|
|
SiE Basic Research Actions
|
|
|
|
|
SiE
|
|
|
|
During the "focusing" phase of SiE's operation, key subjects for
further research were identified within the broad area of modelling
and simulation.
In particular, five Action Clusters were defined. These clusters
correspond to needs of the users of Information Technology for
Modelling and Simulation.
The Working Group has started the second phase.
During this phase, actual Basic Research is encouraged in the areas
identified during the first phase.
In the sequel, we will describe the scientific content of these
activities.
Within the five Action Clusters, specific Subjects
were identified. These subjects and the corresponding Actions
reflect both end user needs and the availability of expertise in Europe.
Names (e-mail addresses) of Action Leaders are given in braces
next to each Action. Where appropriate, a link to the Action Home
Page is given.
1. Multi Paradigm Modelling
(
Hans Vangheluwe)
Given the growing complexity of systems and their models,
building and sharing heterogeneous (using multiple formalisms)
models is becoming increasingly important.
Action 1.1: Glossaries, Definitions,
(Jacques Lefevre)
Researchers from different backgrounds, in different
application areas use different terminology to describe
modelling and simulation concepts. This Action
classifies and discusses those concepts and their terminology.
This is particularly important in the area of multi-paradigm
modelling.
As complex multi-formalism models are developed and shared,
the problem of expressing the "meaning" of these models becomes
apparent. This Action investigates and explains the problem
and suggests solutions which will be of use in the standardisation of
modelling languages.
Action 1.3: Model Exchange
(Francis Lorenz)
From a practical point of view, a Neutral Exchange Format
for models has to be agreed on to allow meaningful model exchange.
Two levels can be identified: the generic model level
where the model type (ODE, PDE, Discrete Event, ... ) is
not yet relevant, and the formalism specific level
where the model type needs to be known.
In the continuous realm, starting points are NMF, PROFORMA, ALLAN,
ULM, Dymola, VHDL-A.
In the discrete realm, the starting point is DEVS.
The ultimate aim is multi-paradigm modelling. The continuous
and discrete starting points are boundary conditions.
Links with STEP/EXPRESS (product modelling) will be investigated.
Action 1.4: Symbolic Manipulation
(Hilding Elmqvist)
To enable the developments in standardised, non causal, continuous
model representations as well as to increase continuous simulator
performance, the symbolic manipulation of models needs to be
investigated. In particular, the needs, techniques, as well as a
standard architecture should be discussed.
As symbolic manipulation techniques are closely related to
the model representation, this Action also investigates continuous
model representations.
2. Simulators and Low Level Model Representation
(Per Sahlin)
Whereas the aim of Modelling is meaningful re use
of knowledge, the aim of Simulation is accurate and performant
``running of these models. Both aims are often conflicting,
which introduces the need for specific simulator level model
representation. At this level (of compiled models) also, re use of
models is desirable.
Action 2.1:
Simulator Standardisation Guidelines
(Brian Ricketts)
In general a simulator consists of a model representation
and a solver which will produce the state trajectory for this model.
If standard API s (Application Programmer s Interfaces)
are introduced for solvers and models, exchange of both
at link level becomes possible.
Standardisation of a solver model architecture will be investigated.
This entails the discussion of the use of multiple models
and multiple solvers. Also, the amount of symbolic information
``compiled into models needs to be determined.
This Action will investigate general Standardisation issues.
These will be used in the following Actions.
Action 2.2: Neutral Model Solver protocol
for continuous systems
(Martin Otter)
Will be based on the current specification of DSblock.
The extension of DSblock towards incorporation of more symbolic
information will be investigated.
Action 2.3: Neutral Model Solver protocol
for discrete systems
(
Herbert Praehofer)
Will be based on the current specification of DEVS.
The extension to existing discrete event simulation solutions will
be investigated.
Whereas the previous Action Cluster deals with the
local development of simulators as constructed from
solvers and low level model representations, this Action
Cluster investigates the interoperability of whole simulators.
Action 3.1: Interacting Distributed Simulators
(
Anders Christensen)
Based on the common behaviour of all simulators (from some
model description, generate a state trajectory), the design
of a generic architecture for interacting simulators can be attempted.
State of the Art technology such as CORBA needs to be investigated.
Action 3.2: Synthetic Engineering,
Distributed Interactive Simulation
(Richard Zobel)
The DIS community is extremely active and this action should provide
a gateway between SiE and the DIS community.
In this action, the civilian impact of DIS will be investigated.
In particular, Synthetic Engineering of a Virtual Enterprise
is a target for the application of Distributed (Interactive) Simulation.
4. Industrial Demonstrator Deployment
(Kaj Juslin)
Modelling and Simulation are powerful tools to analyse,
design and control complex systems. The use of different
modelling paradigms as well as the proliferation of
simulators will increase the quality of products while reducing
their cost. To effectively deploy modelling and simulation
techniques to support the full life cycle of tools and products,
users need more insight in the ``process of modelling and simulation.
Action 4.1: Industrial Deployment Handbook
(
Brian Hollocks)
Incorporating industrial input, this action will provide
guidelines for successful deployment of modelling and simulation
methods and technology.
Action 4.2: Multi Paradigm Demonstrator
(Jacques Lefevre)
It is claimed that complex problems can only meaningfully
and efficiently be solved using a combination of models/techniques
from different paradigms.
This Action will develop a demonstrator which shows the feasibility
of a multi paradigm approach. In particular, neural net,
genetic algorithm, fuzzy, continuous and discrete
models/techniques will be mixed.
5. End User Involvement,
User Simulator Interfaces
(Emmanuel Vergison)
The importance of appropriate user simulator interaction
is paramount. In this context, human factors issues were proven
essential in addition to purely technical considerations.
This action cluster will provide a link with the
EC Human Comfort and Security activities.
A. Additional Actions
The Working Group can only attain its goals no matter how good its
research and conclusions may be if there is a pervasive dissemination
of information about previous, current and future activities of SiE and
about modelling and simulation in general.
Action A.1: WWW server
The WWW server will contain general information about SiE
as well as present 3 different views of modelling and
simulation to the world (to assure broad impact):
- For users who are not aware of what modelling and simulation is.
- For current simulation practitioners who are interested in
- State of the art techniques/technology and standards.
for modelling and simulation researchers and tool builders
who want insight in new developments and standards.
The WWW server tries to help coordinate research in and use of
modelling and simulation by structuring and distributing
knowledge. It must support interaction through feedback from
its clients.
