sensors

Article
Announcement Signals and Automatic Braking Using Virtual
Balises in Railway Transport Systems
Enrique Santiso , José A. Jiménez , Manuel Mazo, Cristina Losada and Francisco J. Rodríguez *

Department of Electronics, University of Alcalá, 28805 Alcalá de Henares, Madrid, Spain;

[email protected] (E.S.); [email protected] (J.A.J.); [email protected] (M.M.);
[email protected] (C.L.)
* Correspondence: [email protected]; Tel.: +34-91-885-6555

Abstract: In rail transport, various automatic protection systems are available to ensure the safe
operation of trains and to facilitate automation and optimization tasks. For this purpose, a set of
physical balises is used, which are placed at fixed points along the railway track. Based on the
information provided by these balises, different information is displayed to the driver, and control
actions are generated. The use of physical balises located at fixed points does not allow for automatic
protection actions on sections of track where they are not installed. This is a major drawback as
in many cases, temporary automatic protection actions are necessary on sections of the railway
line without balises due to various circumstances (work on the track, accidents, etc.). To solve this
problem, this paper presents a solution called announcement signals and automatic braking using
virtual balises (ASAB-VB). This proposal allows the incorporation of virtual balises at points on the
track where it is necessary to temporarily perform automatic protection actions. For this purpose,
the ASAB-VB system allows obtaining the train position in real-time and storing a digital map of the



track that will be made by each train. This digital map includes geographic information about the


balises (both physical and virtual ones) located on the track. At the same time, the train position is
Citation: Santiso, E.; Jiménez, J.A.;
obtained by merging the information provided by a GNSS, an odometer, and an inertial system (gyro
Mazo, M.; Losada, C.; Rodríguez, F.J.
and accelerometers).
Announcement Signals and
Automatic Braking Using Virtual
Balises in Railway Transport Systems.
Keywords: railway transport; railway transport safety; virtual balise; digital map; automatic train
Sensors 2022, 22, 1943. https:// protection systems; ASAB-VB
doi.org/10.3390/s22051943

Academic Editors: Hossam A.

Gabbar and Justyna Tomaszewska
1. Introduction
Received: 7 January 2022 Railway transport is becoming increasingly important worldwide, mainly because
Accepted: 24 February 2022 it is one of the safest modes of transport when compared to other means of passenger
Published: 2 March 2022 transportation. For example, according to the Worldwide Railway Organisation (UIC), the
Publisher’s Note: MDPI stays neutral accident rate is almost 200 times higher for road than for railway transport. This, together
with regard to jurisdictional claims in with the contribution of railways to sustainable transport policies, justifies the promotion
published maps and institutional affil- of this mode of transport in most countries. All this is also favored by the incorporation of
iations. new technologies, which allow increasing the safety performance of this means of transport.
Additionally, thanks to the incorporation of these technologies, train crashes are becoming
rarer in the European Union. Recently published Eurostat data confirm this positive trend.
In 2019, there were 516 significant railway accidents with 802 fatalities in the EU, down from
Copyright: © 2022 by the authors. 853 deaths in 2018 [1]. According to the “Report on Railway Safety and Interoperability
Licensee MDPI, Basel, Switzerland.
in the EU-2020” [2], the fatality risk for a train passenger is a quarter of the risk for a
This article is an open access article
bus/coach passenger. The use of individual transport means, such as passenger car, carries
distributed under the terms and
a substantially higher fatality risk: car occupants are almost 50 times more likely to die
conditions of the Creative Commons
than train passengers traveling the same distance. The fatality risk for an average train
Attribution (CC BY) license (https://
passenger is now about 0.05 fatalities per billion passenger-kilometers, making it the safest
creativecommons.org/licenses/by/
4.0/).
mode of land transport in the EU.

Sensors 2022, 22, 1943. https://doi.org/10.3390/s22051943 https://www.mdpi.com/journal/sensors

Sensors 2022, 22, 1943 2 of 17

In the field of railway transport, automatic train control (ATC) and automatic train
protection (ATP) systems enable the safe operation of trains, as well as the automation and
optimization of their functioning. Train protection systems perform a variety of actions,
such as preventing the train from running a red light, controlling train speed, warning
the train driver in advance, or activating emergency brakes in case of danger. In Europe,
each country has developed its automatic protection system, the most significant difference
being the type of connectivity they have between the track and rolling stock (induction,
radio, etc.) as well as the speed control they perform (punctual or continuous). However,
the establishment of trans-European rail networks has led to the development of the Euro-
pean Rail Traffic Management System (ERTMS), which ensures interoperability within the
European rail network. ERTMS consists of a train control system (European Train Control
System—ETCS) and a communications system (GSM Railways—GSM-R) to carry out a
continuous transmission via-train, each time the train passes over a Eurobalise (balises
installed on the track). Within the ERTMS technology, there are three levels of complexity
in its operation, and all of them provide continuous on-board supervision. In Spain, differ-
ent automatic train protection systems coexist, with the digital ASFA (Automatic Signal
Announcement and Braking) being one of the most widely used due to its characteristics.
Most automatic train protection systems (ERTMS, ASFA, etc.) use physical balises
(PB) installed at known geographical coordinates along railway tracks, together with a
detector (receiver) system onboard the train. When the train passes the balises, the onboard
system recognizes their status and displays information in the cabin on the appearance
of the side signals and the maximum speed of the section to facilitate driving. Moreover,
it can control train speed and driver actions, activating emergency braking in the event
of detecting abnormal driving situations. As an example, in Spain, the digital ASFA
system uses passive physical balises installed on the track, along the railway lines, and a
detector system (receiver) onboard the trains. When the train passes over the balises, the
receiver system recognizes their status by means of a magnetic coupling balise-receiver and,
from this information, obtains the speed restrictions on the track sections following the
balise location points. The onboard system includes in its receiving part an oscillator that
oscillates at a certain frequency in the absence of balise detection (free oscillation frequency).
When the train passes over the balises located on the track, the oscillator changes its free
oscillation frequency to the resonance frequency forced by the balise circuit. The digital
ASFA is designed to detect up to nine different frequencies, in the range of 55–115 kHz.
Each frequency corresponds to what is known as the “balise aspect” (in the case of the
digital ASFA, up to 9 “balise aspects” are considered: L1—stop announcement/pre-stop
announcement/etc., L2—conditional free lane, L3—free lane/protected level crossing, etc.,
and for each one a braking curve profile is set. In addition, there is vertical signalling which
also informs about the status of the balise.
Regarding the train positioning, several proposals use GNSS and inertial systems.
Thus, in [3] a study was carried out on the reliability of low-cost GNSS receivers for
estimating the position of trains. The authors of [4] propose a data fusion scheme using
GNSS and odometry to locate trains and detect if the integrity of a railway convoy is
accidentally broken (some wagon is uncoupled from the convoy). Whereas [5] describes
a train location system on low traffic lines that is based on GNSS and in which the train
communicates with a central system, via radio, to receive running orders based on its
position. Systems for accurate train location based on the fusion of information provided
by GNSS, odometers and inertial systems are presented in [6,7]. An integrated INS/GNSS
train positioning system with odometric data is proposed in [8]. Finally, a detailed analysis
of the most relevant works related to data fusion between GNSS and inertial system can be
found in [9].
About the use of PBs as milestones to correct the train position each time one of
them is read, there have also been works carried out using digital maps to improve the
accuracy and reliability of the information provided by satellites. In this context, the authors
in [10] point out the relevance of digital maps within the map-matching process required
Sensors 2022, 22, 1943 3 of 17

