Information Technology Reference
In-Depth Information
A comparison of agent-based modeling with established modeling
techniques (e.g., cellular automata or queueing networks) can be found
in [71]. Another approach in microscopic modeling is object-oriented
simulation which has its roots in the 1960s [20, 62]. Basically, 'an
object-oriented simulation consists of a set of objects that interact
with each other over time' [62]. There are many commonalities
between object-oriented simulation and agent-based modeling (like
classes of similar entities, encapsulation of data and methods) [62, 20],
[36, p. 45f.]. Nevertheless, there are also some differences, although
mainly on a conceptual level. Within object-oriented simulations the
simulated entities are usually not modeled using mentalistic concepts
(e. g., beliefs, desires, intentions), the notion of space is less stressed,
and objects are typically purely reactive (in contrast to autonomously
acting agents) [27].
A precise distinction between agent-based modeling and object-
oriented simulation cannot be made. In a way, agent-based modeling
is more advanced on the conceptual level with ideas like autonomous
agents, variable interactions and decision-making autonomy [68]. At
the same time, each agent-based model uses a lot of ideas originally
developed in the context of object-oriented simulation. Therefore,
agent-based modeling 'is not the same as object-oriented simulation,
but the object-oriented paradigm is a useful basis for agent modeling,
as an agent can be considered a selfdirected object with the additional
capability of action choice' [83].
Within the context of discrete event simulation several modeling
views are common (e.g., process view or activity view). Similarly,
agent-based modeling may be considered as another modeling view
for describing a system under investigation. Therefore agent-based
modeling is not something completely new but rather provides a new
mindset.
Discrete event simulation uses events as central element, where
events happen at a specific point in time and trigger state changes.
The central notion of agent-based modeling is the agent and agents are
supposed to be heterogeneous, active entities with individual behavior.
Accordingly, agent-based modeling focusses on the behavior of single
