sensors

Article
Analysis of DME/DME Navigation Performance and
Ground Network Using Stretched-Front-Leg
Pulse-Based DME
Euiho Kim
Department of Mechanical & System Design Engineering, Hongik University, 04066 Seoul, Korea;
[email protected]




Received: 16 August 2018; Accepted: 27 September 2018; Published: 29 September 2018


Abstract: Global navigation satellite systems (GNSS) have become a primary navigation means for
aircraft. However, the signal power of GNSS is very weak, and its service can be disrupted at any
time when there is interference or jamming. For this reason, the Federal Aviation Administration
(FAA) in the United States has recently chosen a distance measuring equipment (DME)-based aircraft
navigation technique, called DME/DME, as an alternative aircraft navigation means for use by
around 2030. The reason that the FAA plans to use DME/DME in such a short duration, by around
2030, is presumed to be because the ranging accuracy of DMEs is between 70 to 300 m, which is
about 7 to 30 times worse than that of GNSS. Thus, a significant loss of positioning performance is
unavoidable with current DMEs. To make DME/DME a more competent alternative positioning
source, this paper proposes an advanced DME that could provide a ranging accuracy of around
30 m by employing a recently developed Stretched-Front-Leg (SFOL) pulse. The paper introduces
optimal ground station augmentation algorithms that help to efficiently transform the current DME
ground network to enable a DME/DME positioning accuracy of up to 0.3 nm or 92.6 m with a
minimal number of new ground DME sites. The positioning performance and augmented ground
network using the proposed SFOL pulse-based DME are evaluated in two regions which have distinct
terrain conditions.

Keywords: DME; GNSS jamming; APNT

1. Introduction
Global navigation satellite systems (GNSS) have been widely used for aircraft navigation, and
their role will be more important in future air traffic control. For example, the USA Next Generation
Air Transportation System (NextGen) and Single European Sky ATM Research (SESAR) in Europe
heavily depend on GNSS to meet future air traffic demands. However, it is well known that the GNSS
signal can be easily lost in the presence of interference and jamming because of its weak signal power.
In fact, South Korea experienced Global Positioning System (GPS) outages in 2010 and 2012 due to
the strong GPS jamming signal transmitted from North Korea [1]. These events caused serious safety
issues for aircraft and ships using GPS as a primary navigation source.
In fact, GNSS outages due to interference or jamming were expected before the events in South
Korea. The Federal Aviation and Administration (FAA) in the USA started an Alternative Position,
Navigation, and Timing (APNT) program in 2010 to develop a GNSS back-up system for possible
GNSS outages [2,3]. The envisioned APNT system by the FAA should have resilience such that it is
difficult to interfere with. In addition, it must be able to continuously operate without being dependent
on GNSS, and it should be able to serve various users such as commercial airliners, as well as general
aviation. Another important APNT system requirement is that its service should be compatible with
the legacy users, as well as future users such as unmanned aircraft systems.

Sensors 2018, 18, 3275; doi:10.3390/s18103275 www.mdpi.com/journal/sensors

Sensors 2018, 18, 3275 2 of 19

Many clever approaches have been proposed for an APNT architecture, including wide area
multilateration (WAM), GNSS-like pseudolites, distance measuring equipment-based navigation
(DME/DME) [4,5], enhanced DME using a carrier phase [6], high-accuracy DME using alternative
pulses [7,8], a mosaic DME/pseudolite hybrid system [9], and L-band Digital Aeronautical
Communication System type 1 (LDACS1) [10]. First, WAM uses aircraft surveillance signals such as
Automatic Dependent Surveillance-Broadcast or 1090 Extended Squitter as ranging sources. Using the
known location of WAM receivers and their range measurements, a WAM master station computes
an aircraft position and sends the position solution to the aircraft. Because of the delivery of the
aircraft position from the ground, there has been concern regarding the security of the system. The
GNSS-like pseudolites operate similarly to GNSS with much higher signal power, except that ranging
sources are installed on the ground rather than in outer space. This approach requires a pseudolite
network infrastructure with a precise time synchronization, which requires significant resources. The
DME/DME is a traditional ground nav-aid system, and a DME ground network is well established
around the world. In DME/DME positioning, the aircraft measures a slant range to the ground DME
transponders in view and computes its position through the multilateration technique. Because the
aircraft position is determined onboard, there is no security concern like with WAM. However, the
drawback of the DME/DME is a poor positioning accuracy, which can be a few hundred meters. To
improve DME/DME positioning accuracy, the enhanced DME was proposed. This method uses a DME
carrier phase to measure the distance between aircraft and a ground DME transponder in centimeter
levels of accuracy, which requires a more precise oscillator than the one used in today’s DME system.
On the other hand, the proposed alternative pulses are to improve DME/DME positioning accuracy by
replacing the conventional Gaussian pulse. The alternative pulses are able to suppress multipath and
noise impact on the range measurements. It seems to be possible to transmit the alternative pulses in a
modern existing DME without a hardware change but a power amplifier pre-distortioner algorithms
tuned for a Gaussian pulse should be adjusted for an alternative pulse. The mosaic DME/pseudolite
hybrid system proposed to use one DME ground station with a few pseudolites. This unique design
enables an aircraft 3D positioning with one DME station, but its positioning accuracy can be up to
several hundred meters, and a pseudolite installation is required. The LDACS1 is a future aeronautical
communication system in L1 band and can be used as a ranging source, as well. This system has been
actively investigated and requires further research at this point.
Among these candidates, the FAA has recently chosen DME/DME as a short-term APNT solution
by 2030, and has planned to augment the currently deployed DME ground transponder network
to fill DME/DME coverage gaps in the conterminous USA [11]. It is presumed that the reason for
using DME/DME as an APNT is the well-established DME ground network and widespread use
of DME avionics in commercial airliners. DME measures the slant range between the aircraft and
the ground transponder based on the radar principle [4]. Using the multiple ranges from ground
DME transponders, the DME/DME positioning technique computes the aircraft position using
multilateration. One of the drawbacks of using the legacy DME equipment for APNT is the lack
of ranging accuracy. The ranging accuracy of today’s DME is around 7 to 30 times worse than GNSS
ranging accuracy [5], which is insufficient to support the targeted APNT positioning accuracy for
navigation (0.3 nm) and surveillance (92.6 m) originally proposed in [4,5]. To improve DME raging
accuracy, alternative pulse waveforms have been proposed [7,8], including smoothed concave polygon
(SCP) and stretched-front-leg (SFOL) pulses. A recent analysis of the SFOL pulse showed that the
multipath-induced ranging error of the SFOL pulse was about four and half times and two and
half times less than the conventional Gaussian and SCP pulses, respectively. Therefore, if an SFOL
pulse is implemented in DMEs, its higher ranging accuracy could significantly improve DME/DME
positioning accuracy.
This paper assesses a possible ranging accuracy of an SFOL pulse-based DME and investigates
DME ground network augmentations to meet a DME/DME positioning accuracy of 0.3 nm or 92.6 m
in two distinct terrain conditions. The augmentation of a DME ground network is planned in a way
Sensors 2018, 18, 3275 3 of 19

that a minimal number of new DME sites is added to the existing DME network while meeting the
desired alternative aircraft positioning accuracy. In this paper, Section 2 briefly gives an overview of the
performance of today’s DME and desirable APNT performance requirements. Section 3 introduces the
proposed SFOL pulse-based DME. Section 4 presents DME ground network optimization algorithms
based on binary integer linear programming. Section 5 presents the optimized DME ground networks
used to support a horizontal positioning accuracy of 0.3 nm and 92.6 m in selected regions. A discussion
of the results follows in Section 5.

