Comparison of Collision Avoidance Systems
and Applicability to Rail Transport
Cristina Rico Garc
´
ıa
∗
, Andreas Lehner
∗
, Thomas Strang
∗
and Matthias R
¨
ockl
∗
∗
German Aerospace Center
Institute of Communications and Navigation
82230 Wessling
Germany
Email: {cristina.ricogarcia, andreas.lehner, thomas.strang, matthias.roeckl}@dlr.de
Abstract—The paper presents an overview of the state
of the art in collision avoidance related with transportation
systems like the Automatic Identification System (AIS)
for maritime transportation, Traffic Alert and Collision
Avoidance System / Automatic Dependent Surveillance-
Broadcast (TCAS/ADS-B) for aircraft, and the Car-2-Car
communication system (C2C) for road transportation. The
examined systems rely on position detection and direct
communication among vehicles. Alike a collision avoidance
system for railway transportation ”RCAS” is introduced.
Focussing on the communication aspects, possible ap-
plicability of the examined state of the art systems to
RCAS is studied. The analysis are performed at different
communication system layers, namely application (APP)
layer, media access control (MAC) layer and physical layer
(PHY), which are the most relevant for a single hop
network broadcast system as favorized in RCAS. Since
multihop and addressed communication are not foreseen
in a first RCAS approach, the network layer is not taken
into account.
I. INTRODUCTION
Actual statistics of the International Union of Rail-
ways (UIC) show that there are three significant train
accidents in Europe every day [3], despite millions of
Euros which have been invested in trackside and in-train
safety equipment. In order to increase safety in railway
traffic, some countries are partially installing control
systems, mainly centralized management ones, specially
the Automatic Train Control (ATC), where the trains
are monitored by devices located along the rail. These
devices send the collected information to an operation
center that can pass specific instructions to the train. A
European ATC standard, European Train Control System
(ETCS), is intended to replace the various European
ATC systems, in order to protect international train
traffic. However according to the estimations of the
German railway company ”Deutsche Bahn” (DB), it
could take up to 20 years and cost up to 8 billion Euros
to introduce ETCS right across Europe [2]. Additionally,
only the operation center has an overall overview of the
traffic situation, and a train driver could only be warned
of a hypothetical collision if the operation center decides
it.
While maritime, air, and road transport have a vehicle
integrated collision avoidance system available or in the
development phase, we find no satisfactory solution of
this type of technology in railway transportation.
Therefore the German Aerospace Center (DLR) is
currently developing a Railway Collision Avoidance
System (RCAS) [1] that will allow the drivers to have
an up-to-date accurate knowledge of the traffic situation
in the vicinity, and act in consequence. The system is
intended to not rely on components in the infrastructure,
this way substantially reducing its rollout- and mainte-
nance costs, as well as inherently providing a migration
strategy. The basic idea of RCAS is to calculate the own
position and movement vector and broadcast this infor-
mation as well as additional data like vehicle dimensions
to all other trains in the area. Thus, the driver’s cabin
could be equipped with a display showing the position of
the other vehicles in the region. Computer analysis of the
received information, the own position and movement
vector and an electronic track map detects possible
collisions, displaying an alert signal, and advising the
driver of the most convenient strategy to follow in order
to avoid the danger. The system can take into account
different danger sources, like advancing trains or road
vehicles or obstacles, and classify them according to a
specific scale.
The purpose of this paper is to discuss the state
of the art of collision avoidance system for different
transport means, i.e. TCAS, ADS-B, AIS, C2C, from the
communication point of view and examine their possible
applicability to RCAS.
The paper is organized as follows: First we describe
the surveillance strategy of the above mentioned systems
and the appropriateness for RCAS are discussed. Sec-
tion 3 introduces the different MAC layer approaches
utilized. Furthermore, the special constraints in RCAS
are identified. Section 4 examines the physical layer, and
the specific requirements of RCAS are analyzed. Finally,
section 5 provides a summary.
