35
1 INTRODUCTION
The vast majority of the world’s goods flow across the
oceans. The shipping industry is, therefore, a major
driver of the global economy. The impact of
temporary breakdowns in shipping routes and supply
chains is illustrated by the recent incident in the Suez
Canal, which was blocked for days by the Ever Given
golden-class container ship. Although not caused by
cyber attacks, a precisely timed execution of an attack
could specifically provoke such incidents.
Threateningly, incidents attributable to cyber crime
are proliferating. Increasingly sophisticated, domain-
specific, and targeted attacks on maritime systems are
being observed [1, 30], making adequate cyber defense
strategies for this sector urgently necessary. In
particular, attacks aimed at misleading ship
navigation not only pose a serious risk from an
economic perspective with massive monetary
consequences due to disrupted maritime value chains,
but can also cause dangerous collisions of vessels and
endanger the environment and human lives. This is
why the International Maritime Organization (IMO)
placed cyber security on its roadmap and required
shipowners to establish cyber risk management by the
beginning of 2021 [17]. An operative implementation
of cyber risk management for shippers is also
supported by industry guidelines [5, 6]. Nevertheless,
studies reflecting the current state-of-the-art in cyber
security technology for the maritime domain conclude
that maritime systems are still highly vulnerable to
cyber attacks [7]. This is confirmed by recent
incidences, cf. [1].
One reason for high cyber risks correlates with the
peculiarities of maritime technology. Maritime
systems have a long life cycle. Although they were
originally designed for local (air-gapped) networks,
they are incrementally interconnected with public
interfaces. It has long been known that maritime
BRAT: A BRidge Attack Tool for Cyber Security
Assessments of Maritime Systems
C. Hemminghaus
1,2
, J. Bauer
1
& E. Padilla
1
1
Fraunhofer Institute for Communication, Wachtberg, Germany
2
University of Bonn, Bonn, Germany
ABSTRACT: Today’s shipping industry is largely digitalized and networked, but by no means immune to cyber
attacks. As recent incidents show, attacks, particularly those targeting on the misleading of navigation, not only
pose a serious risk from an economic perspective when disrupting maritime value chains, but can also cause
collisions and endanger the environment and humans. However, cyber defense has not yet been an integral part
of maritime systems engineering, nor are there any automated tools to systematically assess their security level
as well-established in other domains. In this paper, we therefore present a holistic BRidge Attack Tool (BRAT)
that interactively offers various attack implementations targeting the communication of nautical data in
maritime systems. This provides system engineers with a tool for security assessments of integrated bridge
systems, enabling the identification of potential cyber vulnerabilities during the design phase. Moreover, it
facilitates the development and validation of an effective cyber defense.
http://www.transnav.eu
the International Journal
on Marine Navigation
and Safety of Sea Transportation
Volume 15
Number 1
March 2021
DOI: 10.12716/1001.15.01.02
36
systems do not comply with state-of-the-art security
practices [3, 25, 27, 28].
However, we believe this is to some extent due to a
lack of technology to integrate cyber security into the
early stages of systems development, rather than
retrofitting and patching. Besides, the resilience of
maritime systems in case of cyber attacks is not
explicitly assessed and validated, neither in system
development nor in certification or operation.
Moreover, common cyber risk assessments typically
focus on external attacks that target at the availability
of different components of maritime systems [30].
However, the class of internal cyber attacks should
not be ignored. Physical access is almost impossible to
prevent in practice. Changing personnel onboard
ships as well as long cable harnesses between
distributed electronics result in a large attack surface
for internal attacks. Such attacks are inherently more
powerful and harmful than external attacks from
cyber and physical space. Nevertheless, they are
rarely considered in risk analyses and are not
supported by automated tools in the context of cyber
defense. Overall, in contrast to other domains, a true
maritime-specific security tool that also considers
application-level properties is still missing.
To close this gap, we present BRAT, a holistic and
user-friendly BRidge Attack Tool. It is designed as a
modular framework that interactively offers various
implementations of cyber attacks targeting the
communication of nautical data in Integrated Bridge
Systems (IBSs). Those attacks can be individually
configured, combined, scheduled, and orchestrated in
an automated manner. To the best of our knowledge,
BRAT is the first domain-specific tool for internal
network attacks against maritime systems. It has the
potential to improve security assessments and allows
system engineers to systematically identify and verify
vulnerabilities in a large number of components.
Furthermore, BRAT also enables the development and
validation of appropriate cyber attack
countermeasures, such as effective detection or
preventive authentication. In addition, it is possible to
integrate BRAT into the Maritime Education and
Training (MET) of bridge crews, thereby increasing
awareness and improving their response to cyber
threats [11, 12].
The core contributions of this paper are twofold:
First, we extend the common threat landscape of
maritime systems to include the class of internal cyber
attacks. Additionally, we develop an attack model for
this kind of threat. Second, based on this attack model,
we introduce a bridge attack tool as a comprehensive
framework to launch domain-specific cyber attacks
against IBSs having the potential to further advance
cyber security in the maritime domain.
The remainder of this paper is structured as
follows. Section 2 gives a brief background on
maritime systems, communication protocols, and
cyber security testing. In Section 3, internal cyber
attacks on IBS are placed in the overall context of
maritime cyber threats and an attack model
implemented by our approach is derived. The design
concept of BRAT and its architecture are introduced in
Section 4, while Section 5 provides a detailed focus on
BRAT’s attack features and implemented attacks
realized in our reference implementation. In Section 6,
we then revisit the main use cases of our attack tool
and reflect on the results of our approach in a security
discussion. A conclusion is finally given in Section 7.