2. Overview of Distance Measuring Equipment and Requirements on Alternative

Aircraft Positioning
DME consists of ground transponders and aircraft avionics. Aircraft DME avionics first interrogate
ground transponders, and the transponder replies back to avionics through a pair of pulses. The
time elapsed from the interrogation to the receipt of the replies is used to measure the slant range
between the aircraft and the ground transponders [4,11]. The operational frequency ranges of the DME
are from 960 to 1215 MHz. This frequency range is divided into 252 channels, and each channel is
separated by 1 MHz, and a ground DME transponder is assigned its own channel frequency. To find
the aircraft position using DMEs, referred to as DME/DME positioning, multi-channel or scanning
DME avionics simultaneously measure slant ranges to multiple ground transponders in view and
compute aircraft position using the known coordinates of the transponders stored in a flight computer
and multilateration principles. The number of available channels in a scanning DME is typically
between two and six.
To use DME/DME as an alternative positioning means to GNSS, its service area should be down
to terminal area where aircraft positioning accuracies satisfy 1 nm (95%) [12]. It would also be desirable
for DME/DME to be capable of supporting an approach phase of flight, which requires up to 0.3 nm
positioning accuracy. In addition, to use DME/DME as a surveillance system, its positioning accuracy
should be better than 92.6 m, which is the aircraft tracking accuracy of secondary surveillance radars
(SSR) [5].
The accuracy requirement for DME/DME-based navigation, i.e., total system error (TSE), is a
statistical sum of navigation system error (NSE), flight technical error (FTE), and path definition error
(PDE) [3], such that
TSE2 = FTE2 + NSE2 + PDE2 (1)

The NSE is also called a position estimation error, and FTE is a path steering error and depends
on the flight control method. The PDE is the error associated with route waypoints and is usually
considered to be zero, because waypoints can be accurately surveyed nowadays. The NSE is caused by
a positioning system and can be formulated for DME/DME as follows

NSE = HDOP·σrange (2)

where HDOP is a horizontal dilution of precision, and σrange is the ranging accuracy of DME. HDOP is
a goodness of the relative geometry between the aircraft and the ground DME transponders in view
and can be computed from a positioning error covariance matrix. For a DME positioning system to
meet an NSE, the required HDOP is given by
s
TSE2 − FTE2
HDOP = 2
(3)
4σrange

For example, an FTE of 0.25 nm (95%) and σrange of 70 m (95%) require a 4.39 or lower HDOP
value to satisfy a TSE value of 0.3 nm. Please note that HDOP depends on a ground network layout,
and the ground network optimization algorithms discussed in Section 4 search for an efficient ground
network layout that provides a targeted HDOP.
Sensors 2018, 18, x FOR PEER REVIEW 4 of 19
Sensors 2018, 18, 3275 4 of 19

3. SFOL Pulse and Proposed SFOL-Based DME

3. SFOL
The Pulse and Proposed
conventional SFOL-Based
DME widely DME
uses Gaussian pulses. The Gaussian DME pulse has a narrow
spectral density that does not interfere with other
The conventional DME widely uses Gaussian pulses. neighborhood channels,
The Gaussian DME butpulse
its multipath and
has a narrow
noise-induced
spectral range
density that error
does could be large,with
not interfere up toother
a hundred meters, due
neighborhood to its slowly
channels, but itsincreasing
multipathrising
and
edges. This section introduces an alternative SFOL DME pulse that is insensitive to multipath
noise-induced range error could be large, up to a hundred meters, due to its slowly increasing rising effects
and significantly
edges. This sectionenhances
introducesa DME ranging accuracy.
an alternative SFOL DME This section
pulse also
that is presents to
insensitive anmultipath
advancedeffects
DME
that significantly
and is based on the SFOL pulse
enhances a DME and DME/P
ranging (precision)
accuracy. Thislike hardware.
section also presents an advanced DME that
is based on the SFOL pulse and DME/P (precision) like hardware.
3.1. Overview of SFOL Pulse
3.1. Overview of SFOL Pulse
The SFOL pulse was designed by using genetic algorithms [8]. The SFOL pulse has a rise time
of 2.8The
μs, SFOL
a width of 3.4
pulse wasμs,designed
and a falling timegenetic
by using of 3.0 μs. Thus, it satisfies
algorithms [8]. Thethe pulse
SFOL shape
pulse has requirements
a rise time of
in the
2.8 µs, acurrent
width ofDME ground
3.4 µs, transponder
and a falling time ofspecifications
3.0 µs. Thus, it[13]. Figure
satisfies the 1pulse
compares a conventional
shape requirements in
Gaussian
the currentpulse,
DME SCP pulse,
ground and SFOLspecifications
transponder pulse. The overall shape1of
[13]. Figure the SFOL
compares pulse is very Gaussian
a conventional different
from the
pulse, SCP other two
pulse, pulses,
and SFOLand theThe
pulse. unique shape
overall of the
shape rising
of the SFOL edge is effective
pulse in reducing
is very different frommultipath
the other
effects.
two pulses, and the unique shape of the rising edge is effective in reducing multipath effects.

Figure 1. Comparison of Gaussian, Smoothed Concave Polygon, and Stretched-Front-Leg pulses.

Figure 1. pulses.

Figure
Figure 22 compares
compares the the ranging
ranging performance
performance of the Gaussian,
of the Gaussian, SCP,
SCP, and
and SFOL
SFOL pulses
pulses with
with respect
respect
to
to the injected multipath. Multipath delays range from 0 to 6 μs, and the multipath to direct
the injected multipath. Multipath delays range from 0 to 6 µs, and the multipath to direct pulse
pulse
amplitude
amplitude ratioratio is
is 0.3. With the
0.3. With the injected
injected multipath conditions, the
multipath conditions, the root
root mean
mean square
square (RMS)
(RMS) of
of the
the
multipath-induced ranging errors of Gaussian and SCP pulses were 26.1 and
multipath-induced ranging errors of Gaussian and SCP pulses were 26.1 and 14.9 m, respectively. 14.9 m, respectively.
The
The RMS
RMS of of SFOL
SFOL waswas 5.9
5.9 m.
m. Figure
Figure 33 shows
shows the
the sensitivity
sensitivity ofof the
the multipath
multipath mitigation
mitigation performance
performance
of
of the
the SFOL
SFOL pulse
pulse under
under aa noisy
noisy environment.
environment. As As the
the signal-to-noise
signal-to-noise ratio
ratio decreases,
decreases, the
the multipath
multipath
mitigation performance of the SFOL pulse tends to decrease as well. However,
mitigation performance of the SFOL pulse tends to decrease as well. However, the SFOL pulse the SFOL pulse is
is still
still able to mitigate multipath range errors by 56.9% and 36.0% of those of Gaussian
able to mitigate multipath range errors by 56.9% and 36.0% of those of Gaussian and SCP pulses, and SCP pulses,
respectively,
respectively, even
even atat an
anSNR
SNRof of24
24dB.
dB.
 x FOR PEER REVIEW
Sensors 2018, 18, 3275 5 of 19
Sensors 2018, 18, x FOR PEER REVIEW 5 of 19

Figure 2. Range error envelopes due to in-phase and out-of-phase multipaths for SCP, SFOL, and
Figure
Figure 2. Range error
2. pulses.
Range error envelopes
envelopes due
due to
to in-phase
in-phase and
and out-of-phase
out-of-phase multipaths
multipaths for
for SCP,
SCP, SFOL,
SFOL, and
and
Gaussian
Gaussian pulses.
Gaussian pulses.