in train positioning systems based, in this case, on GALILEO. The authors of this work
highlight the importance of the digital map in positioning systems in safety-critical railway
applications. In [11], positioning methodology for train navigation applications based in a
satellite system called BDS (BeiDou navigation satellite system), inertial system, odometer,
and map-matching is proposed to solve the problem of positioning when satellite system
outages occur when trains pass through signal obstructed areas, such as under bridges,
inside tunnels, and through deep valleys.
The virtual balise (VB) concept arises in the context of ERTMS to obtain an improved
ERTMS. The virtual balise concept consists of replacing a physical balise by a virtual
analogue, thus reducing the number of physical balises on the track and resulting in cost
savings [12]. The principle of operation is based on storing the list and coordinates of virtual
balises and using them together with GNSS positioning information to know the location of
the train [13]. The VB concept has been successfully investigated in many research projects,
e.g., ESA3InSat [14], GSA ERSAT-EAV [15], GSA RHINOS [16], and GSA STARS [17]. The
ASAB-VB system is intended for conventional lines with low traffic density where ERTMS
is not available. Therefore, the concept of VB is different in this application, where VBs are
used to facilitate the incorporation of temporary speed limits (TSL) on the railway track.
Essentially, all the previous works are based on how to solve the problem of knowing
the exact position of the train as reliably as possible, using different sensory systems (GNSS,
odometric, inertial systems, radar, etc.) and, in some cases, including digital maps. This
paper presents a novel system called announcement signals and automatic braking using
virtual balises (ASAB-VB) that can obtain the position of the train in real-time, as well
as store the digital map including the coordinates of all the balises detected along the
route. The train position and the digital map are used to generate virtual balises when it is
necessary to temporarily perform automatic protection actions. The main contributions
of the proposed system lie in (i) the fact that it allows the incorporation of virtual balises
(VB) that complement and/or replace the PBs and (ii) the incorporation of virtual marks
(VM) used to correct the error accumulated by the odometric system. It is based on the
train position obtained by the fusion of GNSS information with odometric and inertial
systems. Additionally, the system may also be used to obtain information on the status
of the physical balises (operational balises, absence of balises, damaged ones) with the
consequent saving in maintenance costs. There are no previous works proposing the use of
VBs and VMs for the purposes indicated in this paper.
The rest of this paper is organized as follows: Section 2 gives an overview of the
proposed solution, Section 3 shows the architecture of the ASAB-VB system, Section 4
details the positioning algorithm, Section 5 presents the main results obtained during the
testing phase and, finally, Section 6 presents the conclusions.

2. General Overview of the Proposal

As explained above, most current automatic train protection systems use physical
balises located at fixed points along the railway track. This means that it is not possible
to carry out automatic protection actions on sections of the track where these PBs do not
exist. This is a major drawback, as in many cases it is necessary to carry out temporary
automatic protection actions, due to various circumstances (maintenance and repair work
on the track, accidents, etc.) on sections of the railway track where there are no physical
balises. This is one of the main reasons that have led the authors of this work to develop a
new system that complements the current automatic protection systems and incorporates
improvements from the point of view of both safety and flexibility. Among the advantages
of this proposal, called Announcement Signals and Automatic Braking using Virtual Balises
(ASAB-VB), the following are worth mentioning:
a. Incorporation of the digital map with VBs based on the train position obtained by the
fusion of GNSS information with odometric and inertial systems. These VBs either
complement, replace, or both, the physical ones.
Sensors 2022, 22, x FOR PEER REVIEW 4 of 17

Sensors 2022, 22, 1943 4 of 17

a. Incorporation of the digital map with VBs based on the train position obtained by the
fusion of GNSS information with odometric and inertial systems. These VBs either
b. complement, replace,
Facilitating the or both, the
incorporation physical ones.
of temporary speed limits (TSL) on the railway track
b. by makingthe
Facilitating useincorporation
of these VBs. ofThese TSLs should
temporary be established
speed limits (TSL) onwhen unexpected
the railway track
situations
by making useoccur
of on theVBs.
these trackThese
(automatic protection
TSLs should actions): maintenance,
be established accidents,
when unexpected sit-
etc. occur on the track (automatic protection actions): maintenance, accidents, etc.
uations
c. Obtaininginformation
Obtaining informationon onthe
thestatus
statusof
ofthe
thePBs
PBs(operational
(operational balises,
balises, absent
absent balises, or
damaged ones) with consequent savings in maintenance
damaged ones) with consequent savings in maintenance costs costs
d. UseofofVMs
Use VMstotocorrect
correctthe
theerror
erroraccumulated
accumulatedby bythe
thepositioning
positioningsystem.
system.
To achieve
To achieveall allthese
thesefunctionalities,
functionalities,the theASAB-VB
ASAB-VB system
systemrequires an an
requires infrastructure
infrastructureon
the the
on ground
groundand and
on-board railway
on-board units,units,
railway to know the position
to know of the train
the position within
of the train the railway
within the
line at any
railway linetime, as well
at any time,as
asthe
welllocation of the PBs
as the location of on
thethe
PBstrack.
on the track.
The ASAB-VB
The ASAB-VBsystem system(patent
(patentnumber
number ESES 2418929
2418929 A1,A1,
[18])[18]) has been
has been designed
designed to
to com-
complement the digital ASFA, although it could be a complement to
plement the digital ASFA, although it could be a complement to other automatic trainother automatic train
protection
protection systems,
systems,such suchasasthe
theERTMS
ERTMS[19].[19].Under
Underthis premise,
this premise, Figure
Figure1 shows
1 showsthethe
general
gen-
configuration of the proposed ASAB-VB
eral configuration of the proposed ASAB-VB system.system.

Figure 1. General configuration of the ASAB-VB system.

Figure 1. General configuration of the ASAB-VB system.

The
The ASAB-VB
ASAB-VBsystem systemincludes
includesthe thefollowing
following elements
elementsin each railway
in each tracktrack
railway (see Fig-
(see
ure 1):
Figure 1):
•• Physical balises
Physical balises (PBs):
(PBs): these
these are
are the ones used
the ones used inin the
the digital
digital ASFA
ASFA system.
system. These
These
balises are read by the train as it passes over them. Depending
balises are read by the train as it passes over them. Depending on the type of balise on the type of balise
(so called “balise appearance”) they have a certain effect on the
(so called “balise appearance”) they have a certain effect on the signalling and thus signalling and thus
on the train speed limitation. It should be noted that the external
on the train speed limitation. It should be noted that the external appearance of the appearance of the
PBs is
PBs is the
the same
sameregardless
regardlessofoftheir theirtypetypeoror“balise
“baliseappearance”.
appearance”. TheThetrain’s balise
train’s de-
balise
tection system
detection systemconsists
consistsof aofresonant
a resonant circuit, that that
circuit, oscillates at a certain
oscillates frequency
at a certain (per-
frequency
manent frequency,
(permanent frequency,PF).PF).
When Whenthe thedetector passes
detector overover
passes a balise, it oscillates
a balise, at the
it oscillates at
resonant frequency of the circuit set in the beacon, to which it is
the resonant frequency of the circuit set in the beacon, to which it is tuned by induc- tuned by inductive
coupling,
tive receiving
coupling, the corresponding
receiving the corresponding “balise appearance”.
“balise FigureFigure
appearance”. 2 shows the exter-
2 shows the
nal appearance
external of a PB
appearance of ainstalled on a on
PB installed track.
a track.
• Virtual balises (VBs): are balises that are established virtually at certain kilometer
points where the passing of the train through them also has a certain effect on sig-
nalling and thus on the train speed. The need for virtual balises may be due to the
requirements to establish speed limits at points on the track where no physical balises
exist, or to replace damaged or stolen physical balises.
• Virtual marks (VMs): only serve as a complement to the train positioning system.
Specifically, VMs are used to correct the error accumulated by the positioning system,
as described below. So, they do not affect signalling.
Sensors 2022,22,
Sensors2022, 22,1943
x FOR PEER REVIEW 55 of
of 17
17

