MO (Modeling Ontology)
DEMO (Discrete Event Modeling Ontology)
Introduction (still a ROUGH SKETCH).
DEMO is an ontology for
discrete-event modeling (system dynamics for discrete systems). Here DES is understood as an event-driven
system with discrete states. System evolution can thus be described by a
sequence of ordered pairs (state, time): {(s0, t0),
(s1, t1), …} with an assumption that for each
change of state sk® sk+1 there exists an event e
(or a set of events E*), that caused this change.
Our first goal is to define a
hierarchical (in some sense) set of relevant modeling formalisms and
relationships between them. We start with a smaller set in order to develop
necessary building tools and approaches.
The taxonomy can be
established based on different criteria and it is not quite clear for us yet,
which one will prove to be the most useful. Still, we feel that the first thing
to do is to establish a hierarchy based on theoretical modeling power. In other words we want the formalisms on top
of the taxonomy to have larger classes of discrete-event systems that they can
model compared to the formalisms below (see the diagram).
We start by placing two
equivalent (in modeling power) formalisms on the top – Generalized Stochastic
Petri Nets and Generalized Semi-Markov Process models.
There are three main
components of the formalisms in consideration:
1.
Underlying
graphical representation;
2.
Probabilistic
transitions; (these may be also considered as a part of 1)
3.
Stochastic clock
structure that introduces time in the model and is used as an input;
Stochastic clock structure is
completely independent of 1 and 2. This suggests that the taxonomy can grow in
(at least) two orthogonal directions: one based on the underlying graph and
another based on stochastic clock. In other words one can view it as (at least)
a two-dimensional hierarchy. In the
first dimension it grows down by putting certain restrictions on graph
properties: topology, set cardinalities, connectivity, probabilistic
transitions, etc. In the second the
restrictions are applied to stochastic clock properties: clock distribution
functions, clock rates, and so on.
(Note: The diagram we have right now does not reflect the
two-dimensional structure yet, but rather both directions are jumbled up in
one.)
Source files in Word document format: DeMo-intro.doc,
DeMo-diagram.doc