MODEL - AND SIMULATION DRIVEN SYSTEMS -
ENGINEERING FOR CRITICAL INFRASTRUCTURE SECURITY
ARCHITECTURES
Sascha Goldner
and Philipp Rech
System Design Center, EADS Deutschland GmbH, Landshuterstraße 26, Unterschleißheim, Germany
Keywords: Critical Infrastructure Protection, Scenario based, Rule based, Process Modelling, Case based Reasoning,
Expert System, Systems Engineering, NATO Architecture Framework, CONOPS, Simulation.
Abstract: The design of systems that support the protection of critical infrastructures or state borders against various
threats is a complex challenge. This requests the use of both modelling and simulation in order to reach the
operational goal. Threat scenarios like terrorist attacks, (organized-) crimes or natural disasters involve
many different organisations, authorities, and technologies. These systems are often operated only by
implicit known processes and diverse operational guidelines based on dissimilar legislations and
overlapping responsibilities. In order to cope with these complex infrastructure systems and their
interconnected processes, a scenario- and architecture based systems engineering approach must be
implemented to design a solution architecture compliant with the requirements, internal and external
demands. This paper gives an overview of the developed approach towards the system architecture and
explains the different engineering steps in order to implement this architecture for real use-cases.
1 INTRODUCTION
Maintaining and improving the security of critical
infrastructures against various known and new
threats is a crucial task and often just tackled by the
implementation of new technology without assessing
the effectiveness of the security measures as a
whole. Besides security technologies (e.g. CBRNE
scanners, CCTV, etc.), nowadays rules and
regulations also have to be considered in the
implementation of security systems and all related
processes in critical infrastructures. Moreover, the
many different involved authorities (i.e. airport
operators, local and federal police, different fire
departments, military etc.), work by experience and
implicit knowledge with regards to their
organisation-specific guidelines for decision making.
Due to the great importance of these non-technical
factors of influence, there is a need for a security
system engineering approach that takes into account
these non-technical parameters and their
harmonization among the different organizations
participating.
Within this paper we present the EADS
engineering approach based on the usage of an
expert system, an architecture framework as well as
a simulation tool and describe in detail the different
engineering steps. Currently this approach is
successfully implemented within studies and
international large-scale projects in the domain of
border security and critical infrastructure security.
2 OBJECTIVE
Our objective is the development of system
architectures for complex security systems. This
requires the detailed modelling of the respective
security mechanisms and relevant processes in
which the many different authorities are involved.
Their expert knowledge, the real-world behaviour
and all required decisions have to be gathered,
analyzed and represented in a consistent way. By
application of our method the following goals will
be met:
Gathering tacit expert knowledge and
transforming it into system / software behaviour
Improving of the functionality of security
systems against new emerging threats by using
424
Goldner S. and Rech P. (2010).
MODEL - AND SIMULATION DRIVEN SYSTEMS - ENGINEERING FOR CRITICAL INFRASTRUCTURE SECURITY ARCHITECTURES.
In Proceedings of the 12th International Conference on Enterprise Information Systems - Artificial Intelligence and Decision Support Systems, pages
424-427
DOI: 10.5220/0002972604240427
Copyright
c
SciTePress
new technologies (e.g. 'naked' scanner) and
processes.
Using of domain expert knowledge in order to
verify and validate the system architecture
3 APPROACH
In order to reach the above described objectives we
developed a generic engineering approach (Fig. 1)
which is implemented in six subsequent steps using
an EADS-tailored tool chain. The following
paragraphs will explain the steps in detail.
Step 1: Analysis of As-is System. A fundamental
precondition of all engineering approaches is a deep
knowledge about the system or object of
investigation. For this reason the first step of the
engineering approach is a system analysis in order to
understand the system itself and its functioning. In a
data gathering phase of this analysis all relevant
elements and parameters are identified which
describe and influence the object of investigation
(e.g. airport, border control, etc.). Further more, it is
analysed how these parameters correlate and
influence each other. The direction of a correlation is
defined and also a qualitative assessment is applied
which allows to describe the degree of correlation of
the parameters (no, medium, strong correlation). The
influence analysis can be done by using either
correlation tables or matrices.
Another aspect of the system analysis
concentrates on the business processes. They
describe the activities and their sequence of
execution in order to produce a specific service or
product on a very fine-grained level. For every
process step it has to be determined e.g. who
performs this step, who is responsible, what are the
input and output parameters, how long does this step
take and what systems and tools are used to
implement this process activity.
The step of analysing the target system is crucial
and the analysis results define the basis for creating
threat scenarios which is done in step 2.
Step 2: Scenario Generation. In our work we
pursue a scenario-driven approach because we
believe that only a real-life scenario can show the
actual benefit and practical relevance of the method.
Further we need a real-life reference in order to
demonstrate how the knowledge of domain experts
is incorporated in our work. Finally the added value
of our method can be measured and compared more
easily when it is based on a real-life example and it
is more comprehensible to any stakeholder.
