Model-based Threat and Risk Assessment for Systems Design
Avi Shaked
1,2 a
and Yoram Reich
2b
1
Cyber Division, Israel Aerospace Industries, Ashdod, Israel
2
Systems Engineering Research Initiative, Faculty of Engineering, Tel Aviv University, Tel Aviv, Israel
Keywords: Threat and Risk Assessment, Model based Engineering, Cybersecurity, Security by Design, Systems Design.
Abstract: Integrating cybersecurity considerations in the design of modern systems is a significant challenge. As
systems increasingly rely on connectivity and software to perform, cybersecurity issues of confidentiality,
integrity and availability emerge. Addressing these issues during the design of a system – a security by-design
approach – is desirable, and considered preferable to patching an existing design with extraneous components
and mechanisms. In this paper, we present a model-based methodology for cybersecurity related systems
design. This field-proven methodology takes into consideration cybersecurity threats alongside the system’s
composition and existing mechanisms, in order to communicate, assess and drive the incorporation of security
controls into the system design. We discuss aspects of the methodology’s design and how it relates to its real-
life applications and usage context.
1 INTRODUCTION
Contemporary systems rely on software components
and on connectivity to perform. Software components
exist in diverse forms, such as Personal Computer
based software, embedded software (e.g., software
used in microcontrollers) or programmable logic
(e.g., configuration of Field Programmable Gate
Arrays components). Connectivity also manifests in
various forms: a system can include internal
connectivity between its components and/or can
interact with external entities and systems (e.g., a
connection to the Internet). Connectivity may be
fixed and continuous, as in internal communication
busses used throughout the system’s operation; or
temporal, as in a connection between a system
component and a dedicated test equipment during
acceptance tests or software version upgrades.
While software and connectivity are used to
promote system capabilities (e.g., rapid deployment
of new features and operation in swarms,
respectively), they also open up the system to cyber-
threats. These threats may impact the confidentiality,
integrity and/or availability of the system, either in its
entirety or with respect to its components. It is
therefore of utmost importance to identify such
a
https://orcid.org/0000-0001-7976-1942
b
https://orcid.org/0000-0002-0922-8381
potential threats and manage them as risks. A popular
term used to denote such activities is Threat and Risk
Assessment (TRA).
When designing a system, TRA may be used not
only to assess the cybersecurity risk posture of the
system under development, but also to affect its
design to take into account the cybersecurity
perspective, and consequently mitigate potential
threats. Such an approach is commonly referred to as
“security by-design.”
In this paper we present a methodology for
incorporating cybersecurity considerations in the
design of systems based on TRA. In Section 2, we
provide the relevant background related to security
by-design methodology. In Section 3, we discuss our
design research method. In Section 4, we present
TRADES: a Threat and Risk Assessment for the
Design of Engineered Systems methodology. In
Section 5, we share information concerning the
multiple case study validation of our methodology,
based on its applications in Israel Aerospace
Industries (IAI). Section 6 discusses related
approaches, and Section 7 summarizes key points and
discusses the potential of further research efforts.
Shaked, A. and Reich, Y.
Model-based Threat and Risk Assessment for Systems Design.
DOI: 10.5220/0010187203310338
In Proceedings of the 7th International Conference on Information Systems Security and Privacy (ICISSP 2021), pages 331-338
ISBN: 978-989-758-491-6
Copyright
c
2021 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
331
2 BACKGROUND
2.1 Contextual System Development
Characteristics
We provide relevant context with respect to systems
development in IAI, which affected the design of our
methodology.
In IAI – as in many other high-tech industries and
companies that develop complex, engineered systems
systems development is typically done within the
scope of a development project. The development
scheme favours a top-down approach, according to
which system-level requirements and design are
established to meet stakeholder requirements (and
specifically, the purchaser of the system). These then
drive the development of lower-level components.
Another related characteristic of systems
development in IAI is its hierarchical approach,
which views the system as a hierarchical organization
of components. This is aligned with the view offered
by Vee-model, a prominent system development
model (Forsberg and Mooz, 1991). For example, the
highest hierarchy of a system comprises several sub-
systems; each of these sub-systems includes several
lower-level components that may be further
decomposed into even lower-level components, and
so forth. The hierarchical organization provides
engineering and managerial context and
responsibilities (e.g., reflecting in a Work-
Breakdown Structure associated with the
development project).