2 BACKGROUND
This section first introduces the main electronic
components of maritime systems. The focus is on
sensors, which are the interface to the environment
and are, therefore, particularly threatened by external
attacks. Also, processing systems, which can be
deceived by false sensor data, are discussed. Then, the
communication protocols for transmitting nautical
information are introduced, which offer poorly
protected attack surfaces. Finally, we briefly review
the current state of security tools and use this to
motivate our bridge attack tool.
2.1 Maritime Systems
Figure 1. An exemplary architecture of a typical maritime
system onboard commercial vessels, cf. [4, 20, 27]. Due to
lack of network security features, the maritime system is
vulnerable to Person-on-the-Side attacks from injected
devices (PotS1) or compromised sensors (PotS2) and Person-
in-the-Middle attacks (PitM) from manipulated
communication paths.
Typical maritime systems onboard vessels are
heterogeneous and distributed IT systems with a
plurality of different sensors and actuators. Therefore,
they can be understood as CPSs connecting the cyber
and physical space. They consist of technical
measurement instruments and nautical processing
assets, along with systems for external communication
and alarms, as schematically shown in Figure 1.
Maritime sensors gather nautical data to provide
an accurate navigational situation picture, which is
the base for navigational decision-support. Typical
sensors are Global Navigation Satellite System (GNSS)
receivers, primarily GPS, radars, echo sounders, and
compasses with speed logs. More sophisticated
sensors not only measure but interpret incoming data,
such as maritime radars that often interpret the
information stream to recognize and track individual
objects. The data gathered by sensors is usually
provided via serial interfaces, connected by sensor
integration units that often bundle multiple sensors,
as exemplarily shown in Figure1.
Onboard processing systems handle sensor data
for either to derive a navigational picture or to control
specific actuators such as rudder and propulsion.
These systems are interconnected into IBSs, which
usually comprise an Electronic Chart Display and
Information System (ECDIS) for route planning and
monitoring, a conning and a radar display as well as
37
an autopilot. The Voyage Data Recorder (VDR) is a
mandatory equipment for the continued logging of
sensor and system states.
External communication systems are used for
different purposes. Automatic Identification System
(AIS) transceivers exchange messages between other
maritime systems in the vicinity for safety and
security reasons, including their identity, position,
and course information. Incoming AIS messages are
usually presented as targets on the ECDIS. Very-
Small-Aperture Terminals (VSATs) are used for
digital welfare communication, but also navigational
needs such as weather and chart updates or route
optimization. Further navigational assistance is
provided by onboard alarm systems. Bridge
Navigational Watch and Alarm Systems (BNWASs)
notify crew members when the current officer on
watch is unresponsive or unable to perform duties. A
central point to monitor and handle alarms from
various bridge sources is the Bridge Alert
Management System (BAMS), which alarms for
unsafe navigation and also monitors sensor integrity
to some extent.
2.2 Maritime Communication Protocols
Save navigation builds upon a reliable interconnection
between bridge components. Hence, the electronic
exchange of nautical data is since long standardized.
A common standardization is the NMEA 0183
interface standard defined by the National Marine
Electronics Association (NMEA) in the 1980s. In this
ASCII-based exchange protocol, nautical information
is encoded to so-called NMEA sentences, which
consist of talker and sentence identifiers, followed by
payload fields and a checksum. Nowadays, NMEA
sentences are widely established as an encoding
format even beyond shipping.
As part of maritime digitization, modern networks
use Ethernet technology as transmission medium.
There are three common maritime communication
protocols based on the IP-stack: Lightweight Ethernet
(LWE), NMEA over IP, and NMEA OneNet. The
former is a modern communication protocol that was
standardized a decade ago with IEC 61162-450 [15]. It
is based on the UDP/IP-stack and uses IPv4 multicast
with individual receiver groups according to the
equipment type. Although the physical
communication medium changed to the faster
Ethernet, NMEA sentences are still used and
encapsulated in UDP datagrams. But LWE also allows
the transmission of larger files in a specific binary
format. NMEA over IP [26] is a predecessor of LWE,
still in use. This protocol is basically a direct portation
from the original NMEA 0183 to the IP stack using
UDP or TCP. As the protocol is not standardized, the
explicit implementation varies between different
manufacturers.
At the end of 2020, the first version of the NMEA
OneNet standard has been published. OneNet is
based on the UDP/IPv6-stack and, in contrast to LWE,
supports additional datagram and application
security measures. However, as the standardization
has just recently been finalized, its market coverage is
still very limited. For this reason, our focus is placed
clearly on LWE standardized in IEC 61162-450 as the
prevailing maritime communication protocol.
2.3 Cyber Security Testing
Cyber security is often neglected in maritime systems
engineering, leading to inherently insecure systems.
Since update cycles and certification processes also
take a long time, fixes for identified vulnerabilities do
not ship in time. Furthermore, with current methods
for testing it is not possible to verify that a weakness
causing a vulnerability is actually remedied. Hence,
already fixed vulnerabilities may reoccur after system
updates or new vulnerabilities may arise from the
same weakness. In the maritime domain, this results
in a high effort for manufacturers and shipowners in
updating, certifying, and deploying new versions of
systems after security weaknesses were identified. To
reduce the costs, necessary updates to patch security
vulnerabilities are left out, which significantly
increases the overall cyber risk.