Sensors 2022, 22, x FOR PEER REVIEW 6 of 17

2. Generation of virtual balises (VBs): when the kilometer point corresponding to a vir-
tual balise (KPVB) stored in the TBMPB is reached, the VB is generated.
3. Detected track balises map (DTBM): during the route, a map of physical balises called
a detected track balises map (DTBM) is generated, in which the PBs detected during
the route are registered, their GNSS coordinates, kilometric point, and the time while
their detection was maintained called time of balise (tB). This time is indicative of the
quality of the detection of the balise. At the end of the route, the DTBM is transferred
to the control center for analysis.
The information provided by the positioning system, together with the track balises
map (TBM), makes it possible to know not only whether or not the physical track balises
are detected correctly or whether they are at the correct kilometric points (KPT), but also
whether there are undetected balises, their detection quality, etc. For this purpose, once
the route
Figure
Figure is completed
2.2.General
General and the
configuration
configurationof DTBM
ofthe
the has been
ASAB-VB
ASAB-VB transferred to the ground system, the in-
system.
system.
formation associated with the physical balises contained in the DTBM is matched with the
The general flow chart of the train.
Announcement Signals and Automatic Braking using
•TBM,Virtual
previously loaded
balises onto
(VBs): arethe
balises that are established virtually at certain kilometer
Virtual Balises system (ASAB-VB) is shown in Figure 3, where two different stages are
points where the passing of the train through them also has a certain effect on signal-
contemplated: before the start of the route (static train), and during the route.
ling and thus on the train speed. The need for virtual balises may be due to the re-
quirements to establish speed limits at points on the track where no physical balises
exist, or to replace damaged or stolen physical balises.
• Virtual marks (VMs): only serve as a complement to the train positioning system.
Specifically, VMs are used to correct the error accumulated by the positioning sys-
tem, as described below. So, they do not affect signalling.
The general flow chart of the Announcement Signals and Automatic Braking using
Virtual Balises system (ASAB-VB) is shown in Figure 3, where two different stages are
contemplated: before the start of the route (static train), and during the route.
As can be seen in Figure 3 before the train starts a certain route, the following two
actions are performed:
1. Obtaining the track on which the train is located: before starting the route, it is nec-
essary to determine the track on which the train is located. For this purpose, a posi-
tioning system based on differential global navigation satellite system (DGNSS) is
used.
2. Loading the track balise map (TBM) in the on-board unit: before the start of a route,
the track balise map (TBM) is stored in the on-board unit, which includes geographic
information on the balises (PB and VB) and virtual marks (VM). The TBM includes
the following information about each balise (PB and VB): the track on which they are
located, the kilometer point of the balise position on the track (KPB and KPVB), and
its appearance (effect on signalling). The information included in the TBM on the
VMs is only the GNSS coordinates (latitude and longitude) and kilometer point of
the virtual mark position on the track (KPVM). The TBM transfer from the geographic
information system (GIS) to the train can be done via radio (3-4-5G/Wimax), or can
be loaded locally via USB if communications are not available. If necessary due to
any unforeseen situation (maintenance requirement, accident on the track, etc.), the
transferred file (TBM) can be modified during the route if there is radio communica-
tion with a ground station (e.g., at an intermediate station on the route).
Once the route has started, the three most relevant activities are:
1. Positioning: the position of the train on the track is defined by the track number on
which it is located (TN) and the kilometric point of the train position on the track
Figure3.3.The
Figure Thegeneral
generalprocedure
procedureof
ofthe proposed
theproposed ASAB-VB
ASAB-VB system.
system.
(KPT). To have full availability of the train position, these data are continuously up-
dated by merging
3. Architecture the data obtained
of the ASAB-VB System by the odometric and inertial system. The error
accumulated by this estimation is reduced each time a PB is detected, or when the
To achieve the objectives, the ASAB-VB system requires an infrastructure on the
GNSS estimated position matches that of a KPB.
ground and onboard railway units. Figure 4 shows the most important modules of track-
side and onboard subsystems.
Sensors 2022, 22, 1943 6 of 17

As can be seen in Figure 3 before the train starts a certain route, the following two
actions are performed:
1. Obtaining the track on which the train is located: before starting the route, it is
necessary to determine the track on which the train is located. For this purpose, a
positioning system based on differential global navigation satellite system (DGNSS)
is used.
2. Loading the track balise map (TBM) in the on-board unit: before the start of a route,
the track balise map (TBM) is stored in the on-board unit, which includes geographic
information on the balises (PB and VB) and virtual marks (VM). The TBM includes
the following information about each balise (PB and VB): the track on which they are
located, the kilometer point of the balise position on the track (KPB and KPVB), and
its appearance (effect on signalling). The information included in the TBM on the
VMs is only the GNSS coordinates (latitude and longitude) and kilometer point of the
virtual mark position on the track (KPVM). The TBM transfer from the geographic
information system (GIS) to the train can be done via radio (3-4-5G/Wimax), or can
be loaded locally via USB if communications are not available. If necessary due to
any unforeseen situation (maintenance requirement, accident on the track, etc.), the
transferred file (TBM) can be modified during the route if there is radio communication
with a ground station (e.g., at an intermediate station on the route).
Once the route has started, the three most relevant activities are:
1. Positioning: the position of the train on the track is defined by the track number on
which it is located (TN) and the kilometric point of the train position on the track
(KPT). To have full availability of the train position, these data are continuously
updated by merging the data obtained by the odometric and inertial system. The
error accumulated by this estimation is reduced each time a PB is detected, or when
the GNSS estimated position matches that of a KPB.
2. Generation of virtual balises (VBs): when the kilometer point corresponding to a
virtual balise (KPVB) stored in the TBM is reached, the VB is generated.
3. Detected track balises map (DTBM): during the route, a map of physical balises called
a detected track balises map (DTBM) is generated, in which the PBs detected during
the route are registered, their GNSS coordinates, kilometric point, and the time while
their detection was maintained called time of balise (tB ). This time is indicative of the
quality of the detection of the balise. At the end of the route, the DTBM is transferred
to the control center for analysis.
The information provided by the positioning system, together with the track balises
map (TBM), makes it possible to know not only whether or not the physical track balises
are detected correctly or whether they are at the correct kilometric points (KPT), but also
whether there are undetected balises, their detection quality, etc. For this purpose, once
the route is completed and the DTBM has been transferred to the ground system, the
information associated with the physical balises contained in the DTBM is matched with
the TBM, previously loaded onto the train.

3. Architecture of the ASAB-VB System

To achieve the objectives, the ASAB-VB system requires an infrastructure on the
ground and onboard railway units. Figure 4 shows the most important modules of track-
side and onboard subsystems.
Sensors 2022,22,
Sensors2022, 22,1943
x FOR PEER REVIEW 77 of
of1717

Figure 4. The architecture of the ASAB-VB system.

Figure 4. The architecture of the ASAB-VB system.