A significant manifestation of the hierarchical
approach is that – throughout systems development –
the systems-engineering team responsible for the
development of a component in a specific hierarchy
defines requirements for the lower hierarchy’s
components, according to which the lower-level
components are designed and eventually verified.
These requirements are expected to relate to multiple
aspects of functionality, performance and quality,
with cybersecurity aspects included.
In general, the cost of performing changes late in
the system development life-cycle or after a system is
fielded is increasingly high (compared to performing
these changes earlier or avoiding them by proper
design in advance), and changes that originate from a
cybersecurity perspective are no exception. It is,
therefore, important to communicate the need for
cybersecurity mechanisms as early as possible in the
development effort (Mead and Stehney, 2005;
Shevchenko et al., 2018).
Also, it is noteworthy that IAI development
projects typically last for several years, and that the
resulting systems remain operational for a long period
of time (a 10-20 year operational period is not
uncommon). During such extended periods of
development and operations, the organization (and
often, its customers) considers it important to have an
up-to-date documentation of the system’s design.
Also, throughout such lengthy life-cycles, the
cybersecurity threat landscape is subject to change
(due to newly emerging attack technologies and
techniques, for example), and this introduces further
challenges to the threat and risk assessment of the
systems.
2.2 Threat and Risk Assessment
Typically, the threat and risk assessment of systems
is expressed as natural language text, tables and free-
form diagrams (that typically provide a specific view
of a given design). A threat model related publication
(Bodeau and McCollum, 2018), for example, depicts
a representative example as a story with free-form
diagrams. Another representative example is a radar
system security research report (Cohen et al., 2019),
in which threats are expressed in the form of a table.
The aforementioned documentation is typically
prepared ad-hoc and is not necessarily aligned with
the actual system design. This does not support
rigorous engineering nor the establishing of the
cybersecurity posture throughout the system life-
cycle. The free-form approach to TRA indicates a gap
in sound, practical methodology.
Some conceptual frameworks relating to TRA and
to security by-design exist, providing important
concepts and guidance. Two prominent examples
follow. First, MITRE a not-for-profit organization
which specializes in systems engineering and
cybersecurity offers a threat modelling framework
(Bodeau and McCollum, 2018), which identifies the
system composition and data flows as critical in
evaluating the potential impact of a cyber-attack.
Second, the United States National Institute of
Standards and Technology (NIST) offers an
authoritative source – NIST SP 800-160 – which
discusses systems security-by-design, and provides
exhaustive natural language guidelines for
introducing security considerations in systems
engineering activities (Ross et al., 2016). Both
aforementioned publications lack a concrete
approach for applying TRA for systems design. They
do, however, stress the importance of introducing
requirements for the system and its constituents
(alternately, introducing capabilities/functions to the
system and its constituents) from a security
perspective. Introducing quality related system
ICISSP 2021 - 7th International Conference on Information Systems Security and Privacy
332
requirements has also been acknowledged as an
effective mechanism to incorporate security concepts
in early stages of development (Mead and Stehney,
2005). However, we are unaware of any rigorous,
field-proven model-based approach to do so. We
discuss some related approaches in Section 6.
3 METHOD
TRADES is a domain-specific methodology for
security by-design. TRADES was developed based
on concepts and insights from another research
project in which we developed another domain-
specific methodology (Shaked and Reich, 2019).
First, TRADES was designated as a model-based
engineering methodology. We consider this as
TRADES’ significant value proposition. Our main
goal was to promote a rigorous, lasting depiction of a
systems cybersecurity design, and addressing this
based on a well-defined, unambiguous data model
(Ramos et al., 2011) seemed essential; especially
when considering the aforementioned lengthy
development and operation life-cycle and the
changing threat landscape. A model-based approach
facilitates the ability to use the model as a single
source of reference, establishing communication and
coordination between cybersecurity experts, system
engineers, managers and regulators. Specifically,
modelling languages were suggested as a prospective
approach to promote the definition and
communication of security related abstractions
(Mailloux et al., 2018). Digital models can also be
maintained throughout the development, reflecting
and communicating any changes to both the system
and its cybersecurity assumptions (e.g., applicable
threats, included security mechanisms and risk
management decisions).