As known from other industries, appropriate tools
for comprehensive security testing have an immense
benefit. Particularly in the domain of secure software
development, automated or semi-automated security
testing is a common technique during system
development [8]. But security testing has also found
its way into a wide range of other domains, e.g.,
industrial control systems [23] and automotive
vehicles [14]. However, in the maritime context, so far
only generic tools have been used to support secure
systems engineering or identify vulnerabilities.
Sviličić et al. used a generic vulnerability scanner
(Nessus) to assess typical maritime components [27
29]. Although their analyses reveal flaws in the
systems’ underlying operating systems, their
assessments do not take into account the maritime-
specific context, i.e., the actual maritime applications,
data types, and communication protocols. Regarding
network security, well-established generic tools like
bettercap and yersinia facilitate the testing of common
protocol vulnerabilities in the IP stack. On the other
hand, several Proofs-of-Concepts (PoCs) on the
exploitability of application-level vulnerabilities in
maritime systems already exist [3, 4, 20, 22]. However,
these PoCs cannot be integrated into system
development in an automated manner.
Overall, the maritime sector lacks a true maritime-
specific framework that provides adequate tools for
testing and assessing cyber security at the application
level. Concerning the plurality of components and
their interconnection within maritime systems,
mentioned in the previous subsections, there is also a
large number of attack vectors that need to be
addressed by such a framework. The vulnerabilities,
from which these attack vectors emerge, will be
analyzed in the following section.
3 ANALYSIS OF MARITIME VULNERABILITIES
As Cyber-Physical Systems (CPSs), maritime systems
are threatened by attacks from both, physical and
cyber space. On the one hand, some maritime sensors
38
use electromagnetic signals to operate and, thus, are
vulnerable to Electronic Warfare (EW). This category
of attacks aims to deny the system’s access to the
electromagnetic spectrum which usually results in a
Denial of Service (DoS). Maritime sensors which are
known to be prone to EW attacks are GPS, AIS, and
radar, cf. [30]. On the other hand, cyber attacks target
the information processed in IT systems. They can be
categorized by the exploited target which can be a
technology, a human, or an organizational process,
and by violation of the protection goals availability,
integrity, and confidentiality. Our main focus in this
paper lies on cyber attacks targeting maritime
technology and threatening the availability and
integrity of data. This initially excludes supply chain
and phishing attacks on employees and suppliers.
Nevertheless, cyber attacks that build on successful
attacks from these categories, e.g., previously
compromised IT systems or corrupt staff members,
were considered. The confidentiality of data is
neglected because nautical data is rarely classified in
civil use cases.
In our threat analysis, we further distinguish
between external and internal cyber attacks
depending on the attacker’s access to the vessel’s
components. External cyber attacks assume the
attacker has access to the communication medium of
external interfaces. In the maritime context, this can be
a public-facing network interface, a USB port, or even
an antenna for radar, VHF, or GPS frequency
spectrum. Compared to EW attacks, external cyber
attacks do not exclusively target the spectrum, but
rather the transmitted information, such as sending
fake AIS messages or seemingly valid but interfering
GPS signals. Furthermore, they can already be
realized with low monetary costs, i.e., less than $400
for VSAT attacks [22] or $2000 for a GPS spoofing
[24].
Internal cyber attacks require that the attacker has
access to one or more IT components within the
internal network. For instance, this can be achieved by
a previously compromised component (e.g., via
malware) or by injecting an additional malicious
device into the network. Compromised components
usually have a worse impact on the security of the
entire system than external attacks, because the data
they provide is generally more trusted by peers.
In [30], Tam and Jones propose a model-based
framework for maritime cyber-risk assessment. The
authors consider vulnerabilities of both categories,
EW (VHF, radar) and external cyber attacks (USB,
Internet, satellite). The effects caused are grouped as
violations of the protection goals of availability (DoS)
and/or integrity (misdirect). Based on this work, our
analysis extended the risk assessment to include cyber
vulnerabilities of internal cyber attacks, which are
neglected as a separate class also in other relevant risk
analyses in the literature. Additionally, relevance
indices for protecting these systems based on recent
incidents according to [1] were added, where
applicable. Furthermore, we also include additional
maritime system components that are expected to be
vulnerable to this new category of attacks, as
highlighted in Table 1. As the extension of the original
table provided by [30] reveals, internal cyber attacks
are expected to expose an additional class of risks to
the reliability of maritime systems (emphasized by
colored entries in the table). The availability and
integrity of these systems may be limited due to a
high rate of incoming messages or intentionally
manipulated messages sent by a malicious device, for
instance. As it has been demonstrated that cyber
attacks can have a huge impact on the navigational
process, e.g., by distorting the crew’s situational
picture and provoking a grounding [4, 10, 19, 20],
internal cyber attacks must necessarily be taken into
account in systems engineering and MET.
Table 1. Internal cyber attacks extend attack vectors on
maritime systems according to threat landscape in [30],
resulting to further cyber effects, as highlighted in the table.
They also have impacts on further components, added in
the bottom rows (separated by the dashed line). Systems are
ranked by relevance based on previous incidents according
to [1].