Figure 3. Averaged root mean square (RMS) range errors in 1000 cases of noise injection with five
Figureof3.signal-to-noise
levels Averaged rootratio
mean square (RMS) range errors in 1000 cases of noise injection with five
(SNR).
levels of signal-to-noise ratio (SNR).
3.2. Proposed SFOL Pulse-Based DME
3.2. Proposed SFOL Pulse-Based DME
3.2. Proposed SFOLrange
The primary Pulse-Based DME in DME include timing biases, multipath, and thermal noise.
error sources
The primary range error sources in DME include timing biases, multipath, and thermal noise.
The ranging error range
The primary budgets in the
error DME in
sources specifications
DME include and modern
timing state-of-the-art
biases, DMEs
multipath, and are listed
thermal in
noise.
The ranging error budgets in the DME specifications and modern state-of-the-art DMEs are listed in
Table 1 [5].
The ranging It is also expected that current DMEs being operated are the mixture of the legacy DMEs
Table 1 [5]. Iterror budgets
is also in the
expected thatDME specifications
current DMEs being and modernare
operated state-of-the-art
the mixture ofDMEs are listed
the legacy DMEs in
Table 1 [5]. It is also expected that current DMEs being operated are the mixture of the
barely meeting specifications and modern state-of-the-art DMEs. Although the SFOL pulse could legacy DMEs
barely meeting specifications and modern state-of-the-art DMEs. Although the SFOL pulse could
Sensors 2018, 18, 3275 6 of 19

barely meeting specifications and modern state-of-the-art DMEs. Although the SFOL pulse could
effectively reduce multipath-induced ranging errors, suppression of other ranging error sources is also
desirable to have a better ranging accuracy, as shown in Table 1.
To maximize DME ranging accuracy with current technologies, it is recommended to transmit
an SFOL pulse in precision DME (DME/P)-like hardware that provides 10–30 m ranging accuracy
depending on operational modes [14]. DME/P was originally developed to support aircraft automatic
landing with a microwave landing system. Because DME/P uses a cos/cos2 pulse shape with an abrupt
change at the rising edge, which causes a wide spectrum, its transmission power is typically less than
100 Watts to avoid being an interference source to other nav-aid signals. Therefore, the service coverage
of DME/P is at most 20 nm or so. If an SFOL pulse is employed in DME/P hardware, referred to
DME/S (DME/SFOL) in this paper, DME/S would be able to provide high-ranging accuracy with the
same coverage area as DME/N (DME Normal) up to approximately 200 nm without any interference
concerns. The projected ranging accuracy budget of DME/S is also listed in Table 1. The timing biases
in the transponder and avionics were taken from DME/P standards [14]. The ranging accuracy of
propagation and noise was based on simulation results that the SFOL pulse mitigated multipath and
noise-induced errors by 56.9% at 24 dB SNR, as shown in Figure 3.

Table 1. Ranging accuracy budget of legacy DME, modern DME, and DME/SFOL.

Modern State-of-the-Art
95% Accuracy (m) Legacy DME (DME Spec.) DME/S
DME Performance [2,13]
Signal in Space
75.0 15.0
Reply Delay (m) 10.0 (DME/P)
(FAA-E-2996) (measured)
127.3 39.0
Propagation to Aircraft/Noise (m) 16.8 (analysis)
(derived) (analysis)
Air/Avionics
315.0 62.0
Avionics Bias (m) 15.0 (DME/P)
(DO-189) (derived)
Propagation to 127.3 54
16.8 (analysis)
Transponder/Noise (m) (derived) (analysis)
Total root sum square (RSS) (m) 370.4 92.1 29.8

4. DME Ground Network Development Strategy

The current DME ground network is mainly the legacy of DME and Very High Frequency
omnidirectional range (VOR)-based navigation, called rho-theta navigation. In rho-theta navigation,
the aircraft needs to see one DME collocated with VOR, while DME/DME positioning must have
lines of sight to two or more ground DME stations. Thus, DME/DME requires a greater number of
ground stations than rho-theta navigation, and the FAA is currently augmenting the DME ground
network to fill the DME/DME coverage gaps in the conterminous USA [12]. This section discusses
how a DME ground network can effectively augment the existing DME network using binary integer
linear programming [15,16].

4.1. Overall Architecture Development Approach

The optimization of a DME/DME ground network is determined through the process shown
in Figure 4. The process takes the following parameters as input: TSE, FTE, existing DME station
network topology, defined airspace, terrain geography, and σrange . Using these inputs, the process
involves searching for an optimal ground station network having a minimum total number of stations
while possessing the maximum number of existing DME stations. Because DME/DME operation
is capacity-limited, it is necessary to perform a capacity analysis to check if the determined optimal
station layout can handle the interrogation requests from the expected future air traffic. In the previous
Sensors 2018, 18, 3275 7 of 19

analysis, it was found that a ground network meeting the positioning accuracy requirement had no
problem in processing the interrogation rates of future air traffic [15]. Therefore, the capacity analysis
Sensors 2018, 18, x FOR PEER REVIEW 7 of 19
is not considered in this paper.

Figure 4. DME/DME
DME/DME network
networkplacement
placement procedures.
procedures.

Coverage Analysis Methodology Using Binary Linear Integer Programming

Coverage Analysis Methodology Using Binary Linear Integer Programming
The proposed process generates an optimized DME/DME ground station network by filling the
The proposed process generates an optimized DME/DME ground station network by filling the
coverage gaps of the currently deployed DME ground network to provide the required alternative
coverage gaps of the currently deployed DME ground network to provide the required alternative
positioning service. When there are only small local coverage gaps, the augmentation problem is
positioning service. When there are only small local coverage gaps, the augmentation problem is
straightforward. However, when the coverage gaps extend to a large area, augmenting new DME sites
straightforward. However, when the coverage gaps extend to a large area, augmenting new DME
to an existing ground network becomes a complex sensor network design problem.
sites to an existing ground network becomes a complex sensor network design problem.
These facility location or sensor network problems have been an active research subject for
These facility location or sensor network problems have been an active research subject for
several decades [17]. The sensor network problem is nondeterministic polynomial time (NP)-complete;
several decades [17]. The sensor network problem is nondeterministic polynomial time (NP)-
therefore, heuristics-based
complete; search algorithms
therefore, heuristics-based search tailored
algorithms for atailored
particular
forproblem
a particular haveproblem
to be used, have because
to be
there are no general efficient algorithms [18]. For example, Roa et al. developed
used, because there are no general efficient algorithms [18]. For example, Roa et al. developed a meta- a meta-heuristic
approachapproach
heuristic which balanced the precision
which balanced and availability
the precision of positioning
and availability and looked
of positioning andforlooked
a minimum for a
number of ranging sources through a diversified local search [19].
minimum number of ranging sources through a diversified local search [19]. Genetic algorithms Genetic algorithms have havebeen
used to design an optimal sensor network using dilution of precision (DOP)
been used to design an optimal sensor network using dilution of precision (DOP) or its variant as a or its variant as a fitness
functionfunction
fitness [20,21]. [20,21].
Ref. [22,23]
Ref.optimized the internode
[22,23] optimized distances distances
the internode of rangingof sources
ranging to find
sourcesa positioning
to find a
positioning network resulting in a minimum average mean square error (MSE) in the estimatedThese
network resulting in a minimum average mean square error (MSE) in the estimated user position. user
prior studies looked for optimal sensor layouts in a limited space with few
position. These prior studies looked for optimal sensor layouts in a limited space with few constraints. constraints. The optimal
planning
The of augmenting
optimal planning ofDME ground networks
augmenting DME ground mustnetworks
consider amust
much larger area
consider and constraints
a much larger areasuch and
as existing DME sites, terrain geometry, radio propagation, and airport location.
constraints such as existing DME sites, terrain geometry, radio propagation, and airport location. This This section presents
heuristics-based
section presentssensor planning algorithms
heuristics-based that are tuned
sensor planning for the DME
algorithms that augmentation
are tuned for problem.
the DME
To develop
augmentation problem.effective heuristics for the DME augmentation problem, it is necessary to understand
the characteristics of DME/DME-based
To develop effective heuristics for theaircraft navigation. First,
DME augmentation as shown
problem, in Figureto5,understand
it is necessary an aircraft
requires
the more stringent
characteristics positioning accuracy,
of DME/DME-based aircraftasnavigation.
its altitudeFirst,
is lower aroundinan
as shown airport.
Figure 5, an However,
aircraft
in the lower altitudes, the line of sight from aircraft to a DME ground
requires more stringent positioning accuracy, as its altitude is lower around an airport. However, becomes limited such that in
positioning accuracy is likely lower than in high-altitude airspace. Second,
the lower altitudes, the line of sight from aircraft to a DME ground becomes limited such that the lower HDOP induces
better positioning
positioning accuracyaccuracy,
is likely and lower
lower HDOP
than can generally
in high-altitude be a result
airspace. whenthe
Second, a user
lower is HDOP
surrounded inducesby
ranging sources. Aircraft at high altitudes do not usually play a role in the
better positioning accuracy, and lower HDOP can generally be a result when a user is surrounded by design of a DME network
because it
ranging can seeAircraft
sources. many DMEs at high with a good do
altitudes azimuth angular
not usually playdiversity and
a role in thecan easilyofmeet
design a DME the network
required
accuracy [15]. Therefore, useful heuristics include first placing
because it can see many DMEs with a good azimuth angular diversity and can easily meet DMEs outside low-altitude regions
the
such that aircraft meet a more stringent positioning requirement. Also,
required accuracy [15]. Therefore, useful heuristics include first placing DMEs outside low-altitudeDMEs on the ground must
have some
regions such distance between
that aircraft meetthem to avoid
a more placing
stringent multiplerequirement.
positioning DMEs in a local Also,area,
DMEs whichon themay result
ground
must have some distance between them to avoid placing multiple DMEs in a local area, which may
result in redundant sites. An additional heuristic is the preference for existing stations over new
stations because the re-use of the existing stations would be expected to reduce the overall APNT
architecture development costs.
Sensors 2018, 18, 3275 8 of 19