A scenario is used to describe a potential threat
respectively a potential attack that can be carried out
against the security system. The scenario description
is made up of the system elements that have been
specified in the previous step. The important point is
the ability to create self-consistent scenarios. This is
because of the correlations between the system
elements that have been specified in step 1. For
example, it makes no sense to use a knife in order to
attack an armoured vehicle. In that case, we would
have defined a negative correlation value between
knife and armoured vehicle.
Another very important aspect of the scenario-
driven approach is that the business processes which
are affected by the scenario can be derived
automatically. Regarding business processes, we act
on the assumption that there are just a few roles that
can be taken on in the system, e.g. the system
"airport" offers for example the roles of a passenger,
an employee, a visitor and a supplier. Further we
assume that an attacker tries to remain undetected
until the actual attack on the target starts by slipping
into at least one of the roles the system provides.
Just in the moment the attacker performs the attack
or the attacker's cover has been compromised, the
attacker's original role becomes obvious.
Business processes describe the intended
behaviour of a system and our goal is to create a
system architecture which on the one hand provides
a functionality that complies with the real-life
behaviour and on the other hand allows to assess and
optimize the systems security mechanisms. Our
main conclusion in this regard is that the system
architecture has to have a high degree of structural
similarity to the systems business processes. The
next section describes how such a system
architecture can be achieved. The result of this step
are self-consistent scenarios.
Step3: Design of the Solution Architecture. After
the as-is system is analysed and threat scenarios are
defined, the goal of the third step in the system
engineering approach is the design of the desired to-
be system architecture. In this architecture we
consider two different views (
Figure 1):
Operational View
System View
The operational view describes how the roles,
organizations and activities have to operate in order
to react on the given threat. This description provide
the operational guidelines which are noted in a
CONOPS (concept of operations). The system view
specifies the technical framework for the
implementation of the CONOPS. This system view
MODEL - AND SIMULATION DRIVEN SYSTEMS - ENGINEERING FOR CRITICAL INFRASTRUCTURE
SECURITY ARCHITECTURES
425
consists of different components such as platforms
(e.g. aircraft-carrier), hardware (camera, radar),
software, facilities and material.
Both the operational and the system view of the
architecture have to be designed in a way that they
enable an organisation to cope with the scenarios
defined in step 2.
Where
Where
How
How
What
What
Generated Objects
Generated Objects
Core Objects
Core Objects
Container
Asset
Function
Product
Who
Who
Operational
Architecture View
Entities
Relationships
Attributes
Relationships
Attributes
Entities
Org
Org
Network
Network
Knowledge
Skills & Abilities
Knowledge
Skills & Abilities
Performance
Performance
Process
& Sub-Activities
Process
& Sub-Activities
Info
Info
CONOPS
(rules)
CONOPS
(rules)
Role
Role
Data
Data
Design
Rules
Design
Rules
System
System
Transfer
Link
Characteristic
Structure
Why
Strategy
Cycles
Cycles
Cycles
Cycles
When
Behavior
Need
Line
Need
Line
Info
Exchange
Info
Exchange
Interface
Interface
Data
Exchange
Data
Exchange
System
Function
System
Function
Op
Node
Op
Node
System
Node
System
Node
Performer Asset
System
Architecture View
Where
Where
How
How
What
What
Generated Objects
Generated Objects
Core Objects
Core Objects
Generated Objects
Generated Objects
Core Objects
Core Objects
Container
Asset
Function
Product
Who
Who
Operational
Architecture View
Entities
Relationships
Attributes
Relationships
Attributes
Entities
Org
Org
Network
Network
Knowledge
Skills & Abilities
Knowledge
Skills & Abilities
Performance
Performance
Process
& Sub-Activities
Process
& Sub-Activities
Info
Info
CONOPS
(rules)
CONOPS
(rules)
Role
Role
Data
Data
Design
Rules
Design
Rules
System
System
Transfer
Link
Characteristic
Structure
Why
Strategy
Cycles
Cycles
Cycles
Cycles
When
Behavior
Need
Line
Need
Line
Info
Exchange
Info
Exchange
Interface
Interface
Data
Exchange
Data
Exchange
System
Function
System
Function
Op
Node
Op
Node
System
Node
System
Node
Performer Asset
System
Architecture View
Figure 1: Elements within the Operational- and System
Architecture (Source: NATO Architecture Framework v3).
First the operational view has to be defined in order
to support the realization of the desired CONOPS.
Therefore the required capabilities of an
organisation have to be gathered. These are for
example surveillance and reconnaissance in the
military or evacuation and crisis management in the
civil environment. In the system engineering
methodology these capabilities will be modelled
using a capability taxonomy.
In a next step the operational nodes have to be
defined which are needed to realize the required
capabilities. Operational nodes are processes or
rather a set of successive activities with information
flows between them in order to exchange data. This
analysis provides a clear picture how operations are
performed and thereby support the system engineer
to specify the required technical equipment to carry
out the processes.
The result of step 3 is (amongst other
architecture elements) a process model that shows
all units of behaviour (actors, systems, processes)
and their condition-based interactions.