3.1 Attack Model for LWE
In the case of network attacks, internal attackers can
interact with communication by eavesdropping and
injecting self-crafted messages. Attackers having these
capabilities are called Person-on-the-Side (PotS). If in
addition, attackers are able to suppress legitimate
messages in existing communication, they are called
Person-in-the-Middle (PitM). Note that in the former
case, attackers cannot simply prevent the receiving of
original messages.
Although PitM attacks can cause more damage and
are more difficult to detect, they are also more
difficult to execute. PotS attacks, on the other hand,
are more realistic and, thus, more likely to occur. In
Ethernet networks, a PotS attack can potentially be
performed by any compromised component in the
maritime system, whereas a PitM attack requires the
attacker to obtain a special position in the network,
e.g., directly between a sensor’s connection to the
ECDIS. Both types of attacks could be performed by
injecting a malicious sensor, a malicious device
(which can be a sensor or a tiny system-on-a-chip
device like a Raspberry Pi [21]), or by compromising
an existing asset [20]. Our attack model explicitly
includes both types of internal cyber attacks, which
are integrated into Figure 1 as examples for
illustration.
39
Similar to other established protocols, LWE was
not designed to be secure against PitM and PotS
attacks. It does not implement supplementary security
measures, despite an optional Message Authentication
Code using the insecure MD5 hash algorithm by
default [13, 15]. However, with the standard
IEC 61162-460 [16], new measures to face network
security are proposed such as network segregation,
firewalls, network access control, and security
monitoring. Although this mitigates external and
internal cyber attacks, IEC 61162-460 is currently not
mandatory and its widespread implementation will
take several years. Thus, PitM and PotS attacks will
remain relevant in practice.
Figure 2. Test- and development environment with sensor
input (either using an NMEA/IEC simulator or real
maritime electronics) and data consumers, particularly the
ECDIS as major HMI device processing and visualizing
incoming sensor data.
4 DESIGN AND CONCEPT OF BRAT
The concept of our framework is based on the attack
model from the previous section and covers PitM and
PotS cyber attacks on maritime networks. BRAT is
designed to be integrated into generic development
environments together with the maritime application
under test. Our reference environment is depicted in
Figure 2 and by default provides an IBS consisting of
usual components introduced in Section 2.1. A
simulator provides data of general maritime systems
via LWE and comprises simulations of common
sensors, such as GNSS, compass, and speed logs, but
also a simulated AIS transceiver for external nautical
information. Our environment can be further
extended by real hardware components using
Ethernet or serial interfaces, which make it possible to
seamlessly connect different IBSs, an autopilot, or
other maritime equipment.
Attacks against the IBS and other systems in the
development environment are conducted by BRAT
that is implemented as an ordinary network device. It
can launch pre-configured attacks automatically for
automated testing or manually be controlled by an
operator. For semi-automatic attack execution, attacks
can be configured and launched interactively using a
graphical Human-Machine Interface (HMI) that is
connected via an out-of-band communication channel,
e.g., WiFi.
In the setup outlined in Figure 2, BRAT
implements PotS attacks against the communication
of nautical data in the maritime network. These
attacks are performed by capturing legitimate traffic
in the network and injecting modified nautical
messages that are processed by receivers. For instance,
BRAT can inject position-related messages with a
relative deviation causing a distorted situation
picture. Note that, despite our focus on PotS attacks
and LWE in the following sections, BRAT can also be
used for PitM attacks as well as for maritime systems
supporting NMEA over IP.
4.1 Architecture
Figure 3. The data flow between BRAT’s main components.
The architecture of BRAT consists of four main
components whose interaction is visualized in
Figure 3. The sniffer captures and filters maritime
traffic from the network and provides that data to the
tracker. To use BRAT in automated testing of specific
scenarios, the sniffer can be configured to use pre-
recorded traffic from a file (e.g., in pcap format). The
tracker maintains the current navigational state
derived from captured packets and stores a history in
a database. The attacker module implements the
actual attack logic and has access to the database.
Based on its configuration and database input,
malicious traffic can be generated and finally injected
into the network by the sender. Wrappers for nautical
payloads provide a convenient language to
manipulate nautical data deliberately, e.g., shift the
position by the command pos.lon += 1.0. Additional
attacks can effortlessly be added to the framework
from scratch or by adjusting already implemented
attacks (cf. Section 5.2). A user interface allows for
configuring, launching, combining, and orchestrating
attacks programmatically or interactively. Malicious
traffic can also be stored in a file from which it can be
replayed or analyzed later.
4.2 Design Principles
Figure 4. BRAT provides a user-friendly HMI to configure,
launch, combine, and orchestrate attacks against the
maritime system.
Our design principles include usability,
modularity, and reliability. To enhance usability and
lower the technical barrier so that also non-technical
users can perform security assessments, BRAT comes
40
with three tailored interfaces. Firstly, it is possible to
use a Python package which makes it easy to
investigate security features of maritime systems in an
automated test and development environment.
Secondly, a command-line interface can be used to
interactively launch attacks in security assessments
either from scratch or using scripts. Finally, an
interactive web-based HMI as depicted in Figure 4
facilitates the use for non-technical users. As use cases
and systems change, an attack framework has to be
extendable, adjustable, and reusable. Therefore, our
architecture follows the modularity principle. In this
context, a plugin system is implemented and the
functionality of the connection to the maritime
network is decoupled from the attack logic. That
allows to add customized attacks and support other
maritime protocols. Since a security assessment is
based on reliable data processing, our framework was
implemented together with a comprehensive test suite
and logging functionality. This assures dependable
application behavior and allows to effortlessly verify
modified code sections.