II. SURVEILLANCE STRATEGY
The top level analysis is related to the surveillance
strategy.
1) Maritime Navigation: The maritime surveillance
application of AIS [4] consists of a continuous inter-
change of driving data from ships as can be seen in Fig.
1, like GNSS position, speed and direction, as well as
relevant information like identification numbers, length
and beam, ships draught, route plan etc.. The received
data from other vessels enables the calculation of the
Closest Point of Approach (CPA) and Time to Closest
Point of Approach (TCPA), which provides information
regarding risk of collision.
Fig. 1. AIS surveillance strategy
2) Aviation: TCAS [8] relies on the secondary
surveillance radar (SSR). By means of this radar, a
TCAS equipped aircraft interrogates other aircraft in its
vicinity and listens for the transponder replies. Computer
analysis of these replies determines which aircraft rep-
resent potential collision threats. Consequently Traffic
Advisories (TA), in TCAS I, and resolution advisories
(RA), in TCAS II and III, can be provided. The TCAS
basic functionality conforms a serie of steps.
• The aircraft announce their presence via squitter,
i.e, a burst of data sent without being prompted by
an interrogating radar. A short downlink message
containing the aircrafts address is sent periodically.
• When an aircraft realizes another aircraft (intruder)
is in its vicinity, it interrogates the intruder asking it
for its altitude. At the same time the radar calculates
the distance to the intruder and its speed.
• Taking into account the gained information, the
aircraft evaluates whether the intruder is a threat
or could be a threat in short time. Steps 2 and 3
are illustrated in Fig. 2, 3, 4.
• In order to control threats, immediate threats, de-
picted in Fig. 3, are often interrogated, while poten-
tial threats, illustrated in Fig. 4 are less frequently
interrogated.
• If TCAS II or TCAS III are used, and an immediate
threat is detected, the aircraft and the intruder
perform a coordinated communication to reach an
agreement about the most suitable resolution advi-
sory as can be seen in Fig. 3.
ADS-B [12] conforms the evolution of TCAS. Based
on the Global Positioning System (GPS), an aircraft
can automatically broadcast its identification address,
Fig. 2. TCAS interrogation sequence in a non threat intruder situation
Fig. 3. TCAS interrogation sequence in a threat intruder situation
GPS derived latitude and longitude, altitude and the
3-D velocity as depicted in Fig. 5. The transmission
of ”intention” will be included as well. This broadcast
information can be picked up by the Air Traffic Con-
troller (ATC) and other aircraft in the range, to track
aircraft more precisely than with radar [9]. Consequently
it is possible to display this information on the cockpit
display without the involvement of ground station equip-
Fig. 4. TCAS interrogation sequence in a potential threat intruder
situation
ment and alert the pilot in case of danger.
Fig. 5. ADS-B surveillance strategy
3) Road Transport: The Car 2 Car Communication
Consortium [13] aims to establish an open European
industry standard for the ad-hoc communication between
vehicles and vehicles to infraestructure, which is cur-
rently in the development phase. Although the scope of
C2C is very wide, it will offer among others safety appli-
cations for collision prevention. The cars are equipped
with various types of sensors. These sensors may for
instance detect the abrupt braking of the vehicle and a
short distance to the preceding vehicle. Together with the
position determined by a GNSS receiver this information
is broadcast to the vehicles in the vicinity which may
detect the presence of a traffic jam by exploiting the
received information as illustrated in Fig. 6. In order to
extend the network, multihop is implemented. Therefore
the range is not limited to the communication range,
which is relatively short.