Second, as in our previous research and in
accordance with others (for example, Pullonen et al.
(2019)), our domain-specific language which is part
of the model-based methodology was designed to
be as minimal as possible, yet expressive, to support
its cognitive usability. We initially identified the key
ontological entities (threat, security control, and
component; as described in the next section) and their
relationships, and designed our model and
representations accordingly. The design of our model
and representations included cognitive considerations
that were established as effective in our previous
research and were reapplied here with domain-related
3
The open source TRADES modelling tool is available at
https://github.com/iai-cyber/TRADES.
adaptations. Furthermore, our domain-specific
methodology was designed to fit the contextual
aspects of systems development (as mentioned in
Section 2.1).
We implemented our domain-specific
methodology using software modelling technologies.
While the details regarding the implementation of
TRADES on top of a modelling infrastructure is
beyond the scope of this paper, we provide an open-
source tool to support the dissemination and use of
the methodology
3
.
The methodology was applied to several real-life
cases, and this further establishes its validity, in
accordance with the multiple case study research
methodology. A brief overview of these applications
is provided in Section 5, with lessons learned
discussed both in Section 4 (where we introduce the
design of the methodology) and in the concluding
discussion.
4 TRADES METHODOLOGY
In this section, we present the TRADES
methodology. Prominent design considerations (of
the methodology) are explained throughout.
4.1 System Design: Ontology and Its
Representation
In TRADES, systems and their constituents are
represented by a component typed element. A
component is an abstract concept, which may
represent logical and/or physical/structural entities.
Considering the context in which we wished to apply
TRADES specifically the hierarchical approach to
systems development system composition is
portrayed in TRADES using the aggregation of
components in hierarchies. In addition to being
aligned with the overall systems development
methodology of IAI, the use of hierarchies
contributes to designing security in various level of
abstraction, while maintaining traceability between
low-level and high-level concepts (Mead and
Stehney, 2005). It is noteworthy that hierarchies are
components themselves. A “box”-like notation is
used to represent a component. A two-dimensional
shape has the ability to be used as a container to
reflect a design hierarchy, and in TRADES, a
component can indeed serve as a container to include
other components.
Model-based Threat and Risk Assessment for Systems Design
333
Another aspect of system design that TRADES
emphasizes is the exchange of data between
components. This exchange is denoted using a
directional arrow between components (representing
an “affect relation” model element), which conveys a
data” typed element. The use of a directional shape
is designed to force the cybersecurity analyst to
“think” in terms of the data, its source and its target,
as opposed to thinking in terms of links between
entities.
TRADES includes another system design
element: security control.” While TRA related
aspects of the security control ontological entity (and
its directly corresponding model element) are
explained in the subsequent section, we note that this
element represents a security control (also commonly
referred to as “mitigation mechanism”) that is
found/designed in the system or in its constituents.
This designation of the security control as a system
design element is reflected in the design of the
TRADES notation (i.e., its representation): a security
control is placed within the boundaries of a
component, denoting that the implementation of the
security control is the responsibility of the component
owner. A security control element is currently
represented as a box with a dashed outline,
differentiating it from a component element (whose
outline is solid).
Figure 1 provides an example of the system design
notation of TRADES. A system SystemA
comprises two subsystems: SubsysA1 and SubsysA2.
The model includes lower level hierarchy details for
SubsysA1, with two components (Component1 and
Component2) exchanging two data items (Data1 from
Component1 to Component2 and Data2 in the
opposite direction). Two security controls are
identified in the model: Control1 – is associated with
the top hierarchy (it is therefore located directly under
the SystemA component); and Control2 is associated
with a component in the second level of hierarchy
(SubsysA1).
Figure 1: TRADES notation for capturing system design.
4
For example, we successfully imported all of MITRE’s
CAPEC attack patterns (https://capec.mitre.org) as threats
to our models, resulting in over 500 threats available for a
cybersecurity analyst to consider when performing TRA.