in redundant sites. An additional heuristic is the preference for existing stations over new stations
because the re-use of the existing stations would be expected to reduce the overall APNT architecture
development
Sensors 2018, 18, xcosts.
FOR PEER REVIEW 8 of 19

Figure 5.
Figure Station placement
5. Station placement heuristics
heuristics set-up.
set-up.

With those heuristics, the method proposed below uses a two-step binary integer linear
With those heuristics, the method proposed below uses a two-step binary integer linear
programming (BILP) formulation to search for valid solution sets. In this approach, terrain, and
programming (BILP) formulation to search for valid solution sets. In this approach, terrain, and
airspace were sectored into grids. The formulation of the first step is
airspace were sectored into grids. The formulation of the first step is
nn
T
min Z= ∑ wwiixxii =
minZ  wwT xx
i=
i11
subject
subjecttoto:: Ax Ax ≥bb
Cx ≤ d (4)
Cx  d (4)
Fx ≤ g
FxT g
w x ≤ Zmin
xi x∈ {Z
T
w 1}
0,min
 
xi  0,1
where Z is the cost function of a candidate DME network. x is the index vector of a candidate station
where Z is
location. If the
xi =cost function
1, then of a candidate
it contains a candidate DME network.
station. x is the xindex
Otherwise, i = 0. vector of a candidate
The vector station
w is a weighting
location.
factor on If xi = 1, thenstation
a particular it contains a candidate
location, station.value
and its element Otherwise,
is lowerxi at= 0.
theThe vector candidate
preferred w is a weighting
station
factor on a The
locations. particular
matrixstation
A is a location,
visibilityand its element
matrix and hasvalue is lower about
information at the preferred
the radio candidate
propagation station
line
locations.
of The matrix
sight between A is in
aircraft a visibility matrix and
a given location andhas theinformation about the
candidate station radio propagation
locations. The ith row line
of ofA
sight between
corresponds to aircraft
an aircraftin alocation
given location
grid and and the jth thecolumn
candidate to a station
candidate locations.
station The ith row
location grid.ofThe A
corresponds
elements to an aircraft
of matrix A also location grid and
take on values of the
0 orjth 1. column to a candidate
If the aircraft at the ithstation location“sees”
row location grid. Thethe
elements
station at of
thematrix A alsolocation,
jth column take on values of 0toor
Ai,j is set 1. 1.Otherwise,
If the aircraftAi,j at theThe
is 0. ith row
vectorlocation “sees” the
b represents
station atminimum
required the jth column
numberlocation,
of visibleAi,jstations
is set to 1. Otherwise,
at the corresponding Ai,j isaircraft
0. Thelocation.
vector b represents the
required
C and minimum
d controlnumber of visible
the minimum stations atbetween
separation the corresponding
stations. If aircraft location.
the distance between ith and jth
C and
stations d control
is less than the theminimum
minimumseparation,
separationCbetween i,j is set to stations. If the distance
1. Otherwise, between
Ci,j is equal ith and
to 0. The vectorjth
stations
d is less
is a vector ofthan the
all 1s, minimum
which meansseparation,
that only one Ci,j isstation
set to is1. allowed
Otherwise, Ci,jinside
to be is equal
thetoseparation
0. The vector limit d
is aanother
of vector of all 1s,location.
station which means that only
The matrix one station
F contains is allowed
records of thetoprevious
be insidesolution
the separation limitthe
sets x, and of
another
vector = (xT 1 location.
g station − 1)1 hasThe matrix
all of F contains
its values equal records
to one less of the
thanprevious
the number solution sets x, in
of stations andthethe vector
previous
g  (x T 1 sets
solution 1)1 inhasF. all
This ofconstraint
its values forces
equal the to one
BILPless than the number
formulation to yield of stationssolution
a unique in the previous
for each
iteration.
solution sets in F.ZThis
Finally, min isconstraint
the minimum forcescosttheamong the valid solution
BILP formulation to yieldsets found through
a unique solution previous
for each
iterations. Thus, the last constraint, w Tx ≤ Z , reduces the search space such that the BILP only
iteration. Finally, Zmin is the minimum cost among min the valid solution sets found through previous
looks for aThus,
iterations. new solution having a cost
the last constraint, w xthat
T ≤ Zisminless than or
, reduces theequal
search Zmin . such that the BILP only looks
to space
for a new solution having a cost that is less than or equal to Zmin.
Please note that Equation (4) does not have the accuracy requirement as a constraint, because it
cannot be directly incorporated into the formulation. Therefore, the solutions from Equation (4) do
not necessarily satisfy the accuracy requirement. For this case, the coverage analysis method
implements the second step or updated BILP in Equation (5).
Sensors 2018, 18, 3275 9 of 19