The more challenging task is the acquisition of
the detailed behaviour for the decision nodes in that
process model. These are required in order to
perform a realistic simulation and optimization of
the process. To cope with this challenge EADS has
build an interface between the architecture
modelling tool and a case-based expert system
which is used to enter the business logic / behaviour
for the decision nodes of the process model. This
procedure is described in more detail in the
following section.
Step 4: Verification and Validation of the System
Architecture. After the process model has been
created, the correctness of the architecture has to be
verified in order to provide the functionality the
domain experts expect. The step of verification and
validation is a quite straight forward task when
dealing with pure technical systems. In domains
where implicit knowledge and human decision
making is part of the overall system behaviour, a
formal verification and validation is difficult to
apply. Far more feasible is the approach of using
concrete examples to validate the system behaviour.
The test data are created by the domain experts,
thereby ensuring that their implicit knowledge is
incorporated into the system functionality. In our
approach we use an expert system to validate and
verify the system architecture. This modus operandi
is actually a two-step process. In the first step a
behaviour model is created within the expert system
whose structure conforms to the system architecture.
In the second step the functionality or behaviour of
this architecture is implemented, verified and
validated.
In order to create a behaviour model within the
expert system the process model of Step 3 is
imported and displayed as a graph. The behaviour of
each model element is created by defining rules for
specific situations (case-based reasoning). Based on
these rules, decisions are made how to process a
specific input and where to send the result. The
model structure remains the same, but depending on
the input data respectively the situation, the output
might be different. The two main features of this
method are as follows:
Data-driven behaviour modelling: Based on our
experiences we realized that knowledge
acquisition and its implementation into systems
can be achieved more easily by using real-life
examples. Examples are close to reality and
easier to describe and to verify compared to
general and highly theoretical statements.
Therefore we use a set of training data, so called
training situations, in order to create a node's
desired behaviour and thereby to train the overall
model.
Consistent behaviour modelling: while defining a
rule for a process node the rule engine
automatically checks the consistency of that rule
with respect to all the existing rules. If an
inconsistency is detected the user is forced to
change the rule until the inconsistency is
eliminated. This mechanism is called
ICEIS 2010 - 12th International Conference on Enterprise Information Systems
426
"conclusion and justification". The user has not
only to define what decision based on the input
data has to be made, but moreover he has to
define why it has to be made. To justify a
conclusion the specific training situation has to
be used and therefore it has to be sufficient.
Once the behaviour model has been created, the
expert is able to validate and verify the behaviour by
using the test data in order to compare the results of
the behaviour model with the intended system. A
further output of step 4 are executable rules which
are used as parameters for the simulation in the next
step.
Step 5: Simulation of the Process- and System
Configuration. After the rules have been defined
using the case-based reasoning approach the process
diagrams have been transferred into a simulation
framework. This simulation is used to analyse and
optimise different process and technology
configuration alternatives based on time, resources
and other (changeable) parameters.
Using process simulation we are able to
determine weaknesses in some scenarios and suggest
and validate an optimized alternative process
configuration. Further an optimal resource allocation
can be computed.
These results give important hints and
justifications to the desired target architecture as
described in the next section.
Step 6: Refinement of the Target Architecture.
Using the described engineering steps we are able to
define an architecture that is consistent and fulfils
the operational and system requirements. The
consistency is reached by carefully designing the
architecture using an architecture model and
modelling tool environment that provides
consistency checks and gap analysis reports. The use
of simulation enables us to define an optimal
architecture configuration and resource assignment.
The use of case-based reasoning enables us to
create and manifest a consistent business logic that
is very close to the way the human expert works
because it is gathered in a machine learning
approach in which all justifications are visible and
traceable.
All previous steps (1-5) are performed iteratively
and enforce a constant update and refinement of the
requirements and the system architecture similar to
an agile method well known in software design.
4 CONCLUSIONS
Constantly new emerging threats in the domain of
infrastructure and border security require existing
and future security systems to be able to cope with
those new threats. Our engineering approach
accommodates these new requirements and enables
the engineering of a highly functional security
architecture by:
Applying case-based reasoning in order to gather
human behaviour and tacit knowledge from
domain experts.
Providing iterative and agile engineering steps
based on modelling and simulation.
Using military and civilian modelling
frameworks and standards for the architecture
models.
Validation and verification of the system
functionality on real case scenarios and expert
knowledge.
REFERENCES
Cole, M., et al., 2004. Optimisation of Critical
Infrastructure Protection: The SiVe Project on Airport
Security. In CRITIS’09, 4th International Workshop
on critical information infrastructure security.
Springer.
Wisnosky, D., 2004. DoDAF Wizdom: A Practical Guide,
Wizdom Press.
International Council on Systems Engineering (INCOSE),
2010. INCOSE Systems Engineering Handbook v3.2,
INCOSE.
NATO Consultation, Command and Control Board under
cover of AC/322-D(2007)0048-AS1, 2007. NATO
Architecture Framework Version 3.
MODEL - AND SIMULATION DRIVEN SYSTEMS - ENGINEERING FOR CRITICAL INFRASTRUCTURE
SECURITY ARCHITECTURES
427