Fig. 6. Car-2-Car surveillance strategy
4) Applicability to RCAS: While the surveillance ap-
plication of AIS, ADS-B and C2C are broadcast systems,
TCAS relies on a duplex addressed communication. A
point to point communication has the advantage that it
allows the reduction in the number of sent messages,
since the intruders that do not constitute a threat interrupt
the communication. Therefore, a larger maximum den-
sity or message length is supported. The first option to
implement is the simpler broadcast protocol. However,
if the message length or maximum permitted density
would be a restriction, which is not the case in a first
asessment, a point to point communication should be
considered. On the other hand, addressed communica-
tion allows the agreement on a coordinated movement
of the involved vehicles.
The special railway transportation characteristics add
major difficulties at this level; the manoeuvring possibil-
ities are considerably reduced and thus, the reaction op-
tions are mainly limited to brake the train. The potential
speed of the trains, combined with the reduced reaction
capabilities and the geographical proximity between
adjacent rails, introduces a high accuracy requirement on
position determination - down to the cm level - in RCAS.
Different location techniques (like GNSS and RADAR)
are utilized in the maritime, aircraft and C2C surveil-
lance systems. No further analysis of the RADAR ap-
proach is carried out, since it can be assured, that in the
majority of the scenarios, no direct line of sight between
RCAS modules is available. GNSS has been proved to
be a suitable localization method for trains in the project
RUNE (Rail User Navigation Equipment) [15], funded
by the European Space Agency (ESA). However, there
is no direct line of sight between satellites and RCAS
modules e.g. in tunnels and when passing areas in forest.
Thus other technologies - especially optical sensors, and
eddy current sensors - should be incorporated in order to
support satellite based positioning. Similarly, movement
vectors are not just vectors in a 2-dimensional (maritime)
or 3-dimensional (aeronautics) space but rely on precise
trackmaps, which includes highly reliable knowledge
about the status of e.g. track switches. Therefore, alike
ADS-B the transmission of intention, i.e., the expected
route, is of special importance in RCAS.
Road transport behavior differs considerably from the
railway transport, the high density of vehicles on roads
that allows network extension through multihop can not
be assumed in railway transport. Nonetheless, a direct
interface of RCAS with C2C, supported by video would
be one of the technologies utilized to avoid accidents on
level crossings.
III. MAC LAYER
The MAC layer has to be carefully designed as it
defines the throughput of the system. Since there is no
upper layer to manage packet collisions, the MAC layer
should avoid packet collisions or ensure a suitable low
level of collision rate for a surveillance application. At
this point appropriate message length and message rate
should be selected in order to complete the specifications
of the MAC layer.
1) Maritime Navigation: Maritime AIS should be
capable of operating autonomously on the high seas.
Consequently a distributed protocol is utilized. The
technology used for this purpose relies on a protocol
called SOTDMA (Self Organized Time Division Mul-
tiple Access) [5], where the data stream is placed in
defined time slots. Depending on the observed data
traffic, each AIS module builds its own timetable for data
transfer and reserves free time slots for its messages. In
order to ensure synchronized time slots, a very precise
clock is needed. Since the navigation equation provides
position and time information, the GPS receiver assures
the required time accuracy [6]. Coming up to the length
and message rate, AIS slots are 256 bits long and it
supports 4500 reports per minute. Message rate depends
on the own speed, the values go from 2s for speeds over
23 knots, to 3 min for speeds under 3 knots, leading
to a density of around 400 ships in range. In order to
maintain the density below SOTDMA is able to reduce
the range radius in necessary.
2) Aviation: Despite its low throughput, TCAS MAC
protocol is based on Aloha. However, colliding transmis-
sions can be greatly diminished by means of different
techniques, i.e, interference limiting, passive detection,
altitude comparison, interrogation frequency, and direc-
tional antenna. Every TCAS unit monitors the number of
TCAS units within the detection range. This information
is used to limit its own interrogation rate and power as
necessary. In order to minimize the number of interro-
gations, the identity and altitude of targets is intended
to be determined by passively monitoring transmissions
received, afterwards, error detection and correction is
applied to the received messages to reduce the number of
erroneous addresses to be processed. Non-threat intruded
are dismissed through altitude comparison. Additionally,
TCAS units transmit an interrogation sequence nomi-
nally once per second for only those intruders that could
become immediate threats. Intruders that are not likely
to become immediate threats should be interrogated less
frequently. Differentiation is made by means of range
and estimated range rate. Finally, a directional antenna
sequentially generates beams that point in the forward,
aft, left and right direction. Together these beams provide
surveillance coverage for targets at all azimuth angles,
this way synchronous garble is reduced, i.e, the side
lobe signals are ignored. On the other hand, a directional
interrogation reduces the size of the interrogation region.