Please note that Equation (4) does not have the accuracy requirement as a constraint, because it
cannot be directly incorporated into the formulation. Therefore, the solutions from Equation (4) do not
necessarily satisfy the accuracy requirement. For this case, the coverage analysis method implements
the second step or updated BILP in Equation (5).
n
minZ = ∑ wi xi = wT x
i =1
subject to : Ax ≥ bmod
Cmod x ≤ d (5)
Fmod x ≤ g
wT x ≤ Zmin
xi ∈ {0, 1}
(
bi + 1 if HDOPi > HDOPreq
where bmod,i = .
bi if HDOPi ≤ HDOPreq
This second step of Equation (5) is repeated until a solution set meeting the required HDOP is
found or no solution is found due to the conflicts with the constraints. The impact of the vector bmod is
that the airspace locations with an insufficient HDOP value need visibility to at least one more station
to improve the HDOP value. The matrix Cmod relaxes the separation between stations at small steps
for each iteration, and Fmod is used only for the updated BILP formulation. The BILP may fail to find
any solutions that meet the constraint in the beginning. This usually happens when the minimum
separation in C is too conservative, such that the possible combinations of the valid ground stations do
not exist. In that case, it is necessary to relax the separation constraint between stations to search for
valid solution sets.
The block diagram in Figure 6 shows the proposed overall method in designing an optimal DME
ground station network using Equations (4) and (5). As shown in the process, the BILP does not try
to find a ground network for the airspace in the entire interested region at once because the search
space that BILP has to handle can be typically extremely large. Instead, BILP searches for an optimal
ground network for each airport and check if there are any coverage gaps in the entire airspace after
finishing the search procedure for all of the airports. This approach not only accelerates the search
space by reducing the dimension of the vectors and matrices in Equations (4) and (5), but also finds a
ground network that provides a complete coverage of the entire region. The reason for this is that a
low-altitude airspace with a limited line of sight to the ground DME transponders only exists near an
airport. With the ground network layout satisfying the low altitude, aircraft at high altitudes of over
5000 ft have no problem in meeting the position accuracy requirements since there are usually several
ground transponders in view at high altitude.
For the preparation of the ground network search around an airport, the BILP first generates grids
of the airspace and terrain. Using the coordinates of the grids, the vector x and the matrices A and C
are generated by using a digital terrain map and a Matlab radio propagation analysis tool. The vector
b can be set to a column vector of 2s, which is the minimum number of a ranging source required to
compute a horizontal position. Please note that the matrices A and C and the vector b in the first step
should be conservatively set to avoid a redundant site. While the first step is implemented once, the
second step is iterated multiple times until every airspace grid meets the required positioning accuracy.
At each iteration of the second step, Cmod allows a smaller distance between ground transponders
and the element value of the vector bmod increases by one such that the airspace grid having an
insufficient positioning accuracy would be able to see an additional ground transponder. The process
in Figure 6 tends to result in a minimum number of required ground transponders, and its optimality
was rigorously discussed in [15]. This process is used to generate optimized DME/DME networks for
the selected regions in the next section.
Sensors 2018, 18, 3275 10 of 19
Sensors 2018, 18, x FOR PEER REVIEW 10 of 19

Figure 6. Overall coverage analysis methodology process.

5. Optimized DME/DME
5. Optimized DME/DME Network
Network with
with DME/S
DME/S in
in Two
Two Selected
Selected Regions
Regions
In
In this
this section,
section, the
the optimized
optimized DME/DME
DME/DME network network with
with the
the proposed
proposed DME/SDME/S is is investigated
investigated
using the coverage analysis procedure for the following two selected regions:
using the coverage analysis procedure for the following two selected regions: South Florida South Florida and Southand
Korea [24,25]. The two regions have very different terrain conditions. In South
South Korea [24,25]. The two regions have very different terrain conditions. In South Florida, most Florida, most terrain is
flat, such that an aircraft has no problem in having lines of sight to multiple DMEs.
terrain is flat, such that an aircraft has no problem in having lines of sight to multiple DMEs. However, the terrain
in South Korea
However, consists
the terrain in of numerous
South Korea high mountains
consists that significantly
of numerous high mountains limitthat
the lines of sight between
significantly limit the
aircraft and ground transponders. These cases studies will shed light on appropriate
lines of sight between aircraft and ground transponders. These cases studies will shed light on APNT positioning
performance
appropriate APNT requirements
positioningbasedperformance
on DME/DME with DME/S
requirements in different
based on DME/DME regions.withIn the studies,
DME/S in
aircraft are assumed to be equipped with a three-channel scanning DME for DME/DME
different regions. In the studies, aircraft are assumed to be equipped with a three-channel scanning positioning.
When
DME for more than threepositioning.
DME/DME DME ground transponders
When more thanare in view
three DMEofground
the aircraft, it is also are
transponders assumed
in view that
of
the scanning DMEs can select a set of ground transponders that minimize the
the aircraft, it is also assumed that the scanning DMEs can select a set of ground transponders that HDOP value. For the
evaluation
minimize the of the
HDOPcoverage,
value.the Forairspace was sampled
the evaluation into grids the
of the coverage, 1 km apart inwas
airspace longitudinal
sampled into andgrids
lateral1
directions. The altitudes of the grids were determined from the airspace definition
km apart in longitudinal and lateral directions. The altitudes of the grids were determined from in [2] and the linethe
of
sight information between aircraft and a ground transponder was provided from
airspace definition in [2] and the line of sight information between aircraft and a ground transponder Matlab radio signal
propagation
was provided software. The coverage
from Matlab of the airspace
radio signal propagationwas determined
software. Theif the computed
coverage of aircraft horizontal
the airspace was
determined if the computed aircraft horizontal position accuracy, i.e., the product of HDOP and the
assumed DME ranging accuracy meets a given positioning accuracy requirement.
Sensors 2018, 18, 3275 11 of 19

position accuracy, i.e., the product of HDOP and the assumed DME ranging accuracy meets a given
positioning accuracy requirement.
Sensors 2018, 18, x FOR PEER REVIEW 11 of 19
Sensors 2018, 18, x FOR PEER REVIEW 11 of 19
5.1. Case Study: South Florida
5.1. Case Study: South Florida
5.1. Case Study: South Florida
In theInSouth
the South Florida
Floridaregion,
region, there
thereareareeight
eight international airports
international airports and
and 23 23
DMEDME sites.
sites. The The airspace
airspace
altitude In the
defined South
in Florida
this work region,
is at there
500 ft are
above eight international
ground level airports
(AGL) and
within 23aDME
5
altitude defined in this work is at 500 ft above ground level (AGL) within a 5 statute mile (SM) radius sites.
statute The
mile airspace
(SM) radius
altitude
of anofairport defined
andand
an airport in this
gradually work
gradually is at
increases 500
increasesup ft above
uptoto5500ground level
5500 ftft [2]. (AGL) within
To conservatively
[2]. To conservatively a 5 statute
evaluate
evaluate mile (SM)
thethe radius
DME/DME
DME/DME
of an airport and gradually increases upintothe
coverage
coverage of the of current
the current ground
ground network
network in the5500
South
Southft [2]. To conservatively
Florida
Florida region,
region,thethe evaluate
ranging
ranging the DME/DME
accuracy
accuracyof legacy
of legacy
coverage
DME is oftothe
set current
370.4 m ground
(95%) (Tablenetwork
1). Within the
the ranging
South Florida region,
accuracy and the ranging
current accuracy
ground DME of legacy
network,
DME is set to 370.4 m (95%) (Table 1). With the ranging accuracy and current ground DME network,
DME is set
Figures to 370.4 m (95%) (Table0.3 1). With the ranging accuracy
accuracy and currentgaps,
ground DME network,
Figures 7 and778 and
show 8 1.0
showand1.00.3and nm positioning
nm positioning accuracyaccuracy
coverage coverage respectively.
gaps, respectively. The
The positioning
positioning gap of 1.0 nm is local and could easily be removed by adding a couple of DMEs near The
Figures and 8 show 1.0 and 0.3 nm positioning coverage gaps, respectively. the
gap ofpositioning
1.0 nm is local
gap the
ofand
1.0 could easilyand
nm is local be removed bybeadding a couple of DMEs near the gaps. However,
gaps. However, positioning gap ofcould
0.3 nm easily
is large, removed by adding
and a significant a couple
number of of DMEs
additional near
DMEsthe
the positioning
gaps. However,gap of the 0.3 nm is large,
positioning gap and
of a nm
0.3 significant
is large, number
and a of additional
significant number DMEs
of would DMEs
additional be needed
would be needed to provide 0.3 nm positioning capability.
to provide
would0.3 nm positioning
be needed to providecapability.
0.3 nm positioning capability.

Figure
Figure 7. DME/DME
7. DME/DME positioningaccuracy
positioning accuracy gaps
gaps of
of1.0
1.0nm
nmwith
withlegacy DME
legacy andand
DME the the
current DMEDME
current
Figure 7.
ground DME/DME positioning accuracy gaps of 1.0 nm with legacy DME and the current DME
network.
ground network.
ground network.