5 IMPLEMENTATION OF ATTACKS
After explaining BRAT’s architecture and design
principles in the previous section, the following
section first introduces BRAT’s attack features and
provides an overview on possible attacks. Then, it
highlights two special attacks to demonstrate the
potential of our approach.
5.1 Attack Features
Attacks implemented in BRAT aim to alter the
presumed and determined navigational state at the
receiver. This is achieved by superimposing legitimate
traffic by attacker-controlled traffic. To increase the
chance of a successful attack, BRAT supports several
attack features that are listed in Table 2.
Table 2. Attack features implemented by BRAT to increase
the chance of a successful cyber attack.
The first feature, network traffic capturing,
eavesdrops on legitimate packets. From the
communicated data, an adversary can estimate the
current navigational state to adjust the attacks.
Additionally, BRAT can perform replay attacks,
which use formerly recorded packets to be resent at a
later point in time. Since the original packets were
generated by a legitimate source, they are more likely
to be accepted by receiving systems. A minimal need
for configuration makes these attacks easy to
automate.
Instead of just repeating traffic, BRAT can alter
packets before emitting them again into the network
(injection of malicious packets). It is possible to
slightly modify the payload of original packets, which
may not attract the attention of the navigator but still
impacts navigational decisions. A fixed deviation of
the data would be rather obvious. Therefore, BRAT
implements continuous attacks with an incrementally
increasing variation over time to further obfuscate
attacks.
Since modern IBSs support integrity checks of
sensor data, the modification of a single nautical data
type may be detected by default. For instance, a bogus
deviation from the real position may not resonate
with the estimated position from the inertial
measurement unit causing the integrity check to raise
an alert. To circumvent the detection, BRAT supports
parallel attacks targeting different sensors
simultaneously to enhance the pretended integrity of
malicious packets.
As already observed in [19], a high frequency of
malicious packets further facilitates cyber attacks.
Depending on the actual system, legitimate packets
may be discarded or presumed to contain measuring
errors, if there are far more packets with nautical data
modified by the attacker. Also, the sampling rate of
the processing systems may be too slow to process
sparsely received benign packets. With BRAT, it is
possible to configure the frequency of sending
malicious packets.
When standard authentication measures are in
place, an active attacker on an IP-based network
would rapidly be detected by revealing an
unregistered source address. Therefore, BRAT can
perform MAC/IP/UDP spoofing, i.e., leave packets up
to the transportation layer intact or spoof the identity
of explored systems. The ability to arbitrarily change
these network addresses can further complicate the
identification of malicious packets in existing systems.
5.2 Implemented Attacks
As mentioned in Section 3.1, BRAT implements PitM
and PotS attacks on the exchange of nautical data in
LWEs. According to the attack features from Table 2,
cyber attacks can be configured and combined in
several ways. For instance, attacks on the availability
of the target system may be implemented by using a
high frequency of replayed packets for DoS attacks,
e.g., by overcharge the receiver or leveraging special
malformed packets to trigger parsing errors and crash
the receiver. Attacks on the integrity may
continuously shift nautical data in multiple messages
to spoof several sensors simultaneously and remain
undetected by possible integrity checks. NMEA
sentences can be parsed, modified, and retransmitted
by BRAT. Since many maritime systems support
NMEA sentences as input format, our approach can
be applied to environments beyond LWE and also to a
variety of components, including maritime processing
41
systems (cf. Table 1) as briefly discussed in the
following.
ECDIS, conning, and VDR display and log the
current navigational situation and, thus, highly
depend on reliable sensor inputs. Interrupted or
manipulated nautical data can lead to incorrect
estimations and cause the navigator to take wrong
and harmful decisions. BRAT’s attacks targeting
autopilots can send commands, overrule manual
steering, and set an intended course. Depending on
the autopilot’s implementation, it may be possible for
an attacker to hijack the autopilot and, thus, to
remotely control the vessel. Onboard alarm systems
use NMEA sentences to exchange alert information.
Hence, BRAT can intercept, manipulate, and suppress
raised alerts, but also fake alerts can be generated at a
high frequency, e.g., to distract the navigator from
forthcoming hazards. Radars use NMEA sentences to
deliver located tracks of maritime objects to other
devices. As with alarms, these tracks can be
intentionally tampered with or injected. Then again,
some systems may use traditional IP-based protocols
to broadcast radar images, which can be tampered
with by non-NMEA-based BRAT attacks. Without loss
of generality, in the following, we demonstrate the
effectiveness of BRAT’s attacks on maritime systems
documenting launched attacks against OpenCPN, an
open-source chart plotter. Even though it is a free
ECDIS that supports only basic functionalities, we
would like to emphasize here that we have also tested
our attack tool with commercial and well-established
products and have come to similar results.
5.3 AIS Attacks
AIS attacks can be considered from two perspectives.
On the one hand, AIS transceivers consume onboard
position and heading information and provide this
data to other traffic participants. Indirect attacks on
positioning sensors could, thus, imply that
transceivers spread incorrect information, which can
have a wide variety of effects. On the other hand, a
successful AIS attack will result in wrong positions or
hidings of targets on the ECDIS, manipulated
dimensions of targets, or flooding of “ghost" targets.