Thus, minimizing the number of transmission collisions.
Message length are 56 or 112 bits while the nominal
density is 30 aircraft in range.
Similarly to TCAS, ADS-B is based on Aloha. In this
case, interference limitation is as well applied. Since the
ADS-B extended squitter includes additionally position
and speed information, its length is 112 bits instead
the 56 bits of the TCAS squitter where, only identity
information is transmitted.
3) Road Transport: For C2C, a derivate of IEEE
802.11 standard and CSMA/CA (Carrier Multiple Ac-
cess with Collision Avoidance) protocol in the MAC
layer are proposed. One of the main limitations is the
large amount of data stored by the multiple sensors it
uses. Therefore, the message length is so large that the
network might collapse. In order to reduce the length of
the message, the inferred information of the combined
sensors’ output might be sent, instead of transmitting the
direct output of the sensor.
4) Applicability to RCAS: Very different approaches
are carried out for each one of the analyzed systems.
While AIS and C2C have well defined and standardized
MAC protocols, ADS-B and TCAS utilize an ”statis-
tical” approach. However, they all share a common
feature, the MAC protocols are distributed in order to be
managed autonomously. Consequently, the RCAS MAC
layer will be distributed.
The drawback of Aloha is its low throughput, it is
extremely sensitive in high density situations like in
shunting yards or train stations. However, this protocol
simplifies considerably the MAC layer.
Nonetheless, SOTDMA and CSMA/CA are not suit-
able protocols for RCAS. CSMA/CA uses the mech-
anism RTS/CTS (Request to send/ Clear to send) to
avoid the hidden terminal problem [16]. Although it is
specially designed for wireless networks, it does not
work properly in broadcast communication, since the
CTSs packets would collide, due to the non defined
number of recipient and the disparity of the reception
conditions between them. On the other hand SOTDMA
does not solve the hidden terminal problem and does not
assure absence of transmission collision as a contention
phase exists, when a newcomer enters the network and
tries to occupy a free time slot. Therefore, in the case two
newcomers are simultaneously entering and by chance
both occupy a free time block a transmission collision
occurs. This type of collision could hardly occur if the
ad-hoc networks are stable, i.e., the host’s situation in
the network changes slowly. On the other hand, if too
many hosts are leaving and entering continuously the
network, the number of packet collisions might increase
significantly. Since ships have a relative low speed
and course change, SOTDMA works properly in AIS.
Nonetheless, in RCAS, due to the potential high speed
of trains and the relative low communication range, the
network is more unstable and therefore, the protocol
might be not suitable.
Unlike C2C, no message length limitations are present
in RCAS, in a first approach. Moreover, the user density
is as well not such a restrictive parameter as in C2C.
In RCAS, the MAC layer design depends strongly
on the speed of the trains, parameters like message rate
and maximum density are inferred from this value. This
way, high speeds require a larger message rate, since the
position information should be updated more frequently.
Aircraft are able to move in any direction, having
more reaction possibilities, which leads to a lower
message rate compared to trains. However, the speed
of aircraft is higher. Therefore, the reaction time is
reduced, and consequently, the message rate is increased.
Summing up, the message rate for trains is expected to
be in the range of the one for aircraft.