Figure 8. DME/DME positioning accuracy gaps of 0.3 nm with legacy DME and the current DME
Figure 8. DME/DME positioning accuracy gaps of 0.3 nm with legacy DME and the current DME
ground network.
Figure
ground 8. DME/DME positioning accuracy gaps of 0.3 nm with legacy DME and the current DME
network.
ground network.
Sensors 2018,
Sensors 2018, 18,
18, 3275
x FOR PEER REVIEW 12 of
12 of 19
19
Sensors 2018, 18, x FOR PEER REVIEW 12 of 19
Figure 9 shows 0.3 nm positioning accuracy coverage gaps when DME/S and the current ground
Figure
DMEFigure
network99 shows
shows 0.3 nm
are used,
0.3 nm positioning
which accuracy
eliminates
positioning coverage
mostcoverage
accuracy gaps
coveragegaps when
gapswhen
of the DME/S
legacyand
DME/S DME
and the current
theexcept
currentforground
some
ground
DME
DME network
smallnetwork
low-altitudeare used,
regions
are used, which
whichnear eliminates
airports. most
eliminates most
Figure coverage
10 shows
coverage gaps of the
theofoptimized
gaps legacy
the legacy ground DME
DME except except
network for some
of DME/S,
for some small
small
whichlow-altitude
provides
low-altitude 0.3regions
regions nm near airports.
nearpositioning
airports. accuracy
Figure Figure
10 shows 10 shows
in the entire
the the optimized
region.
optimized The
ground ground
network
networkinnetwork
Figure ofrequired
10
of DME/S, DME/S,
which
which
provides provides
a total of0.3
15 nm
DME 0.3 nm positioning
sites, with
positioning accuracy
three new
accuracy in the in
sites, the entire
andregion.
entire region.
couldThe
eliminate The
network network
11incurrent in
FigureDME Figure 10 required
sites. Figure
10 required a total 11
of
ashows
15 total ofsites,
DMEthe 15 DME
withsites,
positioning with
new three
accuracy
three new
of coverage
sites, sites,
and could gapsandof could eliminate
92.6 m11
eliminate with DME/S
current 11
DME current
and theDME
sites. current
Figure sites. Figure
ground
11 shows DME 11
the
shows
network.thePlease
positioning positioning
note of
accuracy accuracy
that ofgaps
coverage
the coverage
coverage ofgap
92.6gaps
ofm92.6of m
with92.6 m with
positioning
DME/S andDME/S
accuracyandis
the current the current
almost
ground ground
identical
DME toDME
that
network.
network.
of 0.3 nm
Please Please note
notepositioning that the
accuracy.
that the coverage coverage
gapFigure gap of
of 92.612mshows 92.6 m positioning
the DME/S
positioning ground
accuracy accuracy
network
is almost is almost
for 92.6
identical identical
to m of 0.3that
to
positioning
that nm
of 0.3 nm positioning
accuracy. This
positioning network
accuracy. accuracy.
requires
Figure Figure 12
a totalthe
12 shows 14 ofshows the
DME sites
DME/S DME/S ground
withnetwork
ground four new network
forsites. for 92.6 m positioning
92.6 m positioning accuracy.
accuracy.
This Thisrequires
network networka requires
total 14 ofa total
DME14 of DME
sites sites new
with four withsites.
four new sites.

Figure 9.
Figure 9. DME/DME
DME/DMEpositioning accuracy
positioning gaps
accuracy of 0.3
gaps of nm
0.3 with DME/S
nm with and theand
DME/S current DME ground
the current DME
Figure 9. DME/DME positioning accuracy gaps of 0.3 nm with DME/S and the current DME ground
network.
ground network.
network.

Figure 10. Optimal DME transponder network for 0.3 nm positioning accuracy with DME/S.
Figure 10. Optimal DME transponder network for 0.3 nm positioning accuracy with DME/S.
Figure 10. Optimal DME transponder network for 0.3 nm positioning accuracy with DME/S.
Sensors
Sensors 2018,
2018, 18,
18, x3275
FOR PEER REVIEW 13
13of
of19
19
Sensors 2018, 18, x FOR PEER REVIEW 13 of 19

Figure 11. DME/DME positioning accuracy gaps of 92.6 m with DME/S and the current DME ground
Figure 11. DME/DME positioning accuracy gaps of 92.6 m with DME/S and the current DME
Figure 11. DME/DME positioning accuracy gaps of 92.6 m with DME/S and the current DME ground
network.
ground network.
network.

Figure 12. Optimal DME transponder network for 92.6 m positioning accuracy with DME/S.
Figure 12. Optimal DME transponder network for 92.6 m positioning accuracy with DME/S.
5.2. Case Figure
Study:12. Optimal
South KoreaDME transponder network for 92.6 m positioning accuracy with DME/S.
5.2. Case Study: South Korea
In the
5.2. Case South
Study: Florida
South Koreacase, a DME/DME horizontal positioning accuracy of 1.0 nm or 0.3 nm could
In the South
be achieved usingFlorida
DME/S case,
witha aDME/DME
few new sites.horizontal positioning
The DME/DME accuracy of
positioning 1.0 nm or
accuracy of 0.3
1.0 nm
nm could
is met
be In the
achieved South
using Florida
DME/S case,
with aaDME/DME
few new horizontal
sites. The positioning
DME/DME accuracy
positioning
in most areas with the current network of the legacy DME. However, the difficult terrain conditionsof 1.0
accuracynm or
of 0.3
1.0 nm
nm could
is met in
be
in achieved
most areas using
with DME/S
the with
current a few
network newof
South Korea make DME/DME positioning more complicated. sites.
the The
legacy DME/DME
DME. positioning
However, the accuracy
difficult of
terrain1.0 nm is
conditions met
in South
in most areas13with
Korea
Figure make
shows theDME/DME
current
the network
terrain of the
positioning
elevation legacy
more
profile inDME.
SouthHowever,
complicated.Korea wherethe difficult
elevation terrain
changesconditions
from 0
in South
FigureKorea
13 make
shows DME/DME
the terrain positioning
elevation more
profile incomplicated.
to 1800 m. The airspace in South Korea is similarly defined to that in South Florida, except that0the
South Korea where elevation changes from to
1800 Figure
m. The 13 shows
airspace the
in terrain
South elevation
Korea is profile
similarly in South
defined Korea
to that where
in elevation
South
minimum AGL altitude is 700 ft. Figure 14 shows the coverage gaps of 1.0 nm positioning accuracy changes
Florida, except from
that 0 to
the
1800
minimum m. The airspace
AGL altitude
with the current in
ground is South Korea
700 ft. Figure
network is similarly
14 shows
of the legacy DME defined to
theincoverage that in South
gaps Unlike
South Korea. of 1.0 nm Florida, except
the positioning
South Florida that
accuracy the
region,
minimum
with the AGL
current altitude
ground is 700
network ft. Figure
of the 14
legacy shows
DME the
in coverage
South gaps
Korea. of
Unlike1.0 nm
the
there are significant coverage gaps even for 1.0 nm horizontal positioning accuracy. Figure 15 shows the positioning
South Florida accuracy
region,
with
there the
are current
horizontal ground
significant
positioning network
coverage
accuracy gaps ofeven
the legacy
coverage for 1.0DME
gaps nm0.3
of in South
horizontal
nm using Korea. Unlike
positioning
DME/S and the
the South
accuracy. Florida
Figure
current DME region,
15ground
shows
there
the are significant
horizontal
network, and Figure coverage
positioning
16 shows gapsan even
accuracy for 1.0gaps
coverage
augmented nm horizontal
of 0.3
network nmpositioning
with using DME/S
DME/S accuracy.
and 0.3
to achieve Figure
the 15 shows
current
nm DME
horizontal
the
positioning accuracy, which has a total 38 sites including all of 14 existing DME sites and 24DME
horizontal
ground network, positioning
and Figureaccuracy
16 coverage
shows an gaps
augmentedof 0.3 nm
networkusing DME/S
with DME/S and tothe current
achieve 0.3 nm
new
ground network,
horizontal and accuracy,
positioning Figure 16which shows hasana total
augmented
38 sites network
includingwith
all ofDME/S to achieve
14 existing DME sites 0.3 and
nm
horizontal positioning accuracy, which has a total 38 sites including all of 14 existing DME sites and
Sensors 2018, 18, x FOR PEER REVIEW 14 of 19
Sensors 2018,18,
Sensors2018, 18,3275
x FOR PEER REVIEW 14
14of
of19
19
24 new sites. Figure 17 shows the coverage gap of 92.6 m with DME/S and the current ground DME
24 new sites.
network, whichFigure 17 shows
is much largerthe coverage
than the 0.3 gap
nm of 92.6 m with
coverage gap. DME/S
Becauseand the current
a significant groundofDME
number the
sites. Figure 17 shows the coverage gap of 92.6 m with DME/S and the current ground DME network,
network, which is much larger than the 0.3 nm coverage gap. Because a significant
ground network augmentations are required even for 0.3 nm positioning accuracy, it seems number of the
which is much larger than the 0.3 nm coverage gap. Because a significant number of the ground
ground network
unfeasible to realizeaugmentations are required
DME/DME positioning even for 92.6
that provides 0.3 mnmpositioning
positioning accuracy,
accuracy it entire
in the seems
network augmentations are required even for 0.3 nm positioning accuracy, it seems unfeasible to
unfeasible to realize DME/DME positioning that provides
region. Thus, no augmented ground network for 92.6 m horizontal 92.6 m positioning accuracy in the entire
accuracy was sought.
realize DME/DME positioning that provides 92.6 m positioning accuracy in the entire region. Thus,
region. Thus, no augmented ground network for 92.6 m horizontal positioning accuracy was sought.
no augmented ground network for 92.6 m horizontal positioning accuracy was sought.

