Information Technology Reference
In-Depth Information
content. The agents memorize a certain number of these messages and select their re-
sponse according to expectations calculated from the entries in their memories. That
approach shows the evolvement of stable interaction patterns from the agents' behavior
under a wide range of parameter conditions [3, sec. 3 and 5].
In an extension of their own model, Dittrich et al. furthermore examine the emer-
gence of social order among an arbitrary number of agents [3, sec. 6]. To this end, they
introduce a random choice of two agents in each simulation step, letting them interact
in the same way as in the basic dyadic setting. Their results show that, for growing
numbers of agents, stable interaction patterns only evolve if alter's behavior reflects the
average agent behavior within the system and if the agents are able to observe more
pairwise encounters than they are involved in themselves [3].
However, those requirements as well as their abstract model of message contents pre-
vent an application of that approach for self-organization in MAS following
particular purposes. Choosing agent pairs for interaction at random contradicts the ob-
jective of emerging agent relationships which define interaction channels. In fact, self-
organization refers to the systematic choice of interaction partners among the set of all
agents in a MAS in its very core. Thus, that selection must be based on expectations
regarding interaction outcomes. In applied self-organizing MAS (e.g., for modeling
supply networks), the semantics of message contents depending on the respective ap-
plication domain is a crucial factor for the determination of such outcomes. Hence, it
has to be considered when generating agent expectations.
Therefore, in the following, a model of double contingency is developed, based on
the basic approach by Dittrich et al. [3], allowing for the application of self-organizing
coordination of an arbitrary number of agents (Section 4.1). Moreover, the original
model using meaningless messages is enriched with semantics derived from the logis-
tics domain, being compatible with a standard agent interaction protocol (Section 4.2).
4.1
Modeling Double Contingency
In this model, agent operations consist of sending FIPA-ACL compliant messages [5].
Observing them refers to their storage in anagent'smemorywhichisusedtocalcu-
late expectations for possible further communicative acts. The observing agent subse-
quently selects its next message to be sent according to these expectations. Thus, an
agent's communicative behavior exclusively depends on its memorized observations
of other agents' behavior, avoiding any further assumptions of their internal properties
and characteristics. Hence, the basic steps enabling the agents to self-organize are as
follows.
1. The observation of incoming messages sent by other agents.
2. The selection of messages to be sent to other agents.
An agent's memory is a vector MEM
=
( mem 1 ,...,
mem n ) with a fixed length n ,where
each entry mem i denotes a tuple of messages m
M ( M being the set of all possi-
ble messages). The second message is the observed response to the first one: mem i =
. An agent possesses two of those memories, MEM ego and MEM alter ,
storing its own reactions to perceived messages and observed others' reactions to its
own messages, respectively. Thus, observation takes place when sending a message
m received , i ,
m sent , i
 
Search WWH ::




Custom Search