4.2 TRA: Ontology and Its
Representation
In TRADES, the TRA revolves around the threats that
are identified as applicable to the system (and to its
constituents). Corresponding with the threat
ontological entity, we use threat” model elements to
denote threats. Since threats are external to the system
design, they appear in the TRADES analysis
diagram outside of the system boundaries (that are
denoted by the outline of the box-like shape of the
component element). In order to further facilitate
cognitive differentiation between design elements
and threat elements, the notation used for threat
elements takes the form of an ellipse.
Threats may appear regardless of their
applicability to a specific design. This promotes the
ability to import a repository of threats whose
applicability to the system under assessment needs to
be determined
4
.
Once a threat element appears in the model, it may
be allocated to one or more components, marking that
it may be applicable to the specific components. This
allocation is graphically denoted in the TRADES
diagram using an arrow from the threat to the
component. A non-trivial aspect of this allocation is
that it is recognized to be a key model element for
TRA: threat allocation. This was identified only
after the two first applications of TRADES in IAI.
After a threat has been allocated to a component, the
allocation is the entity according to which the risk
assessment of the system is performed. Specifically,
potential impact and difficulty (with high difficulty
denoting low feasibility) are assessed with respect to
the specific threat allocation. The TRADES model
includes elements relating to both impact and
difficulty, and these elements may be associated with
a threat allocation, characterizing its assessment with
respect to the two risk management traits (“impact
and difficulty elements exist in the TRADES
model, corresponding with the respective TRA
scoring system ontological attributes, but are not
represented in the TRADES diagram design).
Another task in assessing the cybersecurity risks
relates to the security controls that are included in the
system design. For this, security controls are linked
with the threat allocation which they are believed to
mitigate, using a directional arrow from control to the
threat allocation arrow, designating a threat
mitigation” element.
ICISSP 2021 - 7th International Conference on Information Systems Security and Privacy
334
Figure 2 demonstrates the full design notation of
TRADES. The system (of Figure 1) is shown
alongside several threat elements. Threat1 was
allocated to SubsysA1, Threat2 was allocated to
SystemA, Threat3 remains unallocated, Threat4 is
allocated to Component2. Control2 is associated as a
mitigation for the Threat1 to SubsysA1 allocation.
Control1 was not associated with any threat allocation
(which may signify, for example, that it is either
redundant or irrelevant to the cybersecurity
perspective
5
). The figure exemplifies an additional
representational aspect of the TRADES
methodology: a colouring notation is used to reflect
attributes of some elements. Specifically, the Threat1
to SubsysA1 threat allocation is marked in green,
indicating it has been estimated as an acceptable risk
(possibly based on the mitigation by Control2); while
the Threat2 to SystemA threat allocation is marked in
red, indicating it has been estimated as a gap (an
unacceptable risk). The Threat4 to Component2
threat allocation appears in black, as remaining
undecided (this is based on the “Assessment”
property value, shown in the “Properties” section
below the diagram). Some other properties of the
threat allocation are also shown. The “Difficulty
score” and “Impact score” properties are of particular
interest, as they are associated with the respective
elements (associating the threat allocation with the
“2” difficulty score model element and the “high (3)
impact score model element).
Figure 2: TRADES full notation.
Table 1 summarizes the identified ontological
concepts for the cybersecurity related systems design
domain and how they are addressed in the TRADES
model and in the TRADES diagram. This table
exemplifies the one-to-one correspondence between
5
Controls may be used for purposes other than for
mitigating threats. For example, a system may include
controls based on its availability objectives or its safety
objectives.
ontology and model elements as well as between
model elements and their representation, based on the
ontological clarity (Wand and Weber, 1993) and the
semiotic clarity (Moody and van Hillegersberg, 2008)
design principles. The table also identifies the leading
design perspective Systems Engineering (SE) or
Cybersecurity (Cy) – that is associated with each
concept, stressing that cybersecurity domain experts
(or – according to NIST’s terminology Systems
Security Engineering) need to collaborate with
systems engineers in order to design a system with
cybersecurity considerations.
