optimization [60] John A. Magliacane. SPLAT! A Terrestrial RF Path Analysis Applica-
and application of the w.c.y. lee propagation model in the 1900 mhz tion for Linux/Unix, March 2008.
frequency band. In Vehicular Technology Conference, 1997 IEEE 47th, [61] B.P. Marchant and R.M. Lark. Optimized sample schemes for geosta-
volume 1, pages 87 –91 vol.1, May 1997. tistical surveys. Mathematical Geology, 39:113–134, 2007.
[31] C. Fretzagias and M. Papadopouli. Cooperative location-sensing for [62] A. Medeisis and A. Kajackas. On the use of the universal Okumura-Hata
wireless networks. In Pervasive Computing and Communications, 2004. propagation prediction model in rural areas. In IEEE VTC’00, 2000.
PerCom 2004. Proc. Second IEEE Annual Conference on, pages 121 – [63] M. Nakagami. Statistical Methods in Radio Wave Propagation, chapter
131, 2004. The m-distribution - A general formula of intensity distribution of rapid
[32] Harald T. Friis. A note on a simple transmission formula. In Proc. IRE fading, pages 3–36. Program Press, Oxford, 1960.
and Waves and Electrons, May 1946. [64] Aleksandar Neskovic, Natasa Neskovic, and George Paunovic. Modern
[33] Vijay K. Garg. Wireless Communications and Networking. Morgan approaches in modeling of mobile radio systems propagation environ-
Kaufmann, 2007. ment. IEEE Commun. Surveys Tuts., 3(3), 2000.
[34] N.C. Goncalves and L.M. Correia. A propagation model for urban [65] NIST. http://w3.antd.nist.gov/wctg/manet/calcmodels r1.pdf.
microcellular systems at the uhf band. IEEE Trans. Veh. Technol., [66] Yasuhiro Oda and Koichi Tsunekawa. Advances los path loss model in
49(4):1294 –1302, July 2000. microwave mobile communications. In 10th International Conference
[35] D.B. Green and A.S. Obaidat. An accurate line of sight propagation on Antennas and Propagation, 1997.
performance model for ad-hoc 802.11 wireless lan (wlan) devices. In [67] Y. Okumura, E. Ohmori, T. Kawano, , and K. Fukuda. Field strengh
Communications, 2002. ICC 2002. IEEE International Conference on, and its variability in VHF and UHF land mobile radio service. Rev.
volume 5, pages 3424 – 3428 vol.5, 2002. Electr. Commun. Lab., 16:825–73, 1968.
[36] L.J. Greenstein and V. Erceg. Gain reductions due to scatter on wireless [68] Hagen Paul Pfeifer. On the validation of radio propagation models.
paths with directional antennas. IEEE Commun. Lett., 3(6):169–171, http://jauu.net/data/pdf/propagation-models.pdf, January 2010.
1999. [69] Caleb Phillips, Douglas Sicker, and Dirk Grunwald. Bounding the error
[37] L.J. Greenstein, V. Erceg, Y.S. Yeh, and M.V. Clark. A new path- of path loss models. In IEEE Dynamic Spectrum Access Networks
gain/delay-spread propagation model for digital cellular channels. IEEE (DySPAN), May 2011.
Trans. Veh. Technol., 46(2):477 –485, may 1997. [70] Leonard Piazzi and Henry L. Bertoni. Achievable accuracy of site-
[38] Daniel Halperin, Wenjun Hu, Anmol Sheth, and David Wetherall. specific path-loss predictions in residential environments. IEEE Trans.
Predictable 802.11 packet delivery from wireless channel measurements. Veh. Technol., 48:922–930, 1999.
In ACM SIGCOMM, 2010. [71] Theodore S. Rappaport. Wireless Communications: Principles and
[39] K.T. Herring, J.W. Holloway, D.H. Staelin, and D.W. Bliss. Path- Practice. Prentice Hall, 2002.
loss characteristics of urban wireless channels. IEEE Trans. Antennas [72] REMCOM. Wireless insight: Wireless em propagation software.
Propag., 58(1):171 –177, jan. 2010. http://www.remcom.com/wireless-insite, January 2012.
[40] A. Hills. Large-scale wireless lan design. IEEE Commun. Mag., [73] M. Riback, J. Medbo, J.-E. Berg, F. Harrysson, and H. Asplund. Carrier
39(11):98–107, Nov 2001. frequency effects on path loss. In VTC 2006-Spring, volume 6, pages
[41] A. Hills and J. Schlegel. Rollabout: a wireless design tool. IEEE 2717 –2721, 7-10 2006.