Theground
The groundinfrastructure
infrastructureconsists
consistsofofthethefollowing
followingelements:
elements:
•• Geographicinformation
Geographic information system
system (GIS):
(GIS): generates
generates the map
the map with with the balises
the balises of eachofroute
each
routebalise
(track (trackmap,
baliseTBM).
map, TBM).
•• Control,analysis,
Control, analysis, and
and registration
registration unit:
unit: in charge
in charge of managing
of managing communication
communication withwith
the
the train,
train, recording
recording incidents
incidents detected
detected alongalong the route,
the route, and updating
and updating TBMsTBMs dynami-
dynamically
incally
thosein cases
thosein cases in which,
which, once aonce
routea has
route has started,
started, it is necessary
it is necessary to include
to include new VBs.new
VBs. It is also responsible for the comparison between the TBM
It is also responsible for the comparison between the TBM and the detected track and the detected track
balisesmap
balises map(DTBM),
(DTBM), which
which contains
contains thethe balises
balisesdetected
detectedduring
duringa aroute
route and
andtheir lo-
their
cation. This
location. Thiscomparison
comparison makes
makesit possible
it possibleto obtain the possible
to obtain incidences
the possible of theofroute:
incidences the
absence
route: of balises
absence at the at
of balises points defined
the points in the in
defined TBM,
the time
TBM,oftimedetection of balises,
of detection balises
of balises,
location
balises errors, and
location balises
errors, anddetected on the route
balises detected androute
on the not contemplated in the TBM.in
and not contemplated
• the TBM.allows differential positioning of the train. In the case described in this paper,
GNSS:
• GNSS:
beforeallows
startingdifferential positioning
a certain route, the trackof the train. In
on which thethe case
train is described
located must in this paper,
be known,
before
and for starting
this, itaiscertain route,
necessary to the
usetrack on whichGNSS.
a differential the train is located must be known,
• and for this, it is necessary
Communications module:tomanages
use a differential GNSS. with the train. In this case, it is
communications
• Communications
based on 3-4-5G/Wimax module:communication.
manages communications with the train. In this case, it is
based on 3-4-5G/Wimax communication.
The infrastructure onboard mobile unit consists of the following elements:
• The infrastructure
Control onboard
unit: manages mobile
all the unit consists
operations of the of the following
elements elements: system in-
of the ASAB-VB
• Control
stalled unit:
onboardmanages all theunits.
the mobile operations of theassuming
Specifically, elements an of the ASAB-VB
onboard system
digital ASFA
installed onboard the
system, its functions are:mobile units. Specifically, assuming an onboard digital ASFA
system, its functions are:
− Manage communications between the elements of the mobile unit, as well as
− Manage
betweencommunications
the ASAB-VB and between the elements
ASFA digital systems. of the mobile unit, as well as
− between the ASAB-VB and ASFA digital systems.
Generate the virtual balises (VB) towards the digital ASFA system in case of fail-
− Generate
ure in thethe virtual of
detection balises (VB)balises
physical towards theifdigital
(PB’s) ASFA system
this functionality in case of
is enabled.
− failure in the detection of physical balises (PB’s) if this functionality is
Process the TBM to create, from the train position (obtained by the “positioningenabled.
− Process the TBM to create, from the train position (obtained by the “positioning
system” module), the corresponding virtual balises at the corresponding kilo-
system” module),
meter points the corresponding virtual balises at the corresponding kilometer
(KPTs).
− points (KPTs).
Obtain information on the time instant at which a PB has been detected by the
− Obtain
ASFA information
system. on the time instant at which a PB has been detected by the
ASFA system.
− Transmit to the ground infrastructure the detected track balises map (DTBM).
− Transmit to the ground infrastructure the detected track balises map (DTBM).
As explained above, the DTBM contains the physical balises detected during the
As explained above, the DTBM contains the physical balises detected during the
route and their location.
route and their location.
• Positioning system: determines the position of the train from the integration of the
• Positioning system: determines the position of the train from the integration of the
different sensorial systems: GNSS, the detector of balises, and a solution that inte-
different sensorial systems: GNSS, the detector of balises, and a solution that integrates
grates odometry and inertial system by means of a Kalman filter.
odometry and inertial system by means of a Kalman filter.
• Communications module: responsible for managing communications between the
• Communications module: responsible for managing communications between the
train and the ground infrastructure.
train and the ground infrastructure.
 Sensors 2022, 22, x FOR PEER REVIEW 8 of 17
Sensors 2022, 22, 1943 8 of 17

4.4.Positioning
PositioningAlgorithm
Algorithm
Asalready
As alreadyexplained,
explained,the the train
train positioning
positioning system
system is one
is one of most
of the the most important
important aspectsas-
pects of the ASAB-VB system. This module makes it possible to generate
of the ASAB-VB system. This module makes it possible to generate the VBs at the desired the VBs at the
desired
points points
along thealong
route the
androute and tothe
to confirm confirm
status the
andstatus
correctand correct
location oflocation
the PBs. of
Allthe PBs.
this is
All this is in order to establish the corresponding actions on the signalling
in order to establish the corresponding actions on the signalling and, thus, on the speed of and, thus, on
thetrain
the speed onofthethedesired
train on the desired
sections of thesections
route. of the route.
Thesolutions
The solutionsfor forthe
theinitial
initialpositioning
positioningof ofthe
thetrain
trainbefore
beforestarting
startingaaroute
routeand
andduring
during
theroute
the routeare aredifferent.
different.

4.1.Initial
4.1. InitialPositioning
Positioning
Beforestarting
Before startingaaroute,
route,thethetrack
trackononwhich
which the
the train
train is
is located
located hashas to
to be
be determined.
determined.
For
Forthis
thispurpose,
purpose,aadifferential
differentialGNSS
GNSS(DGNSS)
(DGNSS)isisused,
used,withwithan aninstantaneous
instantaneouscorrection
correction
obtained
obtainedthrough
throughwireless
wireless communication
communication (Wi-Fi/Wimax)
(Wi-Fi/Wimax) withwiththethe
terrestrial infrastructure.
terrestrial infrastruc-
This
ture.positioning systemsystem
This positioning ensures that thethat
ensures train’s position
the train’s is obtained
position with anwith
is obtained accuracy of less
an accuracy
than 1 m. Since the distance between the longitudinal axes of two tracks
of less than 1 m. Since the distance between the longitudinal axes of two tracks is between is between 3.8 and
43.8
m,and
an error of less
4 m, an errorthan 3.8/2
of less = 1.9
than m =would
3.8/2 1.9 m be enough
would to guarantee
be enough the discrimination
to guarantee the discrim-
of the track
ination ontrack
of the whichonthe train
which theistrain
located. This error
is located. will be
This error willidentified
be identifiedby εinby
, therefore
εin, there-
in ≤ ε1.9
εfore in ≤m.
1.9 m.

4.2.
4.2.Positioning
Positioningduring
duringthe
theRoute
Route
During
During the route, theASAB-VB
the route, the ASAB-VBsystem
systemperforms
performspositioning
positioningby byfusing
fusingthe
theodometric
odometric
and
andinertial
inertialsystem,
system,corrected
correctedwith
withGNSS
GNSSand
andPBs.
PBs.Figure
Figure55shows
showsthe the general
generalarchitecture
architecture
of the positioning system during the route. As can be seen, two methods of position
of the positioning system during the route. As can be seen, two methods of position meas-
measurement are envisaged: (1) a GNSS and a system that integrates odometry and inertial
urement are envisaged: (1) a GNSS and a system that integrates odometry and inertial
information (which will be identified in what follows by “Kalman-position”); and (2)
information (which will be identified in what follows by “Kalman-position”); and (2) ad-
additionally, there is an input from the control unit that provides information (kilometer
ditionally, there is an input from the control unit that provides information (kilometer
point of the balise position on the track (KPB) and time of balise (tB )) when a PB is detected.
point of the balise position on the track (KPB) and time of balise (tB)) when a PB is detected.

Figure Generalblock
Figure5.5.General blockdiagram
diagramofofthe
thepositioning
positioningsystem
systemonboard
onboardthe
thetrain.
train.