There are already proposals for securing AIS [2, 9, 18],
but an exhaustive implementation is still a long way
off. Hence, those systems processing incoming AIS
data have to challenge the reliability of that
information and implement countermeasures
themselves.
BRAT can audit whether an ECDIS application is
vulnerable to AIS attacks by simulating a manipulated
AIS transceiver and analyzing the application’s
behavior. Figure 5 shows a successful AIS attack
targeting an ECDIS application, which is intentionally
obvious for demonstration purpose and inspired by
[3]. BRAT manipulates the information sent from the
AIS transceiver to flood the chart plotter with fake
targets on collision course. Attack options can be
adjusted to either hide real targets in a vast number of
fake targets or decrement the targets to make the
attack less obvious. An unaware navigator may start
an evasion maneuver and risk grounding in shallow
water.
Modern ECDIS implement radar overlays which
makes it possible to validate AIS targets and detect
attacks through cross-checking. However, depending
on the particular integration of radar tracks, BRAT can
circumvent the detection by combining AIS attacks
with compatible radar attacks. A more generic
approach to disguise fake targets and hide real targets
despite radar overlay is subject to our future work.
Figure 5. BRAT can launch AIS attacks on the target ECDIS,
e.g., to flood the screen of OpenCPN with numerous ghost
targets on collision course.
5.4 Positioning Attacks
By its very definition, navigation relies on Positioning,
Navigation, and Timing (PNT) data. GPS receivers are
the central sensors providing not only position and
time but also speed and course information. Internal
cyber attacks on PNT data have been shown feasible
by Lund et al. [20]. Their attack aims to shift the
position of the own ship on the ECDIS to make the
navigator presumably correct the displayed position,
which leads to leaving the actual course. BRAT can
simulate such attacks, which is crucial to develop and
validate cyber security corresponding
countermeasures, but can also be integrated into MET.
In Figure 6, the original course displayed by the
ECDIS is shifted and injected fake AIS targets
fabricate a collision course. In contrast to [20], this is
conducted by PotS attacks using an injected device
instead of presuming the presence of malware on the
ECDIS system as outlined in [19].
Figure 6a. OpenCPN displays the vessel’s original course.
42
Figure 6b: A cyber attack fabricates a collision course.
Figure 6. PotS attacks on PNT data distort the navigational
situation displayed by OpenCPN to fabricate a collision
course.
6 SECURITY DISCUSSION
Table 3. Internal cyber attacks (direct) and external cyber
attacks (simulative) implemented by BRAT. Due to
proprietary radar protocols, simulative radar attacks are
only implemented exemplarily, indicated by the bracketed
checkmark.
In Section 3, we discussed internal and external
attack vectors for maritime systems. BRAT
implements attacks addressing these attack vectors.
An overview of already implemented attacks on
maritime systems is given in Table 3.
Internal cyber attacks use the network interface to
attack various system components. Using BRAT
complex attacks can be executed targeting AIS,
ECDIS, VDR, RADAR, autopilot, conning, and
BNWAS/BAMS. With a specially configured PitM
attack, BRAT enables full control over the final output
of maritime sensors, which makes it further possible
to create mock-ups for externally attacked or
compromised devices. Thus, external attacks can be
“simulated” for all data producers, including AIS,
GNSS, speed logs, echo sounder, compass, and
BNWAS/BAMS, cf. Table 3.
With respect to related work, it should be
emphasized that BRAT is capable of reproducing
published cyber attacks, either by executing internal
or simulating external attacks. In [19], an attack
targeting an ECDIS is proposed. Basically, it uses the
injection of malicious packets at a high frequency to
attack the integrity of PNT data. This can also be
conducted by our approach, as already demonstrated
in Figure 6. External attacks targeting AIS are
described in [3]. As shown by Figure 5, AIS
information displayed in the ECDIS can be arbitrarily
manipulated. In this way, the reception of fake AIS
data can also be simulated at the same time. Whereas
direct attacks on vessels’ GNSS receivers cannot be
realized by internal cyber attacks, external attacks,
particularly GPS spoofing attacks as in [4], are enabled
by our tool, as well. For this purpose, internal PitM
attacks can mimic the effects of their external
counterparts.
As an interactive attack generation framework,
BRAT contributes to maritime cyber security in the
areas of security assessment, system resilience, attack
detection, and security training. BRAT’s attacks can be
used in penetration tests to assess the cyber security of
maritime systems. It uses a common communication
interface to attach new systems, provides an
interactive HMI to exploratory test launch and
combine attack modules, and a scriptable interface to
reproducibly trigger identified vulnerabilities. Once
defined, those scripts can be automated and
integrated into system development cycles to test
against vulnerabilities in current and future system
versions. This makes it possible to validate the
functional behavior and resilience of maritime
systems under cyber attacks as part of system
development and certification.
As cyber attacks cannot be fully prevented,
maritime systems and crews have to be accordingly
prepared. A crucial point is the detection of possible
attempts on time to prevent the system from further
damage. Therefore, security mechanisms, so-called
Intrusion Detection Systems, have to be integrated
that not only detect cyber attacks but also provide
helpful instructions for the navigator to remedy the
damage caused by the attack or navigate safely
despite flawed systems. BRAT can significantly
improve both the development of detection methods
and the training of mariners under realistic
conditions.
6.1 Outlook
So far, BRAT’s underlying attack model exploits
security weaknesses in the LWE protocol to
manipulate the communication of nautical data in
maritime systems using PitM and PotS attacks.
