Information Technology Reference
In-Depth Information
At this point, it should be mentioned that DEVS strictly distin-
guishes between model and simulator [151], and that extensions of
original DEVS specifically address agent-based modeling [131].
5.3.2 Simulated environment
Given the importance many authors ascribe to the modeling of
the simulated environment, these remarks have to be taken into
account [52, 146, 76, 70]. The environment is seen as a mostly pass-
ive background providing the common playground for the agents.
Nevertheless, the environment may possess its own processes which
may alter its state (e.g., some kind of diffusion processes) and may
therefore generate events which can be perceived by agents. In this
context it is important to mention that agents have to perceive their
environment explicitly and that this perception (depending on model
purpose and an agents sensors) can be erroneous (cp.
internal model
in [132, 131]).
Due to the variety of areas where agent-based models are em-
ployed, a reference model has to support different types of environ-
ments. Although mostly spatial environments (e. g., 2-dimensional or
3-dimensional) are used, other types of environments must be possible
(e. g., graph-like environments).
5.3.3 Simulation time
The simulation time has to be modeled explicitly and completely in-
dependent, i.e., without restrictions to issues like execution paradigm
(time-stepped, event-driven, etc.). Simulation engines may be imple-
mented rather arbitrary as long as they are conforming to the specified
reference model.
Explicit modeling of simulation time is hardly new to anyone fa-
miliar with modeling and simulation. It is nevertheless mentioned
explicitly due to many limitations currently often present in agent-
based models and simulations (like time-stepped execution with fixed
time increments, see Section 2.3.7). Another reason is that the terms
