Advancing Engineering with AI through the Next Generation of Strategic Projec...
Agena adsif – a language for simulation & agent
1. Agent driven Simulation Framework
With mult-paradigm and AI Support
AdSiF – A Language for Simulation
& Agent Programming
2. AdSiF: Simulation Language
What is AdSiF (Agent driven Simulation Framework)
AdSiF is a family of commercial off-the-shelf (COTS)
software products that provide the framework and all
basic functionality needed for constructive simulations
as well as the ability to interface with live, virtual, and
other constructive simulations.
Using AdSiF, it is possible to develop a simulation
customized for your exact requirements that is more
flexible and easier to use than simulations that are not
based on AdSiF. Using AdSiF can also substantially
lower the cost of developing, upgrading, maintaining,
and using your simulation. If you don’t have your own
simulation development team, Agena’s experts can
develop a custom, AdSiF-based simulation for you.
3. Language Properties
AdSiF is a declerative general purpose simulation language.
The language is developed for generic pupose – there is no
domain dependency.
It is successful in simulation and intelligent agent programming.
Multi-paradigm design environment
Object oriented
Aspect oriented
Agent based modeling
Logic programming
Combine whole paradigms it supports in State oriented
programming paradigm.
Easy-to-Extend simulation models
State diagrams (SOP)
Atomic function plug-ins (SOP, AOP)
Logic programming predicates
AdSiF
LP
AgP
OOPAOP
SOP
4. Language Properties
Extending simulation models by adding new methods,
even in run time
Support for Aspect oriented programming by changing
behavior containers in run time.
Supports discrete and continuous event simulation
Dynamic execution frequency
Stiffing
Controlling double precision and fourier error
Behavior management by deductive reasoning
Ability to Draw whole Conceptual Model
Starting from requirements up to simulation execution
Generic logic based solution for time delayed systems
5. Ontology & Aspect Orientation
Supports Ontology based Modeling
Ontological commitment
Existance queries
Relation concept
Provides tools
Basic concerns of AOP
Tangled requirements/concerns
Scattered requirements/concerns
Swithing to an aspect satisfies a croscutting concern.
What does it mean in simulation ?
Thinking with agenthood
Evaluating states, making decision and changing
attitude
6. Some More Features
Logic
programming,
Agent based
programming,
Aspect oriented
programming
enhance
reasoning
mechanism,
Planning
capability
modelling,
Commitment
strategies,
and
Knowledge and
decision
intensive tasks.
Artificial
Intelligence
In run time
simulation
models can
change their
aspect
depending on
the conditions
user defined.
The capabilitry
provides more
intelligent and
flexible
simulation
environment.
Dynamic
Aspects
Interaction
between
entities are
achieved
publish and
subscribe
mechanism
and event
passing. No
code
dependency
required that
makes high
coherence and
low
dependency.
Interchangability
Using AdSiF
plugin and
Aspect
Oriented
Programming
paradigm it is
highly easy to
extend legacy
simulation
models and
applications
even in
runtime.
Reusability
Possible to
extend models
with plugins
even in Run
Time
Extendability
Extend
simulation
models in run
time
Modular, model
based modeling
w/o Coupling
High
Coherence, low
dependency
Change model
behavior aspect by
autonomous
reasoning
Uses
Reasoning
Technologies
7. Distributed Simulation
Supports HLA and DIS
Dynamic paralalization algorithm selection in run
time
Optimistic algorithms
Pessimistic Algorithms
..
Altering simulation resources in run time
Extending capabilities of simulation models in run
time
Extending by new functions
Adding new attributes
Adding new behaviors
Switching behavior categories in run time
8. Replications & Snaphots
Create a new replication and a snapshot by
user interaction
as a result of decision making
Depending on statistics computed during execution
Schedule replication plan & snapshot plan in
scenario design time
Design a behavioral plan for each simulation
model for both replications and snaphots.
9. Run Time Analysis & Decision
Making
The purpose of run time analysis is to
Pick execution data during execution
Analyze them
Make a decision
Handle execution dependig on the decision given
Handling an execution consists of
Altering resources
Adding new functions, attributes, and behaviors
Swithcing parallel simulation synchronization
algorithms
10. Intelligent Traces Mechanism
The data generated by a simulation entity can be
traced depending on designer choice. These are
Atomic actions
Attributes
Logical premises
Behaviors executed
State transitions
State durations
Event flows
There is no software dependency between trace
declaration and simulation software
Behavior traces give a grammar to learn and query
what the simulation model did during execution.