Maximum
Maximumaccuracy
accuracyin inthe
thetrain
trainpositioning
positioningisisonlyonlynecessary
necessaryat atpoints
pointswhere
wherethere
thereisis
some type of signalling, and at these points, the most important thing is relative
some type of signalling, and at these points, the most important thing is relative position- positioning,
since the signals
ing, since affecting
the signals the the
affecting movement
movement of the train
of the generally
train generallyappear appearininpairs
pairsat at pre-
pre-
established
established distances establishaagiven
distances to establish givenrestriction.
restriction.WithinWithin thethe distances
distances between
between the the
two
two signals,
signals, the train
the train mustmust
carrycarry out certain
out certain actions, actions,
and if and if ittofails
it fails do so, to automatic
do so, automatic
braking
braking
would be would be applied.
applied. For theseFor these reasons,
reasons, the position
the position shall be shall be periodically
periodically estimated estimated
by “Kal-
by “Kalman-position” and corrected using the error in the points before
man-position” and corrected using the error in the points before the location of the signals. the location of the
signals.
The train positioning can be reduced to a one-dimensional positioning, where it is
The train
sufficient positioning
to know can be reduced
the longitudinal to a one-dimensional
coordinate of the track on which positioning,
the trainwhere it is
is located.
sufficient to know the longitudinal coordinate of the track on which the
To keep this coordinate updated, known as the kilometer point of the train position on the train is located. To
keep
trackthis coordinate
(KPT), the fusionupdated, known asand
of the odometer the inertial
kilometer pointdata,
system of the train out
carried position
by meanson the of
track (KPT), the fusion of the odometer and inertial system data, carried
a Kalman filter, is used. The ASFA standard [20] requires positioning systems with these out by means of
acharacteristics
Kalman filter,to is have
used.anThe ASFA
error standard
of less than 5% [20] requires
over positioning
distances of up to systems with works
800 m. Some these
Sensors 2022, 22, 1943 9 of 17

characteristics to have an error of less than 5% over distances of up to 800 m. Some works
comfortably meet these specifications, with errors less than 5 m in 1 km [21], so it can be
assumed that the odometric-inertial system will keep an estimation of the position with an
error less than these values.
In the case of tunnels, in the absence of GPS coverage, the positioning system is based
on the information provided by the odometric system. It should be noted that in most
tunnels there are physical balises that allow correcting the odometric error, being able to
position a VM at the closest point to the tunnel entrance at a point where GPS coverage
is good. Correcting the odometric error when GPS coverage is expected to be lost in the
vicinity of the VMs is precisely the purpose of VMs.

4.3. Reduction of the Positioning Error during the Route

ERTMS systems place balises at every kilometer to reset the above-mentioned errors.
The system proposed in this paper is aimed at conventional lines with low traffic density,
where this signalling does not exist. To reduce the odometric-inertial system error, the
information contained in the TBM, the positioning with GNSS (GPS, Galileo, etc.), and the
detection of conventional signalling balises (ASFA) are used.
Currently, in satellite positioning the uncertainty is less than 3 m 99% of the time
with coverage, although it can reach values less than 1 m. In fact, this positioning, with
the service EGNOS SoL (Safety of Life), meets aviation standards and is used in landing
maneuvers [22,23].
The TBM contains the position of the virtual marks (VMs) distributed along the route.
These VMs are located at points where the orography allows better satellite coverage. The
density of VM in the map is a function of the proximity to the points where there are signs.
If there is no signalling, the positioning error is not relevant. However, at the points where
there are signals that condition the actions of the train it is important to have more precision
in positioning. VMs allow the position to be corrected using satellite positioning. This
procedure is carried out from 1 km before the situation of the VM until the position of the
VM is exceeded. The distance between the current train position (Xi ) and the position of
the virtual mark (XVM ), called D2VM, is approximated by the line that joins the current
train position to the mark using the Equation (1).

D2V M = Arc( Xi − XV M ) × R, (1)

where Xi are the GNSS coordinates of the train at the moment Ti , XVM are the GNSS
coordinates of the mark and R is the radius of the earth, Equation (2).
v
u r2 · cos( Lat) 2 + r2 · sin( Lat) 2
u


1 2
R = Altitude + t (2)
(r1 · cos( Lat))2 + (r2 · sin( Lat))2

where r1 and r2 the radius at the Equator and at the Pole at sea level, 6378.137 km and
6356.752 km respectively, and Lat is determined by using the Equation (3).

Lat Xi + Lat XV M
Lat = (3)
2
To avoid stopping rail traffic by the use of auscultation vehicles, the definition of the
position of the virtual marks (VMs) is done on a map or with support vehicles, at points
close to the track. For this reason, this position does not coincide with the exact passage
of the train, so that, in the approach phase, the coordinates of the virtual marks (VMs)
are transformed into those corresponding to the projections of these on the trajectory of
the train.
Sensors 2022, 22, 1943 10 of 17

The procedure, in that case, consists of the following steps (see Figure 6):
1. The last five positions provided by the satellite positioning system, Xi-4 to Xi in
Sensors 2022, 22, x FOR PEER REVIEW Figure 6, are recorded at 1-s intervals. To reduce the effect of positioning uncertainty,
10 of 17
if the distance between consecutive positions is less than 30 m, then 30 m of distance
between consecutive positions are fixed.
2. The train’s trajectory is approximated to a straight line using least squares (blue line
2. The train’s trajectory is approximated to a straight line using least squares (blue line
in Figure 6).
3. in Figure
The 6). train position (X ) obtained by GNSS and the positioning mark X
last known i VM
3. The last known
are projected onto train
theposition (Xi)line
calculated obtained by GNSS
(red points and the6),
in Figure positioning
obtaining mark
XiP andXVM
Xare projected onto the calculated line (red points in Figure 6), obtaining XiP and XVMP,
VMP , respectively. Then, the distance between the two projections, XiP and XVMP , is
respectively. Then, the D2VM
calculated by obtaining distance
(seebetween
Figure the
6). two projections, XiP and XVMP, is calcu-
lated by obtaining D2VM (see Figure 6).

Figure6.6.Train
Figure Trainposition
positioncorrection
correctionusing
usingvirtual
virtualmarks
marks(VM).
(VM).

4.4. IfIfD2VM
D2VMisisless
lessthan
thanone
onekilometer,
kilometer,the
thelongitudinal
longitudinalposition
positionofofthe
thetrain
trainisiscorrected
corrected
usingthe
using theEquation
Equation(4):
(4):

𝐾𝑃𝑇 = 𝐾𝑃𝑉𝑀 − 𝐷2𝑉𝑀 (4)

KPT = KPV M − D2V M (4)
5. If a PB is detected, the kilometer point of the train position on the track (KPT) is im-
5. If a PB is detected, the kilometer point of the train position on the track (KPT) is
mediately updated using Equation (5):
immediately updated using Equation (5):
𝐾𝑃𝑇 = 𝐾𝑃𝐵 (5)
KPT = KPB (5)
6. If the kilometer point of the train position on the track (KPT) is greater than or equal
6. Iftothe
thekilometer
kilometerpoint of of
point thethe
train position
virtual on position
balise the trackon
(KPT)
the is greater
track than or
(KPVB), theequal
VB is
to the kilometer
generated. point of the virtual balise position on the track (KPVB), the VB is
generated.
Figure 7 shows a schematic of the operations performed when the train is en route
Figure
to: (i) 7 shows
correct a schematic
positioning, of thePBs
(ii) detect operations
to createperformed
DTBM, and when the train VBs.
(iii) generate is en route to:
(i) correct positioning, (ii) detect PBs to create DTBM, and (iii) generate VBs.
Sensors 2022,
Sensors 22,22,
2022, 1943
x FOR PEER REVIEW 11
11ofof1717

Figure7.7.Operations
Figure Operationsperformed
performedtoto(i)(i)correct
correctpositioning,
positioning,(ii)
(ii)detect
detectPBs
PBstotocreate
createDTBM,
DTBM,and
and(iii)
(iii)
generate VBs (train on route).
generate VBs (train on route).

