=Paper=
{{Paper
|id=Vol-2010/paper15
|storemode=property
|title=None
|pdfUrl=https://ceur-ws.org/Vol-2010/paper15.pdf
|volume=Vol-2010
}}
==None==
https://ceur-ws.org/Vol-2010/paper15.pdf
Alberto Falcone
Distribution, Reuse and
Interoperability of simulation
models in heterogeneous
distributed computing
environments
Ph.D. Thesis
September 30, 2017
Copyright © held by the author.
Abstract
Systems Engineering is an interdisciplinary field of engineering and engineer-
ing management that is gaining a central role in a variety of industrial and
scientific domains ranging from e-science to space exploration, due to the
increasing complexity of system requirements and thus of the related engi-
neering problems. Systems Engineering represents one of the most important
and e↵ective methods for designing and studying both Large-Scale System
and System of Systems (SoS) over their life cycles.
A Large-Scale System is defined as:
“A group of subsystems that are interconnected and organized so to form a
whole system with clearly defined boundaries. Each subsystem is self-contained
and independent from the other ones but it cannot work individually.”
Whereas a SoS is defined as:
“A complex purposeful whole that is composed of complex, independent,
self-organizing, component parts whose high levels of interoperability enable
them to be recomposed into di↵erent configurations and even di↵erent systems
of systems; is characterized by poorly-defined issues that significantly a↵ect
its behavior and make it difficult to understand; has ambiguous boundaries
with critical contextual influences involving a mix of technical/non-technical
factors; and exhibits emergent nonlinear properties. The complexity of a sys-
tem of systems is a function of the number and diversity of its components
and their linkages. System of systems linkages range from loosely to closely
connected, but all systems of systems exhibit non-deterministic evolution and
behavior and are cybernetically self-organizing.”
In both the cases (Large-Scale System and SoSs), each component con-
tributes to the functioning of the entire system but, in general, the behavior
of the whole system cannot be straightforwardly derived from the behavior of
its components.
VI Abstract
Systems engineering utilizes systems thinking principles to organize this
body of knowledge and uses tools that include modeling and simulation
(M&S), requirements analysis and scheduling to manage the complexity of
such systems. By using M&S methods, tools and techniques, it is possible to
reproduce the structure and behavior of systems over the time so as to observe
and analyze them. The use of M&S techniques o↵er many advantages, such as
the possibility to study the behavior of a system without physically building
it, and the evaluation and comparison of di↵erent design choices, policies, and
operating procedures through experiments in a controlled (virtual) environ-
ment. Despite the above sketched advantages, M&S has important challenges
many of those related to the significant e↵orts required for producing a full-
fledged simulation model and analyzing simulation results. Moreover, it is
often hard to reuse already available simulation models; indeed, there is a
lack of mechanisms to make interoperable simulation models built on di↵erent
simulation platforms and a scarce support to enable their execution on dis-
tributed infrastructures. To overcome these challenges, many research e↵orts
are focusing on the definition of methods, models and techniques to support
the reuse and interoperability of simulation models and their execution on
distributed computing environment. Two of the most popular e↵orts going
in these directions are the Functional Mock-up Interface (FMI) and IEEE
1516-2010 - High Level Architecture (HLA) standards. However, each of the
two mentioned proposals addresses part of the above issues and great benefits
could derive from their combined exploitation.
In this context, the research presented in this Ph.D. thesis has been fo-
cused on the definition of models, methods and techniques to address, in an
integrated way, the issues of reuse, distribution, and interoperability among
heterogeneous simulation models. The research activity has been conducted
in cooperation with the Software, Robotics, and Simulation Division (ER) -
Simulation and Graphics Branch (ER7) of the NASA’s Lyndon B. Johnson
Space Center (JSC) in Houston (Texas, USA) where I spent nine months of
my Ph.D. program.
Starting from the research objectives above described, the main contri-
butions resulting from the research activity presented in this Ph.D. thesis
concern the definition of:
1. A software framework, which is called HLA Development Kit, that aims
at facilitating the design and develop of distributed simulators compliant
with the IEEE 1516.2010 - High Level Architecture (HLA) standard.
2. A Model-Driven method, which is called MONADS, that makes easier
for Systems Engineers to design a Complex System and simulate it on
a distributed simulation environment, without asking them to explicitly
deal with the intricacies and difficulties of currently available standards
and technologies.
3. Two methods, HLA for FMI and FMI for HLA, to address in an in-
tegrated way the issues of reuse, distribution and interoperability among
Abstract VII
heterogeneous simulation components through the integration of the func-
tionalities o↵ered by the HLA and FMI standards.
Concerning the first contribution, the HLA Development Kit software
Framework (DKF) is a general-purpose, domain-independent framework, fully
implemented in the Java language and released under the open source policy
Lesser GNU Public License (LGPL), which facilitates the development of HLA
Federates. The DKF allows developers to focus on the specific aspects of their
own Federates rather than dealing with the common HLA aspects such as the
management of the simulation time; the connection/disconnection to/from the
HLA RTI; the publish, subscribe and updating of HLA ObjectClass and HLA
InteractionClass elements. The DKF has been designed and developed in the
context of the research activities carried out within the SMASH-Lab (System
Modeling And Simulation Hub - Laboratory) of the University of Calabria
(Italy) working in cooperation with the Software, Robotics, and Simulation
Division (ER) of the NASA’s Lyndon B. Johnson Space Center (JSC) in Hous-
ton (Texas, USA). It has been successfully experimented in the SEE project
since the 2015 edition. In the last edition, the Universities of Calabria (Italy),
Bordeaux (France), Brunel (London, UK) and the Faculdade de Engenharia
de Sorocaba, FACENS (Brazil) developed their SEE-Federates by using the
Kit.
Concerning the second contribution, the result is a Model-Driven method
called MONADS (MOdel-driveN Architecture for Distributed Simulation).
The MONADS method aims at facilitating the distributed simulation of com-
plex systems, specified by using UML/SysML, according to the Model-Driven
Systems Engineering (MDSE) paradigm. Moreover, the HLA simulation code,
generated starting from SysML models by a chain of model-to-model and
model-to-text transformations, is based on the HLA Development Kit software
Framework (DKF). This research activity is carried out working in coopera-
tion with the Laboratory of Software Engineering, Department of Enterprise
Engineering of the University of Tor Vergata (Rome).
In the end, regarding the third contribution. Although the HLA and FMI
standards start from di↵erent objectives and are based on di↵erent techniques,
they have several common features that can be jointly exploited so as to create
a full-fledged solution to enable reuse, interoperability and distributed execu-
tion of simulation models. In this context, two solutions on how to fruitfully
combine the HLA and FMI standards have been defined: (i) HLA for FMI, in
which a FMU is enriched with HLA features and services; and, (ii) FMI for
HLA, in which simulation modules that are available as FMUs are reused in
a HLA simulation environment without modifying them.
The results achieved lead to an improvement in the Systems Engineer-
ing research domain. Specifically, the presented solutions provide a new and
flexible way on how to design and manage complex systems in distributed
computing environments.