Table 1: TRADES methodology with respect to domain
ontology.
Ontological
concept
Leading
design
p
ers
p
ective
Model
element
Representation notation in
TRADES diagram
System
components
(logical/
p
h
y
sical
)
SE Component
Hierarchical
composition
SE
Component
as container
Data flow SE
affect
relation
Data SE Data
(
data conve
y
ed b
y
data flow
)
Security
control
Cy Control
Security
control’s
allocation to
component
Cy
composition
relation
(component
comprises
control
)
Threat Cy Threat
Association
of threat with
com
p
onent
Cy
Threat
allocation
relation
Security
control’s
threat
miti
g
ation
Cy
Threat
mitigation
relation
Impact Cy & SE Impact score
Out of diagram scope
(represented in other model
views / as attributes
Feasibility Cy
Difficulty
score
Out of diagram scope
(represented in other model
views / as attributes
Model-based Threat and Risk Assessment for Systems Design
335
In terms of systems development yields, the
primary output of applying TRADES to a system
design is the identification of security controls that
need to be included in each component, in the
respective level of hierarchy. Once a security control
is allocated to a component, it is established as a
function of the component. The security related
function is subsequently defined in terms of
requirements; and from this point on, they enter the
development cycle just like any other requirement.
This approach is well-aligned with the approach
encouraged by NIST (Ross et al., 2016).
5 APPLICATION REPORT
Since its development, TRADES has been used to
analyse and support the design of more than 10
different systems, including remotely piloted aerial
vehicle systems, radar systems, airborne systems and
missile systems.
TRADES was applied to systems in various stages
of the system development life-cycle. Some
applications were designated to provide an existing
system’s security posture in order to support the
planning of future development blocks. Other
applications were made during the course of
development, for supporting the system design as it
was devised (by system engineers, hardware
engineers and software engineers). Also, several
applications were made during the project proposal
phase, in order to communicate cybersecurity aspects
with stakeholders, specifically with the acquisition
authority (customer) and regulators.
The resulting system models ranged from a model
of 10 system components in three hierarchies, 5 data
items, 5 threats and 4 controls to a model of about 900
system components in four hierarchies, 150 data
items, 30 threats and no controls
6
.
In all of the applications, TRADES was evaluated
as promoting communication and coordination
between the parties involved.
One difficulty we experienced in some of the
applications relates to the lack of a dedicated element
for the communication link itself, as opposed to the
data it conveys. We found that system-engineers (in
6
Specifically, in the largest model yet, the identification of
controls was deemed out of scope, and was left to a
potential future work. The maximum number of control
elements in a specific TRADES application so far is about
50.
7
These should not be confused with model-based
engineering methods. Model-based engineering requires
the establishment of an explicit underlying model
practice) often prefer to present the system
architecture with physical links and protocols as
opposed to data exchanges; and that some
cybersecurity experts like to identify risks primarily
based on these communication links.
6 RELATED WORK
TRADES’ leading motivation is incorporating
security related perspective into systems design. As
such, TRADES shares a common underlying
approach with NIST SP800-160 (Ross et al., 2016)
and the SQUARE (Security Quality Requirements
Engineering) methodology (Mead, 2007). The NIST
publication provides considerations, and does not
provide a concrete methodology as TRADES.
SQUARE does provide high level process
definitions, but – unlike TRADES – it is not a model-
based approach, and it does not provide any model
which corresponds with the domain ontology. As an
example, one of the steps in SQUARE is “Categorize
requirements,” with initial requirements and
architecture identified as inputs to the association of
the requirements with the relevant hierarchy and
component (Mead and Stehney, 2005). TRADES
offers a more concrete take on this, as it specifies and
captures the aspects of the design as well as of the
TRA that are needed in order to assign requirements
to the appropriate hierarchy and component. With
respect to this, we also note that SQUARE’s
identification of inputs for this step (of its prescribed
process) misses an important input: the identification
of a threat allocation, and particularly the
identification of the hierarchy/component to which a
threat is allocated. This allocation has implications to
the association of requirements for security controls
with a specific hierarchy/component.