Commun. Mag., 42(2):132 – 138, February 2004. [74] K. Rizk, J.-F. Wagen, and F. Gardiol. Two-dimensional ray-tracing
[42] G.A. Hufford. The ITS irregular terrain model, version 1.2.2, the modeling for propagation prediction in microcellular environments.
algorithm. http://flattop.its.bldrdoc.gov/itm.html. IEEE Trans. Veh. Technol., 46(2):508 –518, May 1997.
[43] G.A. Hufford, A.G. Longley, and W.A. Kissick. A guide to the use of [75] Joshua Robinson, Ram Swaminathan, and Edward W. Knightly. As-
the ITS irregular terrain model in the area prediction mode. Technical sessment of urban-scale wirelesss networks with a small number of
Report 82-100, NTIA, 1982. measurements. In MobiCom, 2008.
[44] F. Ikegami, T. Takeuchi, and S. Yoshida. Theoretical prediction of mean [76] A. J. Rustako, Noach Amitay, G. J. Owens, and R. S. Roman. Radio
field strength for urban mobile radio. IEEE Trans. Antennas Propag., propagation at microwave frequencies for line-of-sight microcellular
39:299–302, 1991. mobile and personal communications. IEEE Trans. Veh. Technol.,
[45] Intermap Technologies. Digital surface models. 40:203–210, 1991.
http://www.intermap.com/digital-surface-models, January 2011. [77] S. R. Saunders and F. R. Bonar. Explicit multiple building diffraction
[46] Magdy F. Iskander and Zhengqing Yun. Propagation prediction models attenuation function for mobile radio wave propagation. Electronics
for wireless communication systems. IEEE Trans. Veh. Technol., Letters, 27(14):1276–1277, July 1991.
50:662–672, 2002. [78] S.Y. Seidel and T.S. Rappaport. Path loss prediction in multifloored
[47] ITU-R. Digital topographic databases for propagation studies. Technical buildings at 914 mhz. Electronics Letters, 27(15):1384 –1387, 1991.
Report P.1058, ITU, 1999. [79] John S. Seybold. Introduction to RF Propagation. Wiley Interscience,
[48] ITU-R. Terrestrial land mobile radiowave propagation in the VHF/UHF 2005.
bands. ITU-R, 2002. [80] Purnima K. Sharma and R.K. Singh. Comparative analysis of propa-
[49] ITU-R. Attenuation in vegetation. Technical Report P.833, ITU, 2005. gation path loss models with field measured databases. International
[50] ITU-R. Prediction procedure for the evaluation of microwave interfer- Journal of Engineering Science and Technology, 2:2008–2013, 2010.
ence between stations on the surface of the earth at frequencies above [81] Hweechul Shin. Measurements and models of 802.11b signal strength
about 0.7 ghz. Technical Report P.452, ITU, 2007. variation over small distances. Master’s thesis, University of Delaware,
[51] ITU-R. Attenuation by atmospheric gases. Technical Report 676-8, 2010.
ITU, 2009. [82] B. Sklar. Rayleigh fading channels in mobile digital communication
[52] Athanasios G. Kanatas, Ioannis D. Kountouris, George B. Kostaras, and systems .i. characterization. IEEE Commun. Mag., 35(7):90 –100, July
Philip Constantinou. A utd propagation model in urban microcellular 1997.
environments. IEEE Trans. Veh. Technol., 46:185–193, 1997. [83] Vinay Sridhara and Stephan Bohacek. Realistic propagation simulation
[53] Abdullah Konak. A kriging approach to predicting coverage in wireless of urban mesh networks. Computer Networks, 51(12):3392 – 3412,
networks. Int. J. Mobile Network Design and Innovation, 2010. 2007.
[54] L-COM. Antenna specifications. http://www.l-com.com/, January 2011. [84] H. Steendam and M. Moeneclaey. Analysis and optimization of
[55] W.C.Y. Lee. Lee’s model [cellular radio path loss prediction]. In the performance of ofdm on frequency-selective time-selective fading
Vehicular Technology Conference, 1992, IEEE 42nd, pages 343 –348 channels. IEEE Trans. Commun., 47(12):1811 –1819, dec 1999.
vol.1, May 1992. [85] TIA. Wireless communications systems - performance in noise and
[56] William C. Y. Lee. Estimate of local average power of a mobile radio interference - limited situations - recommended methods for technology
signal. IEEE Trans. Veh. Technol., VT-34:22–27, 1985. - independent modeling, simulation, and verifications. Technical Report
[57] William C. Y. Lee and David J. Y. Lee. Microcell prediction in dense TSB-88-B-1, May 2005.
urban area. IEEE Trans. Veh. Technol., 47:246–253, 1998. [86] Saúl Torrico, Henry L. Bertoni, and Roger H. Lang. Modeling tree
[58] Le-Wei Li, Tat-Soon Yeo, Pang-Shyan Kooi, and Mook-Seng Leong. effects on path loss in a residential environment. IEEE Trans. Antennas
Radio wave propagation along mixed paths through a four-layered model Propag., 46(6):872–880, 1998.