5.5.Results
Results
ToToverify
verifythetheoperation
operationofofthe
thesystem,
system,several
severaltests
testshave
havebeen
beenconducted.
conducted.ThisThissection
section
describesthe
describes thetests
testsperformed
performedtotovalidate
validatethetheoperation
operationofofthethesoftware
softwareand andhardware
hardwarethat that
constitutethe
constitute the ASAB-VB
ASAB-VB system
system proposed
proposed in this
in this paper.
paper. Figure
Figure 8 shows
8 shows a picture
a picture of theoftest
the
platform used for
test platform usedtesting the ASAB-VB
for testing system.system.
the ASAB-VB The platform is madeisup
The platform of the
made upfollowing
of the fol-
components: the ASFA
lowing components: system
the ASFA(onboard indications),
system (onboard the balise
indications), thedetector and odometric
balise detector and odo-
information receiver, and the ASAB-VB system.
metric information receiver, and the ASAB-VB system.
Due
Duetotothethe
difficulty of performing
difficulty of performingteststests
on the
onrailway track with
the railway trackthe complete
with equip-
the complete
ment, tests have
equipment, testsbeen
haveperformed with a with
been performed balisea emulator connected
balise emulator to the to
connected real
theequipment
real equip-
inment
Figurein 8. Tests8.have
Figure Testsalso
havebeen performed
also on trains
been performed onusing
trainsinertial
using navigation and GNSS
inertial navigation and
positioning, the results
GNSS positioning, theshown
resultsbelow
showncorrespond to these to
below correspond tests.
these tests.
Sensors 2022,
Sensors 22, x1943
2022, 22, FOR PEER REVIEW 1212ofof17
17
Sensors 2022, 22, x FOR PEER REVIEW 12 of 17

Figure
Figure 8.
8. Test
Test platform.
platform.
Figure 8. Test platform.
AA GNSS
GNSS receiver
receiver with
with EGNOS
EGNOS correction
correction onboard
onboard the
the train
train waswas used
used for
for the
the tests
tests
and
and the A GNSS receiver
the differential with
differential correctionEGNOS
correction was correction
was performed
performed with onboard the
with aa ground train
ground station. was
station. To used for the
To determine
determine thetests
the
and
positionthe with
position differential
highercorrection
with higher resolution was
resolution at performed
at which
which withgenerates
the system
the system a groundthe
generates station.
the To GNSS
VBs, the
VBs, the determine
GNSS the
position
position
position
is linearly
is linearlywith higher resolution
interpolated
interpolated from theat
from the which
before
before theafter
and
and system
after generatescaptured
coordinates
coordinates the VBs,at
captured the
at GNSS
the
the position
instant
instant of VB
of VB
is linearly interpolated
generation.
generation. These from
These are the
the the before
coordinates
coordinates and afterinin
presented
presented coordinates
Tables
Tables1 1andcaptured
and 2. 2.
The atlatency
the instant
latency
The of of
thethe ofsys-
VB
system
generation.
is 1 ms,
tem ms,These
is 1 which which are
allows coordinates
the generation
allows ofpresented
the generationVBs VBs in
of with Tables
a position
with 1resolution
and 2.
a position Theoflatency
resolution5.5 of of
cm5.5at the sys-
a speed
cm at a
tem
of is
200 1 ms,
km/h.
speed of 200 km/h.which allows the generation of VBs with a position resolution of 5.5 cm at a
speed of 200 km/h.
Table 1.
Table VB and
1. VB and VM
VM data.
data.
Table 1. VB and VM data.
KP (km)
KP (km) D2VBD2VB(km) (km) (°) (◦ )
Lat Lat Lon (◦ )
Lon(°)
VB VB
1 1 KP (km)
39.47539.475 D2VB 0 (km) 0 Lat (°)
N-40.539961
N-40.539961 Lon (°)
W-3.291680
W-3.291680
VB
VM VM
11 1 39.47537.895
37.895 0 1.580
1.580 N-40.539961
N-40.529683
N-40.529683 W-3.291680
W-3.304593
W-3.304593
VM VM1
2 2 37.895
38.642 38.642 1.580
0.833 0.833 N-40.534558
N-40.529683
N-40.534558 W-3.298513
W-3.304593
W-3.298513
VM3 39.052 0.423 N-40.537163 W-3.295077
VM32 38.642
39.052 0.833
0.423 N-40.534558
N-40.537163 W-3.298513
W-3.295077
VM3 39.052 0.423 N-40.537163 W-3.295077
Table
Table 2. VB data..
VB generation data
Table 2. VB generation data.
Lat((°)
Lat ◦) Lon(◦(°)
Lon ) Error (m)
Error (m)
Lat (°)
N-40.5399758 Lon (°)
W-3.2917061 Error
0.51(m)
N-40.5399758 W-3.2917061 0.51
N-40.5399758 W-3.2917061 0.51
As an example, a test in which a virtual balise (VB) is generated is detailed. This test
As an
As an example,
example, a testtest in
in which
which aa virtual
virtual balise
balise (VB)
(VB) is generated
generated is is detailed.
detailed. This
This test
test
involves the VB to beagenerated and three VMs located atisdifferent distances from the VB.
involves
involves the VB to
the VB tothebe generated
be location
generated and three VMs located at different distances from the VB.
Figure 9 presents of and
the three VMs and located
the atVB.different distances
It is observed thatfrom
the the VB.
greater
Figure
Figure 9 presents
9 presents the location
the the
location of the
of the three VMs and the VB. It is observed that the greater
the distance between VB and the three VMs
VM, the and the
greater theVB. It is
error in observed thatatthe
the position greater
which the
the distance
the distance between
between the the VB
VB and
and the
the VM,
VM, the
the greater
greater the
the error
error inin the
the position
position atat which
which thethe
VB is generated with respect to the position stored in the TBM.
VB is generated with respect to the position stored
VB is generated with respect to the position stored in the TBM. in the TBM.

Figure 9. VMs and VB in the test.

Figure 9. VMs and VB in the test.
Figure 9. VMs and VB in the test.
Table 1 presents the locations of the VMs and the VB corresponding to this test, cor-
Table 1to
responding presents thebetween
the route locations of thede
Alcalá VMs and the
Henares andVB corresponding
Guadalajara to this
shown test, cor-
in Figure 10.
responding to the route between Alcalá de Henares and Guadalajara shown in Figure 10.
Sensors 2022, 22, 1943 13 of 17

Sensors 2022, 22, x FOR PEER REVIEW 13 of 17

Table 1 presents the locations of the VMs and the VB corresponding to this test,
corresponding to the route between Alcalá de Henares and Guadalajara shown in Figure 10.
The kilometer point (KP), the distance to the VB (D2VB), and the GNSS coordinates (latitude
The kilometer point (KP), the distance to the VB (D2VB), and the GNSS coordinates (lati-
and longitude) are provided in this table.
tude and longitude) are provided in this table.

Figure 10. Map of the route with the location of the PBs (https://www.google.es/maps/ (accessed
Figure
on 10. Map of
8 November the route with the location of the PBs (https://www.google.es/maps/ (accessed
2021)).
on 8 November 2021)).
Several tests have been performed to generate the VB using the three VMs as refer-
ence. In these tests, the KPT is obtained with the inertial system.
In the first test, only the VM2 mark is used to perform the positioning. Figure 11 pre-
sents the last positions of the train before passing VM2. The blue asterisk represents the
projection on the estimated train path (linear approximation) of the last position detected
Sensors 2022, 22, 1943 14 of 17

Several tests have been performed to generate the VB using the three VMs as reference.
In these tests, the KPT is obtained with the inertial system.
Sensors 2022, 22, x FOR PEER REVIEW 14 of 17
In the first test, only the VM2 mark is used to perform the positioning. Figure 11
presents the last positions of the train before passing VM2 . The blue asterisk represents the
projection on the estimated train path (linear approximation) of the last position detected
by the GNSS before the VM22 mark.mark. The
The green
green asterisk
asterisk represents
represents the
the position
position of
of VM
VM22and
and
the red asterisk is its projection on the estimated trajectory of the train.
trajectory of the train.