Threat modelling approaches that rely on
diagrammatic representation exist
7
, with data flow
diagrams (DFDs) being a popular diagrammatic form,
as discussed in a review of available threat modelling
methods (Shevchenko et al., 2018). The review
specifically identifies that using DFDs is insufficient
for threat modelling; further emphasizing that the
common, DFD-based approaches fail to derive
commonly known as “meta-model” and its realization
in the form of a database of modelling elements; with
diagrams providing a viewpoint into the database
elements and/or supporting updates to the database (e.g.,
adding or updating elements). Purely diagrammatic
methods – such as those presented shortly – often lack the
formal ontological foundations and the establishment of a
rigorous data model.
ICISSP 2021 - 7th International Conference on Information Systems Security and Privacy
336
pertinent domain ontology, which is required in order
to perform a systematic TRA. Specifically, upon
examination of a known DFD-based tool (Microsoft’s
Threat Modelling Tool), we found that DFD has not
been adapted – as a formal representation to include
threat modelling elements; and that its system design
perspective in threat modelling remains information-
flow oriented. Correspondingly, the DFD
representation misses the representation of security
controls as well as the hierarchical organization of
systems, and it was deemed inappropriate for system
level TRA scenarios in IAI (and companies
employing similar development paradigms).
MITRE’s conceptual threat modelling framework
(Bodeau and McCollum, 2018) illustrates cyber threat
modelling key constructs as a meta-model. Its focus
is on threat-related elements, without providing any
system-related elements. MITRE’s Common Attack
Pattern Enumeration and Classification (CAPEC)
scheme exhibits similar focus on the threat element,
and lacks a systematic approach (Yuan et al., 2014).
In comparison, TRADES sponsors a more holistic
view of conducting TRA, taking into account
contextual aspects, and specifically those related to
the system design (that is under assessment) as well
as to the management of threat allocations with
respect to the relevant design hierarchy.
TRADES supports top-down systems
development approach that facilitates security by-
design, introducing a dedicated cybersecurity
perspective that can be used to design security
features throughout systems development. In
comparison, TLDR (Mahler et al., 2020) and PE-
BPMN (Pullonen et al., 2019) were deemed
inappropriate as security by-design approaches for
such context. TLDR (Threat identification, ontology-
based Likelihood, severity Decomposition, and Risk
integration) is a threat and risk assessment
methodology which focuses on the analysis of an
existing design, and does not offer a security design
perspective. PE-BPMN (Privacy Enhanced Business
Process Model and Notation) is business process
oriented, and is primarily aimed for privacy-related
analysis of a given process design; as opposed to
TRADES which is systems oriented, and aims to feed
a wider scope of cybersecurity aspects into the system
requirements and design. PE-BPMN primarily
focuses on confidentiality aspects (and, to some
extent, also includes data integrity aspects), whereas
TRADES is more inclusive, and can be used to
introduce aspects of confidentiality, of integrity –
8
See, for example, footnote #4 for our preliminary take on
this.
with respect to both performance and data and of
availability.
7 DISCUSSION
Designing systems with cybersecurity considerations
is essential, as systems increasingly depend on
software and connectivity to perform. We presented
an applicable methodology TRADES for
supporting the incorporation of a cybersecurity
perspective to systems design.
Specifically, TRADES supports the elicitation of
security related requirements. This is based on
identifying desirable security controls in a
contextualized form (of pertinent threat allocations
and risks); and associating these controls as
functional capabilities in different hierarchical levels
of the system design.
Based on multiple real-life applications of
TRADES, its contribution to the communication and
coordination of the cybersecurity perspective and of
pertinent system design between those involved in
development efforts was established. The TRADES-
based models are maintainable throughout the
development life-cycle.
We attribute the aforementioned qualities of
TRADES to its underlying domain-specific, model-
based approach. Specifically, the TRADES model
and notation were derived from profound ontological
understanding of the domain; and the resulting
models are consistent and unambiguous (as opposed
to commonly used free-form diagrams and text, for
example).