of rain forest: an analytic approach. IEEE Trans. Antennas Propag., [87] J.W. van Groenigen, W. Siderius, and A. Stein. Constrained optimisation
46(7):1098 –1111, jul 1998. of soil sampling for minimisation of the kriging variance. Geoderma,
[59] Lorne C. Liechty, Eric Reifsnider, and Greg Durgin. Developing the 87(3-4):239 – 259, 1999.
best 2.4 ghz propagation model from active network measurements. In [88] Lewis E. Vogler. An attenuation function for multiple knife-edge
VTC’07, 2007. diffraction. Radio Science, 17(6):1541–1546, 1982.
Authorized licensed use limited to: Telecom ParisTech. Downloaded on April 15,2024 at 16:38:27 UTC from IEEE Xplore. Restrictions apply.
270 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 15, NO. 1, FIRST QUARTER 2013

[89] René Wahl, Gerd Wölfe, Philipp Wertz, Pascal Wildbolz, and Friedlich Caleb Phillips is a Ph.D. candidate in the De-
Landstorfer. Dominant path prediction model for urban scenarios. In partment of Computer Science at the University
14th IST Mobile and Wireless Communications Summit, 2005. of Colorado, Boulder. He works in the Systems
[90] J. Walfisch and H.L. Bertoni. A theoretical model of uhf propagation Laboratory under Professors Douglas Sicker and
in urban environments. IEEE Trans. Antennas Propag., 36(12):1788 Dirk Grunwald. His research is predominantly con-
–1796, dec. 1988. cerned with measurement and modeling strategies
[91] Matthias Wellens, Janne Riihijärvi, and Petri Mähönen. Spatial statistics for wireless networks and other complex systems.
and models of spectrum use. Comput. Commun., 32:1998–2011, He received his B.Sci. in Computer Science from
December 2009. Portland State University and his M.Sci in Computer
[92] J.H. Whitteker. Physical optics and field-strength predictions for wireless Science from the University of Colorado, Boulder.
systems. IEEE J. Sel. Areas Commun., 20(3):515 –522, apr 2002.
[93] EDX Wireless. Smart planning for smart networks.
http://www.edx.com/, January 2012.
[94] G. Wölfle, R. Hoppe, and F.M. Landstorfer. Radio network planning
with ray optical propagation models for urban, indoor, and hybrid
scenarios. In 11th IEEE Wireless Conference, 1999. Dirk Grunwald Dirk Grunwald is the Wilfred and
[95] H.H. Xia and H.L. Bertoni. Diffraction of cylindrical and plane waves Caroline Slade Endowed Professor at University of
by an array of absorbing half-screens. IEEE Trans. Antennas Propag., Colorado at Boulder, in the Department of Computer
40(2):170 –177, February 1992. Science. He received his Ph.D. from the University
[96] Zhe Xiang, Hangjin Zhang, Jian Huang, Song Song, and K.C. Almeroth. of Illinois, Urbana-Champaign under Dr. Daniel
A hidden environment model for constructing indoor radio maps. In Reed in the Department of Computer Science. He
World of Wireless Mobile and Multimedia Networks (WoWMoM 2005), has been a member of the faculty of the Department
pages 395 – 400, Jun 2005. of Computer Science at the University of Colorado
[97] N. Youssef, T. Munakata, and M. Takeda. Fade statistics in nakagami since 1989. He holds joint appointments in the Dept.
fading environments. In IEEE 4th International Symposium on Spread of Electrical and Computer Engineering and the
Spectrum Techniques and Applications Proceedings, volume 3, pages Interdisciplinary Telecommunications Program.
1244–1247, 1996.
[98] Wei Zhang and N. Moayeri. Closed-form expressions for the prediction Douglas Sicker Doug is currently serving as the
of microcellular mobile radio propagation in urban environments. IEEE Chief Technology Officer and Senior Advisor for
Trans. Antennas Propag., 51(5):952 – 956, may 2003. Spectrum at the National Telecommunications and
[99] Stanislav Zvanovec, Pavel Pechac, and Martin Klelpal. Wireless lan net- Information Administration. Doug is also an as-
works design: Site survey or propagation modeling? Radioengineering, sociate professor in the Department of Computer
12(4):42–49, December 2003. Science at the University of Colorado at Boulder
with a joint appointment in the Interdisciplinary
Telecommunications Program. Doug holds a Ph.D.
from the University of Pittsburgh.

Authorized licensed use limited to: Telecom ParisTech. Downloaded on April 15,2024 at 16:38:27 UTC from IEEE Xplore. Restrictions apply.