Figure 13. Terrain elevation profile in South Korea.

Figure 13. Terrain elevation profile in South Korea.
Figure 13. Terrain elevation profile in South Korea.

Figure 14. DME/DME coverage of 1.0 nm horizontal positioning accuracy in South Korea with legacy
Figure 14. DME/DME
DME performance andcoverage of 1.0
the current DMEnmground
horizontal positioning accuracy in South Korea with legacy
network.
Figure
DME 14. DME/DME
performance and coverage of DME
the current 1.0 nmground
horizontal positioning accuracy in South Korea with legacy
network.
DME performance and the current DME ground network.
Sensors 2018,
Sensors 2018, 18,
18, 3275
x FOR PEER REVIEW 15 of
15 of 19
19
Sensors 2018, 18, x FOR PEER REVIEW 15 of 19

Figure 15. DME/DME coverage of 0.3 nm horizontal positioning accuracy in South Korea with DME/S
Figure 15. DME/DME coverage of 0.3 nm horizontal positioning accuracy in South Korea with DME/S
Figure
and the 15. DME/DME
current coverage
DME ground of 0.3 nm horizontal positioning accuracy in South Korea with DME/S
network.
and the current DME ground network.
and the current DME ground network.

Figure 16. Augmented DME network for 0.3 nm horizontal positioning accuracy with DME/S in
Figure 16. Augmented DME network for 0.3 nm horizontal positioning accuracy with DME/S in South
South Korea.
Figure 16. Augmented DME network for 0.3 nm horizontal positioning accuracy with DME/S in South
Korea.
Korea.
Sensors 2018, 18, 3275 16 of 19
Sensors 2018, 18, x FOR PEER REVIEW 16 of 19

Figure 17.
Figure DME/DMEcoverage
17. DME/DME coverageof
of92.6
92.6m
mhorizontal
horizontalpositioning
positioning accuracy
accuracy in South Korea with
with DME/S
DME/S
and the
and the current
current DME ground network.

6. Discussion
6. Discussion
The coverage analysis and augmented network placement shed light on which types of
The coverage analysis and augmented network placement shed light on which types of
DME/DME operation is appropriate with legacy DME or DME/S. In the South Florida region, a
DME/DME operation is appropriate with legacy DME or DME/S. In the South Florida region, a
DME/DME horizontal positioning accuracy of 1.0 nm is currently provided in most areas with legacy
DME/DME horizontal positioning accuracy of 1.0 nm is currently provided in most areas with legacy
DME performance. However, Figure 8 indicated that non-minimal DME augmentation was needed to
DME performance. However, Figure 8 indicated that non-minimal DME augmentation was needed
provide 0.3 nm horizontal positioning accuracy with legacy DME in this region. If DME/S is used,
to provide 0.3 nm horizontal positioning accuracy with legacy DME in this region. If DME/S is used,
a ground network in Figure 12 providing 92.6 m horizontal positioning accuracy seems to be more
a ground network in Figure 12 providing 92.6 m horizontal positioning accuracy seems to be more
appropriate than the ground network for 0.3 nm horizontal positioning accuracy shown in Figure 10,
appropriate than the ground network for 0.3 nm horizontal positioning accuracy shown in Figure 10,
because one additional new DME site is needed to enable 92.6 m horizontal positioning accuracy,
because one additional new DME site is needed to enable 92.6 m horizontal positioning accuracy,
which is sufficient to replace current radar-based aircraft surveillance systems such as SSR.
which is sufficient to replace current radar-based aircraft surveillance systems such as SSR.
In South Korea, the difficult terrain condition makes it hard to achieve a 1.0 nm DME/DME
In South Korea, the difficult terrain condition makes it hard to achieve a 1.0 nm DME/DME
horizontal positioning accuracy with the legacy DME performance. Most coverage gaps exist in
horizontal positioning accuracy with the legacy DME performance. Most coverage gaps exist in low-
low-altitude regions which have a limited line of sight to DMEs. The difficult terrain conditions
altitude regions which have a limited line of sight to DMEs. The difficult terrain conditions also
also require a significant augmentation of the ground network, even with DME/S, to achieve 0.3 nm
require a significant augmentation of the ground network, even with DME/S, to achieve 0.3 nm
horizontal positioning accuracy. The positioning gap of 92.6 m is much larger than that of 0.3 nm, as
horizontal positioning accuracy. The positioning gap of 92.6 m is much larger than that of 0.3 nm, as
shown in Figure 17; therefore, the appropriate positioning accuracy of alternative DME/DME would
shown in Figure 17; therefore, the appropriate positioning accuracy of alternative DME/DME would
be 0.3 nm or less. To relax a required ground network augmentation in South Korea, a scanning DME
be 0.3 nm or less. To relax a required ground network augmentation in South Korea, a scanning DME
of five channels can be used instead of three-channel scanning DME, since two additional ranging
of five channels can be used instead of three-channel scanning DME, since two additional ranging
sources would help to lower HDOP values. Figure 18 shows the augmented ground DME network for
sources would help to lower HDOP values. Figure 18 shows the augmented ground DME network
0.3 nm
for 0.3 horizontal positioning
nm horizontal accuracy
positioning with five
accuracy withchannels. The network
five channels. consists consists
The network of a totalof
32 anetworks,
total 32
with 11 current sites and 21 new sites. This requires six fewer ground transponders
networks, with 11 current sites and 21 new sites. This requires six fewer ground transponders than than three-channel
scanning DMEs.
three-channel Another
scanning way to
DMEs. relax the
Another wayground network
to relax augmentation
the ground is to allocate aisDME/DME
network augmentation to allocate
a DME/DME service area by reflecting heavy air traffic regions as shown in example,
service area by reflecting heavy air traffic regions as shown in Figure 19. For Figure 19.an augmented
For example,
DME network can be placed to provide 0.3 nm positioning accuracy in the shaded
an augmented DME network can be placed to provide 0.3 nm positioning accuracy in the shaded airspace only, which
will be further
airspace investigated
only, which will bein the author’s
further futurein
investigated research.
the author’s future research.
Sensors 2018, 18, 3275 17 of 19
Sensors 2018, 18, x FOR PEER REVIEW 17 of 19
Sensors 2018, 18, x FOR PEER REVIEW 17 of 19

Figure Augmented
18.18.
Figure AugmentedDME
DMEnetwork
networkfor RNAV
forRNAV 0.3 nm
RNAV0.3 nm with
with DME/S
DME/S(five-channel
DME/S (five-channelscanning
scanning DME)
DME) in in
South Korea.
South Korea.