Figure 11.
Figure 11. MV
MV and
and train
train position
position locations
locations in
in the
the test.
test.

The
The information
information at
at that
that point
point is:
is:
𝐷2𝑉𝑀 = 23.56 m
𝐾𝑃𝑇 = 𝐾𝑃𝑉𝑀 − 𝐷2𝑉𝑀 D2V =M 38.642 − 0.02356
2 = 23.56 m = 38.61844 (6)
KPT =𝐷2𝑉𝐵KPV M = 20.833
− D2V + 0.02356 = 0.85656
M2 = 38.642 km = 38.61844
− 0.02356 (6)
D2VB = 0.833 + 0.02356 = 0.85656 km
From that moment on, the train uses the inertial system to keep the KPT updated,
and when it reaches point 39.475 it generates the VB. Table 2 presents the data recorded
From that moment on, the train uses the inertial system to keep the KPT updated, and
at the time of VB generation:
when it reaches point 39.475 it generates the VB. Table 2 presents the data recorded at the
Only the kilometric point to virtual balise (KPVB) stored in the TBM is used to gen-
time of VB generation:
erate the VB. In this case, the coordinates (lat and lon) of the VB have been used to deter-
Only the kilometric point to virtual balise (KPVB) stored in the TBM is used to generate
mine the positioning error. This error is calculated as the distance between the projections
the VB. In this case, the coordinates (lat and lon) of the VB have been used to determine
of the train position (XiP) (black asterisk in Figure 12) and the VB (KPVBP) (red asterisk in
the positioning error. This error is calculated as the distance between the projections of
Figure 12) on the linear trajectory of the train (Figure 12), at the time when the balise is
the train position (XiP ) (black asterisk in Figure 12) and the VB (KPVBP ) (red asterisk in
generated.
Figure 12) on the linear trajectory of the train (Figure 12), at the time when the balise is
generated.
Table 3 presents the errors obtained in the other test using VM1 and VM3 to perform
the positioning.

Table 3. Results summary.

Lat (◦ ) Lon (◦ ) Error (m)

VM1 N-40.5399730 W-3.2917095 0.92
VM2 N-40.5399758 W-3.2917061 0.51
VM3 N-40.5399812 W-3.2916992 0.33
 Sensors2022,
Sensors 2022,22,
22,1943
x FOR PEER REVIEW 15
15 of 17
17

Figure12.
Figure 12. Position
Position error
error in
in the
the generation
generation of
of VB.
VB. VB
VB position
position according
according to
to TBM
TBM (green
(green asterisk),
asterisk),VB
VB
projectionon
projection onthe
thelinear
lineartrajectory
trajectoryof ofthe
thetrain
train(red
(redasterisk),
asterisk),and
andVB
VB generation
generationpoint
point(black
(blackasterisk).
asterisk).

As can3be
Table seen, the
presents thedistance between the
errors obtained VMs
in the andtest
other theusing
VB is VMdeterminant
1 and VMfor 3 tothe error
perform
in the position
the positioning. at which the VB is generated. The shorter distance between VM and VB, the
smaller the error. It should be considered that GPS coverage limitations in many points of
the
Tablerailway route
3. Results make it. impossible to place the VMs close to the VBs since the VMs must
summary
be located at points where there is good GPS coverage.
Finally, it should be noted Lat (°) several tests were Lon
that (°)
conducted with similar Error (m) Tests
results.
VM1 performed between N-40.5399730
were September and November W-3.2917095
in different weather conditions 0.92 on the
VM2 line shown in Figure
railway N-40.5399758
10. During the tests,W-3.2917061
VBs and VMs were placed0.51 at different
locations.
VM3 It should be noted that
N-40.5399812 the key aspect for accurate
W-3.2916992 of VB generation (minimum
0.33
error) is not the railway line length, but the maximum distance between the VMs and VBs.
Therefore,
As can theberesults obtained
seen, the distance arebetween
extrapolated
the VMs to railway
and thelines VB isofdeterminant
any length. for the error
in the position at which the VB is generated. The shorter distance between VM and VB,
6. Conclusions
the smaller the error. It should be considered that GPS coverage limitations in many points
of theThe ASAB-VB
railway routesystem
makedeveloped
it impossible andto
presented
place the inVMs
this paper
close highlights
to the VBsthe advantages
since the VMs
itmust
can bebring to rail traffic by introducing very
located at points where there is good GPS coverage. significant improvements concerning the
performance of the ASFA digital system. All of this is achieved
Finally, it should be noted that several tests were conducted with similar results. by introducing minimal
modifications
Tests were performed to the onboard
betweensystem September and andwithout
Novemberintroducing any modification
in different weather conditionsto the
track
on the infrastructures.
railway line shown In addition,
in Figure the10.
proposed
Duringsystem
the tests, makes
VBs theandlocation
VMs were of the VBs at
placed more
dif-
flexible to replace or complement the information of the PBs.
ferent locations. It should be noted that the key aspect for accurate of VB generation (mini- Thus, the system allows
loading
mum error) the VBsis notcoordinates
the railwaythat lineestablish
length, but thethe
speed limits (temporary
maximum distance betweenspeedthe limits,
VMsTSL) and
at different points along the route. This makes railway signaling
VBs. Therefore, the results obtained are extrapolated to railway lines of any length. more flexible, dynamic
and safe. Additionally, the system may also be used to obtain information on the status
of
6. the physical balises (operational balises, absence of balises, damaged ones) with the
Conclusions
consequent maintenance cost saving.
The ASAB-VB system developed and presented in this paper highlights the ad-
To do this, it is only necessary to accurately record the coordinates and kilometer point
vantages it canofbring
of the location each to
ofrail
the traffic
balisesby introducing
(PBs and VBs) very and somesignificant improvements
auxiliary points calledconcern-
virtual
ing the(VMs),
marks performance
that allow of the
the system
ASFA digital system.
integration errorsAlltoofbethis is on
reset achieved by introducing
the route. In this way,
minimal
thanks tomodifications
the VMs, the error to thein onboard system
the position at and
which without
the VBs introducing any modification
are to be generated is less
than 1 m. This error is completely acceptable for this type of application. location of the
to the track infrastructures. In addition, the proposed system makes the
VBs It more flexible totoreplace
is important note that or complement
a substantialthe information
number of the PBs.
of references basedThus,
on the
the system
use of
allows
VBs loading
in the the VBs
context coordinates
of ERMS signaling thatsystems
establishcan thebespeed
found limits (temporary
in the literature. speed limits,
However,
TSL) at different points along the route. This makes railway
the proposed ASAB-VB system is intended for conventional lines with low traffic density signaling more flexible, dy-
namic and safe. Additionally, the system may also be used to
where ERTMS is not available. Therefore, the concept of VB is different in this application, obtain information on the
Sensors 2022, 22, 1943 16 of 17