IV. PHYSICAL LAYER
The physical layer (PHY) conforms the main differ-
ence among the systems. In the PHY layer such im-
portant parameters as frequency, modulation, bandwidth,
power, channel coding etc., are specified. In order to
design it, these parameters must provide the required
range and data rate which are given by the propaga-
tion channel characteristics, train speed, and necessary
breaking distance of the trains.
The range is defined as the maximum distance be-
tween transmitter and receiver so that a sufficient signal
to noise ratio is guaranteed at the input of the receiver.
The frequency f influences the decrease experimented
by the signal level due to distance, which is given by
the free space loss L
F S
equation
L
F S
= (
4πRf
c
)
2
(1)
where R is the distance to the transmitter and c is the
speed of light. A higher transmitted power guarantees
with a directly proportional relation a larger range.
On the other hand, the influence introduced by the
propagation channel and the protection given by the
utilized digital modulation scheme, define the noise level
at the receiver.
The upper bound of the data rate is related with the
Shannon-Hartley theorem,
C = B log
2
(1 +
S
N
) (2)
where B expresses the bandwidth. That means that
the amount of information an electromagnetic wave can
carry, is related to its bandwidth.
S
N
indicates the signal
to noise ratio at the output of the demodulator and
not at the input of the receiver like in the range case.
This signal to noise ratio is related with the bit error
rate (BER), i.e, a low signal to noise ratio leads to a
high BER, thus, decreasing the data rate. Otherwise,
high spectral efficient digital modulation increases the
data rate automatically due to larger bits per symbol
values. However, high spectral efficiency implies a larger
probability of error for the same noise level at the input
of the demodulator. Despite adding bits, channel coding
might correct the errors in the signal due to noise,
leading to an increment in the data rate.
1) Maritime Navigation: Table I summarizes the
most important features of the maritime AIS [7]. In
order to allow a backup frequency, to avoid interference
problems and to allow channels to be shifted without
communication loss from other ships, the communi-
cation is performed over two reserved VHF channels.
The data rate is 9.6 kbps, modulated over a nominal
bandwidth of 12.5 kHz. Due to the small available
bandwidth a modulation such as GMSK with a high
spectral efficiency is necessary. The nominal power is
12.5 W leading to a range depending on the antenna
height and the environmental conditions of 15 to 30
nautical miles.
TABLE I
AIS FEATURES
Frequency AIS1: 161.975 MHz
AIS2: 162.025 MHz
Bandwidth 25 or 12.5 kHz
Modulation GMSK FM
Power 12.5 W
Message length Variable.
Dynamic report 256 bits
Data Rate 9.6 kbps
Message Rate Minimum: 2 sec
Maximum: 3 min
User density Around 400
MAC Layer SOTDMA
Data Link Layer HDLC
Range 28-55 km
Attenuation characteristics Air, fog, rain, island
2) Aviation: As observed in Tables II and III, ADS-B
and TCAS parameters are very similar. TCAS utilizes
two different frequencies for the interrogation-answer
sequence, while due to the absence of interrogation,
ADS-B uses a single frequency.
TABLE II
ADS-B FEATURES
Frequency 1090 MHz
Bandwidth 10 MHz
Modulation Pulse position
Power 250 W
Message Length 112 bits
Data Rate 1 Mbps
Message Rate 0.4-0.6 sec
MAC Layer Interference limiting
Data Link Layer Parity check code for address
Range < 370 km
Attenuation characteristics Air, clouds
The radar system used in TCAS introduces a strong
limitation in range. For aircraft further away, the beam
of the antenna becomes wider making the measure
information less accurate. Therefore, despite the same
transmitted power, the range of both systems is very
different.
3) Road Transport: Due to its technological accessi-
bility and favorable cost, a technology based on standard
Wireless LAN is intended to be used in C2C [14], as
summarized in Table IV.
4) Applicability to RCAS: The suitable low AIS fre-
quency band could be reused for RCAS, since geograph-
ical separation would fulfill the interference require-
ments stated by the European Navigation Committee
(ERC). However special consideration should be taken in