However, there are complementary onboard protocols
to exchange information between IT systems,
especially for radar images, chart updates, and
automation control. Although not necessarily
standardized, some protocol implementations used in
operation may reveal flaws similar to LWE, which
makes it possible to adapt BRAT to other devices. For
instance, the exchange of recorded images in modern
radars is based on Ethernet that is vulnerable to the
same attacks as LWE unless further security measures
are in place. BRAT could, therefore, be enhanced to
manipulate those radar images, i.e., inject, replace,
delete tracks, by means of internal cyber attacks.
Currently, BRAT’s attack model focuses on IBSs
onboard commercial vessels. A key assumption in this
context is that there is always a human in the loop
assessing the navigational situation and validating
incoming nautical data. Hence, attacks so far try to
43
obfuscate malicious behavior primarily in the
displayed information to finally decoy the navigator.
However, there are other environments in which
visual obfuscation is of minor importance.
Autonomous surface or underwater vehicles are
designed to operate mostly unmanned, eliminating
the need and the opportunity for human validation. In
consequence, incoming sensor data has an automated
impact on how onboard actuators are controlled,
which increases not only the likelihood but also the
possible damage of a successful cyber attack on the
maritime system.
7 CONCLUSION
In this paper, we extended the current threat
landscape of maritime systems by internal cyber
attacks against integrated bridge systems, which aim
to tamper with the communication of nautical data
and are usually neglected in existing cyber risk
assessments. Moreover, we introduced a BRidge
Attack Tool (BRAT) that, to the best of our
knowledge, is the first maritime-specific security tool
that enables the interactive launch of numerous PitM
and PotS cyber attacks. BRAT supports various
common network attack features, including packet
capturing, replay, and injection attacks along with
classical identity spoofing. It can be deployed in
common development environments which
implement (simulated) sources for nautical data and
are compatible to LWE. Thus, it greatly supports
existing processes to technically assess, prevent, and
detect cyber attacks on maritime systems by using
offensive security methods. In addition, Maritime
Education and Training can benefit from BRAT as
navigators can be trained to adequately react to cyber
attacks in realistic scenarios. By using BRAT, we
further demonstrated how internal cyber attacks can
violate the availability and integrity of common
onboard systems and exemplarily highlighted their
impacts with regard to AIS and GNSS attacks
targeting an ECDIS.
As part of our future work, we plan to extend
BRAT’s range of applications to support further
maritime system interfaces for radar images, chart
updates, and automation control. Also, we will widen
the context to investigate cyber attacks on
autonomous systems.
REFERENCES
1. Awan, M.S., Al Ghamdi, M.A.: Understanding the
Vulnerabilities in Digital Components of an Integrated
Bridge System (IBS). Journal of Marine Science and
Engineering. 7, 10, (2019).
https://doi.org/10.3390/jmse7100350.
2. Aziz, A., Tedeschi, P., Sciancalepore, S., Pietro, R.D.:
SecureAIS - Securing Pairwise Vessels Communications.
In: 2020 IEEE Conference on Communications and
Network Security (CNS). pp. 19 (2020).
https://doi.org/10.1109/CNS48642.2020.9162320.
3. Balduzzi, M., Pasta, A., Wilhoit, K.: A Security
Evaluation of AIS Automated Identification System. In:
Proceedings of the 30th Annual Computer Security
Applications Conference. pp. 436445 Association for
Computing Machinery, New York, NY, USA (2014).
https://doi.org/10.1145/2664243.2664257.
4. Bhatti, J., Humphreys, T.E.: Hostile Control of Ships via
False GPS Signals: Demonstration and Detection.
Navigation. 64, 1, 5166 (2017).
https://doi.org/10.1002/navi.183.
5. Bimco: The Guidelines on Cyber Security Onboard
Ships, https://www.bimco.org/about-us-and-our-
members/publications/the-guidelines-on-cyber-security-
onboard-ships, last accessed 2021/04/19.
6. BSI: IT-Grundschutz Profile for Shipping Companies -
Minimum Protection for Ship Operations,
https://www.bsi.bund.de/SharedDocs/Downloads/EN/B
SI/Grundschutz/profiles/Profile_for_Shipping_Compani
es_Minimum_Protection_for_Ship_Operations.pdf, last
accessed 2021/04/19.
7. ENISA: Cyber security aspects in the maritime sector,
https://www.enisa.europa.eu/publications/cyber-
security-aspects-in-the-maritime-sector-1, last accessed
2021/04/19.
8. Felderer, M., Büchler, M., Johns, M., Brucker, A.D., Breu,
R., Pretschner, A.: Chapter One - Security Testing: A
Survey. In: Memon, A. (ed.) Advances in Computers. pp.
151 Elsevier (2016).
https://doi.org/10.1016/bs.adcom.2015.11.003.
9. Goudosis, A., Katsikas, S.: Secure AIS with Identity-
Based Authentication and Encryption. TransNav, the
International Journal on Marine Navigation and Safety
of Sea Transportation. 14, 2, 287298 (2020).
https://doi.org/10.12716/1001.14.02.03.
10. Hassani, V., Crasta, N., Pascoal, A.M.: Cyber Security
Issues in Navigation Systems of Marine Vessels From a
Control Perspective. In: OMAE2017. , Volume 7B: Ocean
Engineering (2017). https://doi.org/10.1115/OMAE2017-
61771.