While the current TRADES model is aligned with
our domain’s ontology, we are still considering how
to address the difficulty associated with some system
engineers and cybersecurity experts’ desire to include
a dedicated element to denote a link and its
characteristics. We are also evaluating the
effectiveness of additional colouring notations. While
we use colouring notations rigorously, based on our
TRADES data model, we have yet to crystalize a
definitive colouring notation (i.e., one that is to be
used as a standard for all applications). We hope that
our further research will be fruitful with respect to
these.
Future research may integrate TRADES with
other modelling domains (e.g., systems engineering
model-based engineering methods or frameworks), as
well as with cybersecurity related threat repositories
8
.
Model-based Threat and Risk Assessment for Systems Design
337
Integration potential also exists in the integration
between TRADES and PE-BPMN, synthesizing the
two approaches to facilitate a security by-design
approach to system development which also takes
system-related processes (such as operational
processes or maintenance processes) into account.
REFERENCES
Bodeau, D. J., & McCollum, C. D., 2018. System-of-
systems threat model. The Homeland Security Systems
Engineering and Development Institute (HSSEDI)
MITRE: Bedford, MA, USA.
Cohen, S., Gluck, T., Elovici, Y., & Shabtai, A., 2019.
Security Analysis of Radar Systems. In Proceedings of
the ACM Workshop on Cyber-Physical Systems
Security & Privacy (pp. 3-14).
Forsberg, K. and Mooz, H., 1991. The relationship of
system engineering to the project cycle. In INCOSE
International Symposium (Vol. 1, No. 1, pp. 57-65).
Mahler, T., Elovici, Y., & Shahar, Y., 2020. A New
Methodology for Information Security Risk
Assessment for Medical Devices and Its Evaluation.
arXiv preprint arXiv:2002.06938.
Mailloux, L.O., Beach, P.M. and Span, M.T., 2018.
Examination of security design principles from NIST
SP 800-160. In 2018 Annual IEEE International
Systems Conference (SysCon) (pp. 1-8). IEEE.
Mead, N. R., & Stehney, T., 2005. Security quality
requirements engineering (SQUARE) methodology.
ACM SIGSOFT Software Engineering Notes, 30(4), 1-
7.
Mead, N.R., 2007. How to compare the Security Quality
Requirements Engineering (SQUARE) method with
other methods (No. CMU/SEI-2007-TN-021).
CARNEGIE-MELLON UNIV PITTSBURGH PA
SOFTWARE ENGINEERING INST.
Moody, D. and van Hillegersberg, J., 2008. Evaluating the
visual syntax of UML: An analysis of the cognitive
effectiveness of the UML family of diagrams. In
International Conference on Software Language
Engineering (pp. 16-34). Springer, Berlin, Heidelberg.
Pullonen, P., Tom, J., Matulevičius, R. and Toots, A., 2019.
Privacy-enhanced BPMN: Enabling data privacy
analysis in business processes models. Software and
Systems Modeling, 18(6), pp.3235-3264.
Ramos, A.L., Ferreira, J.V. and Barceló, J., 2011. Model-
based systems engineering: An emerging approach for
modern systems. IEEE Transactions on Systems, Man,
and Cybernetics, Part C (Applications and Reviews),
42(1), pp.101-111.
Ross, R., McEvilley, M. and Oren, J., 2016. NIST Special
Publication 800-160: Systems Security
Engineering. National Institute of Standards and
Technology.
Shaked, A., and Reich, Y., 2019. Designing development
processes related to system of systems using a modeling
framework. Systems Engineering, 22(6), 561-575.
Shevchenko, N., Chick, T.A., O'Riordan, P., Scanlon, T.P.
and Woody, C., 2018. Threat modeling: a summary of
available methods. Carnegie Mellon University
Software Engineering Institute Pittsburgh United
States.
Wand, Y. and Weber, R., 1993. On the ontological
expressiveness of information systems analysis and
design grammars. Information Systems Journal, 3(4),
pp.217-237.
Yuan, X., Nuakoh, E.B., Beal, J.S. and Yu, H., 2014.
Retrieving relevant CAPEC attack patterns for secure
software development. In Proceedings of the 9th
Annual Cyber and Information Security Research
Conference (pp. 33-36).
ICISSP 2021 - 7th International Conference on Information Systems Security and Privacy
338