Figure
Figure 19.Heavy
19. Heavyair
airtraffic
trafficroutes
routes in
in South
South Korea
Korea are
are represented
representedasasshaded
shadedregions.
regions.

Figure 19. Heavy air traffic routes in South Korea are represented as shaded regions.
Sensors 2018, 18, 3275 18 of 19

Funding: This research was supported by the Basic Science Research Program through the National Research
Foundation of Korea (NRF), funded by the Ministry of Education (NRF-2017R1D1A1B03028146). This research
was also supported by a grant (18CTAP-C143427-01) from the Technology Advancement Research Program,
funded by the Ministry of Land, Infrastructure, and Transport of the Korean government.
Conflicts of Interest: The author declares no conflict of interest.

References
1. Faria, L.A.; Silvestere, C.A.M.; Correia, M.A.F. GPS-dependent systems: Vulnerabilities to electromagnetic
attacks. J. Aerosp. Technol. Manag. 2016, 8, 423–430. [CrossRef]
2. Narins, M.; Eldredge, L.; Enge, P.; Harrison, M.; Kenagy, R.; Lo, S. Alternative position, navigation,
and timing: The need for robust radionavigation. In Global Navigation Satellite Systems: Report of a Joint
Workshop of the National Academy of Engineering and the Chinese Academy of Engineering; Royal Institute of
Navigation—NAV 10; The National Academies Press: Washington, DC, USA, 2012.
3. Lo, S.; Enge, P.; Niles, F.; Loh, R.; Eldredge, L. Preliminary assessment of alternative navigation means for
civil aviation. In Proceedings of the 2010 ION ITM, San Diego, CA, USA, 25–27 January 2010.
4. Kelly, R.J.; Cusick, D.R. Distance measuring equipment and its evolving role in aviation. In Advances
in Electronics and Electron Physics; Hawkes, P.W., Ed.; Academic Press, Inc.: Cambridge, MA, USA, 1986;
Volume 68, pp. 73–75.
5. Lilley, R.; Robert, E. DME/DME for Alternative Position, Navigation, and Timing (APNT); APNT White Paper;
Federal Aviation Administration: Washington, DC, USA, 2012.
6. Pelgrum, W.; Li, K.; Smearcheck, M.; Van Graas, F. eDME architecture development and flight-test evaluation.
In Proceedings of the 25th International Technical Meeting of The Satellite Division of the Institute of
Navigation (ION GNSS 2012), Nashville, TN, USA, 17–21 September 2012; pp. 812–825.
7. Kim, E. Improving DME performance for APNT using alternative pulse and multipath mitigation. IEEE Trans.
Aerosp. Electron. Syst. 2017, 53, 877–887. [CrossRef]
8. Kim, E.; Seo, J. SFOL pulse, a high accuracy DME pulse for alternative aircraft position and navigation.
Sensors 2017, 17, 2183. [CrossRef] [PubMed]
9. Kim, O.-J.; Kim, C.; Song, J.; Yun, H.; Kim, D.; Kee, C.; Lee, T. A new concept of APNT: MOSAIC/DME 3-D
positioning with a single DME station. In Proceedings of the 2012 International Technical Meeting of the
Institute of Navigation, Newport Beach, CA, USA, 1–30 January 2012; pp. 142–150.
10. Thiasiriphet, T.; Shutin, D.; Schnekenburger, N. Application of bayesian filtering for multipath in
LDACS1-based APNT applications. In Proceedings of the 27th International Technical Meeting of the
Satellite Division of the Institute of Navigation (ION GNSS+ 2014), Tampa, FL, USA, 8–12 September 2014;
pp. 3065–3075.
11. Helfrick, A. Principles of Avionics, 6th ed.; Avionics Communications Inc.: Leesburg, VA, USA, 2010; pp. 55–56.
12. FAA. Performance-Based Navigation (PBN) National Airspace System (NAS) Navigation Strategy 2016;
U.S. Department of Transportation: Washington, DC, USA, July 2016.
13. FAA. Performance Specification Distance Measuring Equipment (DME); FAA-E-2996; U.S. Department of
Transportation: Washington, DC, USA, 2008; pp. 7–16.
14. RTCA Special Committe-149. Minimum Operation Performance Standards (MOPS) for Airborne Distance
Measuring Equipment (DME) Operating within the Radio Frequency Range of 960-1215 MHz; RTCA DO-189;
RTCA Inc.: Washington, DC, USA, 1985; p. 18.
15. Kim, E. Investigation of APNT optimized DME/DME network using current state-of-art DMEs: Ground station
network, accuracy, and capacity. In Proceedings of the IEEE/ION PLANS 2012, Myrtle Beach, SC, USA,
24–26 April 2012.
16. Kim, E. A Coverage analysis methodology for APNT pseudolite ground network. In Proceedings of the 2011
ION GNSS, Portland, OR, USA, 20–23 September 2011.
17. Farahani, R.Z.; Hekmatfar, M. Facility Location Concepts, Models, Algorithms and Case Studies; Physica-Verlag:
Heidelberg, Germany, 2009; pp. 1–4.
18. Dasqupta, S.; Papadimitriou, C.; Vazirani, U. Algorithms, 1st ed.; McGraw-Hill Education: New York, NY, USA,
2006; pp. 247–278.
Sensors 2018, 18, 3275 19 of 19

19. Roa, J.O.; Jiménez, A.R.; Seco, F.; Prieto, J.C.; Ealo, J. Optimal placement of sensors for trilateration:
Regular lattices vs meta-heuristic solutions. In Computer Aided Systems Theory—EUROCAST 2007; Springer:
Berlin, Germany, 2007; pp. 780–787.
20. Ray, P.K.; Maha, A. A genetic algorithm-based approach to calculate the optimal configuration of ultrasonic
sensors in a 3d position estimation system. Robot. Auton. Syst. 2012, 41, 165–177. [CrossRef]
21. Leune, T.; Wehs, T.; Janssen, M.; Koch, C.; Von Cölln, G. Optimization of wireless locating in complex
environments by placement of anchor nodes with evolutionary algorithms. In Proceedings of the
2013 IEEE 18th Conference on Emerging Technologies & Factory Automation (ETFA), Cagliari, Italy,
10–13 September 2013; IEEE: Piscataway, NJ, USA, 2013; pp. 1–6.
22. Monica, S.; Ferrari, G. Optimized anchors placement: An analytical approach in uwb-based tdoa localization.
In Proceedings of the 2013 9th International Wireless Communications and Mobile Computing Conference
(IWCMC), Sardinia, Italy, 1–5 July 2013; IEEE: Piscataway, NJ, USA, 2013; pp. 982–987.
23. Monica, S.; Ferrari, G. Uwb-based localization in large indoor scenarios: Optimized placement of anchor
nodes. IEEE Trans. Aerosp. Electron. Syst. 2015, 51, 987–999. [CrossRef]
24. Kim, E. Benefit analysis of a GA-based DME/N pulse on PBN. In Proceedings of the 2017 Integrated
Communications, Navigation and Surveillance Conference (ICNS), Herndon, VA, USA, 18–20 April 2017;
pp. 4B1-1–4B1-7. [CrossRef]
25. Kim, E.; Cho, H. High accuracy SFOL DME/N pulse and its recommended implementation in South Korea
for alternative aircraft navigation. In Proceedings of the 2018 IEEE/ION Position, Location and Navigation
Symposium (PLANS), Monterey, CA, USA, 23–26 April 2018; pp. 1039–1041. [CrossRef]

© 2018 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).