11. Heering, D.: Ensuring Cybersecurity in Shipping:
Reference to Estonian Shipowners. TransNav, the
International Journal on Marine Navigation and Safety
of Sea Transportation. 14, 2, 271278 (2020).
https://doi.org/10.12716/1001.14.02.01.
12. Heering, D., Maennel, O.M., Venables, O.M.:
Shortcomings in cybersecurity education for seafarers.
Presented at the 5th International Conference on
Maritime Technology and Engineering , Lisbon,
Portugal (2020).
13. Hemminghaus, C., Bauer, J., Wolsing, K.: SIGMAR:
Ensuring Integrity and Authenticity of Maritime
Systems using Digital Signatures. Presented at the
ISNCC-TSP (2021).
14. Huang, T., Zhou, J., Bytes, A.: ATG: An Attack Traffic
Generation Tool for Security Testing of In-Vehicle CAN
Bus. In: Proceedings of the 13th International Conference
on Availability, Reliability and Security. Association for
Computing Machinery, New York, NY, USA (2018).
https://doi.org/10.1145/3230833.3230843.
15. IEC 61162-450:2018: Maritime navigation and
radiocommunication equipment and systems Digital
interfaces Part 450: Multiple talkers and multiple
listeners Ethernet interconnection. (2018).
16. IEC 61162-460:2018: Maritime navigation and
radiocommunication equipment and systems Digital
interfaces Part 460: Multiple talkers and multiple
listeners Ethernet interconnection Safety and
Security. (2018).
17. International Maritime Organization: Guidelines on
Maritime Cyber Risk Management MSC-FAL.1/Circ.3.,
https://www.imo.org/en/OurWork/Security/Pages/Cyber
-security.aspx, last accessed 2021/04/19.
18. Kessler, G.C.: Protected AIS: A Demonstration of
Capability Scheme to Provide Authentication and
Message Integrity. TransNav, the International Journal
on Marine Navigation and Safety of Sea Transportation.
14, 2, 279286 (2020).
https://doi.org/10.12716/1001.14.02.02.
44
19. Lund, M.S., Gulland, J.E., Hareide, O.S., Jøsok, .,
Weum, K.O.C.: Integrity of Integrated Navigation
Systems. In: 2018 IEEE Conference on Communications
and Network Security (CNS). pp. 15 (2018).
https://doi.org/10.1109/CNS.2018.8433151.
20. Lund, M.S., Hareide, O.S., Jøsok, Ø.: An Attack on an
Integrated Navigation System. Necesse. 3, 2, 149163
(2018). https://doi.org/10.21339/2464-353x.3.2.149.
21. Michalas, A., Murray, R.: Keep Pies Away from Kids: A
Raspberry Pi Attacking Tool. In: Proceedings of the 2017
Workshop on Internet of Things Security and Privacy.
pp. 6162 Association for Computing Machinery, New
York, NY, USA (2017).
https://doi.org/10.1145/3139937.3139953.
22. Pavur, J., Moser, D., Strohmeier, M., Lenders, V.,
Martinovic, I.: A Tale of Sea and Sky On the Security of
Maritime VSAT Communications. In: 2020 IEEE
Symposium on Security and Privacy (SP). pp. 13841400
(2020). https://doi.org/10.1109/SP40000.2020.00056.
23. Pfrang, S., Borcherding, A., Meier, D., Beyerer, J.:
Automated security testing for web applications on
industrial automation and control systems.
Automatisierungstechnik. 67, 5, 383401 (2019).
https://doi.org/10.1515/auto-2019-0021.
24. Psiaki, M.L., Humphreys, T.E., Stauffer, B.: Attackers
can spoof navigation signals without our knowledge.
Here’s how to fight back GPS lies. IEEE Spectrum. 53, 8,
2653 (2016).
https://doi.org/10.1109/MSPEC.2016.7524168.
25. Santamarta, R.: White paper: Last Call for SATCOM
Security, https://ioactive.com/wp-
content/uploads/2018/08/us-18-Santamarta-Last-Call-
For-Satcom-Security-wp.pdf, last accessed 2021/04/19.
26. Stripydog: NMEA-0183 over- IP: The unwritten rules for
programmers,
https://stripydog.blogspot.com/2015/03/nmea-0183-over-
ip-unwritten-rules-for.html.
27. Svilicic, B., Kristić, M., Žuškin, S., Brčić, D.: Paperless
ship navigation: cyber security weaknesses. Journal of
Transportation Security. 13, 3, 203214 (2020).
https://doi.org/10.1007/s12198-020-00222-2.
28. Svilicic, B., Rudan, I., Frančić, V., Mohović, D.: Towards
a Cyber Secure Shipboard Radar. Journal of Navigation.
73, 3, 547558 (2020).
https://doi.org/10.1017/S0373463319000808.
29. Svilicic, B., Rudan, I., Jugović, A., Zec, D.: A Study on
Cyber Security Threats in a Shipboard Integrated
Navigational System. Journal of Marine Science and
Engineering. 7, 10, (2019).
https://doi.org/10.3390/jmse7100364.
30. Tam, K., Jones, K.: MaCRA: a model-based framework
for maritime cyber-risk assessment. WMU Journal of
Maritime Affairs. 18, 1, 129163 (2019).
https://doi.org/10.1007/s13437-019-00162-2.