where VBs are used to facilitate the incorporation of temporary speed limits (TSL) into
the railway track. TSL is used in unexpected risk situations: falling objects on the track,
unscheduled maintenance operations, landslides, unexpected accidents, etc.
The operation of the system is conditioned to the detection of a PB or GNSS coverage
in the vicinity of the point where the VB is to be located. As shown in the results section,
for distances of 1.5 km, the position error is less than 1 m. The accuracies shown in the
experiment are calculated using differential GNSS correction with a ground station. It
must be considered that if there is no such possibility during the journey, the error would
be increased by the error provided by the GNSS itself with EGNOS correction, which as
mentioned above, is less than 3 m 99% of the time. This error combined with that of the
odometric-inertial system, is still sufficient for the location of temporary speed limitation
VBs (which are the main application). If the correction is made with PBs, the results would
be the same as shown. It is not easy to quantify the level of safety provided by the ASAB-VB
system. For this purpose, information should be obtained over a long period of time to
compare the number of accidents before and after the implementation of the ASAB-VB
system. Qualitatively, the ASAB-VB system represents an unquestionable improvement in
safety. Improvement in the safety is due to:
1. In the ASB-VB system, VBs are used to facilitate the incorporation of temporary speed
limits (TSL) into the railway track. TSL is used in unexpected risk situations: falling
objects on the track, unscheduled maintenance operations, landslides, accidents, etc.
2. The system makes it possible to identify broken down or damaged physical balises
(PB). These broken down or damaged balises could cause accidents. In addition,
damaged balises are detected when trains are in service, i.e., without the need for
specific maintenance operations.
Future lines of research include (i) adding redundancy and diversity to the system to
increase its reliability, and (ii) complementing the VM generation based on the information
provided by GNSS with MV generation based on information obtained from natural and
artificial marks (poles, traffic lights, trees, signs, etc.). This avoids dependence on GNSS
coverage. Artificial vision systems could be used to detect natural and artificial marks.

7. Patents
The result from the work reported in this manuscript (ASAB-VB system) is patented
in Spain with patent number ES 2418929 A1.

Author Contributions: Conceptualization, E.S.; Formal analysis, C.L.; Funding acquisition, M.M.;
Investigation, E.S. and J.A.J.; Project administration, M.M.; Resources, F.J.R.; Software, C.L.; Super-
vision, E.S.; Validation, J.A.J.; Writing—original draft, J.A.J.; Writing—review & editing, F.J.R. All
authors have read and agreed to the published version of the manuscript.
Funding: This research was funded by the Spanish Ministry of Science and Innovation of the
Government of Spain under the MANTIS Project (PPT-460000-2009-001). The project is co-financed
by the European Regional Development Fund (ERDF).
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Available online: https://www.statista.com/chart/24014/total-number-of-passenger-fatalities-in-railway-accidents/ (accessed
on 28 January 2022).
2. Available online: https://www.era.europa.eu/sites/default/files/library/docs/safety_interoperability_progress_reports/
report_on_railway_safety_and_interoperability_in_the_eu_2020_en.pdf (accessed on 28 January 2022).
3. Bertran, R.; Delgado-Penín, J.A. On the Use of GPS Receivers in Railway Environments. IEEE Trans. Veh. Technol. 2004, 53,
1452–1460. [CrossRef]
Sensors 2022, 22, 1943 17 of 17

4. Acharya, A.; Sadhu, S.; Ghoshal, T.K. Train localization and parting detection using data fusion. Transp. Res. Part C 2010, 19,
75–84. [CrossRef]
5. Stadlmann, B. Automation of operational train control on regional branch lines by a basic train control. In Proceedings of the
IEEE Intelligent Transportation Systems Conference, Toronto, ON, Canada, 17–20 September 2006.
6. Mazl, R.; Preucil, L. Sensor data fusion for inertial navigation of trains in GPS-dark áreas. In Proceedings of the IEEE Intelligent
Vehicles Symposium, Columbus, OH, USA, 9–11 June 2003.
7. Bedrich, S.; Gu, X. GNSS-Based Sensor Fusion for Safety-Critical Applications in Rail Traffic; NAVITEC workshop on Satellite
Navigation User Equipment Technology: Noordwijk, The Netherlands, 15 December 2004.
8. Meng, Y.; Feng, J.; Wang, W. An enhanced INS/GNSS integrated system with ODO for train positioning. In Proceedings of
the IEEE 4th Advanced Information Management, Communicates, Electronic and Automation Control Conference (IMCEC),
Chongqing, China, 18–20 June 2021.
9. Otegui, J.; Bahillo, A.; Lopetegi, I.; Díez, L.E. A survey of train positioning solutions. IEEE Sens. J. 2017, 17, 6788–6797. [CrossRef]
10. Gerlach, K.; Meyer zu Hörste, M. A precise digital map for GALILEO-based train positioning systems. In Proceedings of the
9th International Conference on Intelligent Transport Systems Telecommunications (ITST), Lille, France, 22–29 October 2009;
pp. 343–347.
11. Jiang, W.; Chen, S.; Cai, B.; Wang, J.; ShangGuan, W.; Rizos, C. A multi-sensor positioning method-based train localization system
for low density line. IEEE Trans. Veh. Technol. 2018, 67, 10425–10437. [CrossRef]
12. Azaola, M.; Fidalgo, J.; Barrios, J.; López, M.; Cezón, A.; Arias, J.J. Introduction of GNSS in Railway Safety of Life Applications:
Virtual Balise Detection in ERTMS. In Proceedings of the 31st International Technical Meeting of the Satellite Division of the
Institute of Navigation (ION GNSS+ 2018), Miami, FL, USA, 24–28 September 2018; pp. 1483–1496.
13. de Andrés, S.H. La tecnología GNSS en el sistema de señalización ERTMS. Rev. Digit. Del Cedex 2020, 197, 74–89.
14. ESA. The ESA 3InSat Project. Available online: https://business.esa.int/projects/3insat (accessed on 7 October 2021).
15. European GNSS Agency (GSA). The GSA ERSAT-EAV Project. Available online: http://www.ersat-eav.eu/ (accessed on 7
October 2021).
16. RHINOS Project. Available online: https://www.euspa.europa.eu/railway-high-integrity-navigation-overlay-system (accessed
on 7 October 2021).
17. STARS Project. 2017. Available online: https://www.stars-rail.eu/ (accessed on 7 October 2021).
18. Santiso, E.; Jiménez, J.A.; Hernández, A.; Marquez, V.; Mazo, M. Procedimiento Y Sistema Para La Mejora Del Sistema ASFA
Digital Incorporando Balizas ASFA Virtuales, Perfil de Balizas Y Posicionamiento Del Tren, Oficina Española de Patentes Y
Marcas. Patent ES 2418929 A1, 8 May 2014.
19. ERTMS. Available online: https://www.ertms.net/ (accessed on 15 September 2021).
20. ET 03.365.003.7 Technical Specification Digital ASFA. Available online: http://descargas.adif.es/ade/u18/GCN/NormativaTecnica.
nsf/v0/033BE64FB79AF9A4C12581550027BEFA/$FILE/ET%2003.365.003.7_ASFA%20V%C3%8DA_Ed2.pdf?OpenElement
(accessed on 12 November 2021).
21. Allotta, B.; D’Adamio, P.; Malvezzi, M.; Pugi, L.; Ridolfi, A.; Vettori, G. A Localization Algorithm for Railway Vehicles. In
Proceedings of the IEEE International Instrumentation and Measurement Technology Conference (I2MTC), Pisa, Italy, 11–14 May
2015.
22. EGNOS Safety of Life (SoL) Service Definition Document. Available online: https://www.gsa.europa.eu/sites/default/files/
brochure-sol-sept-2016_v8_hr.pdf (accessed on 12 November 2021).
23. Fernandez-Hernandez, I.; Vecchione, G.; Díaz-Pulido, F.; Jeannot, M.; Valentaite, G.; Blasi, R.; Reyes, J.; Simón, J. Galileo High
Accuracy: A Programme and Policy Perspective. In Proceedings of the 69th Intrnational Astronautical Congress, Bremen,
Germany, 1–5 October 2018.