Information Technology Reference
In-Depth Information
is not an obvious one. As the possible configurations of message receivers for each
order correspond to the power set of the set of available suppliers (without the empty
set), in a network with
n
tiers and
m
parallel actors at each tier, the total number of
potential relationships is (
m
1))
n
−
1
(the possible communication paths through the
network)
2
. Thus, in the chosen scenario the self-organizing agents can choose between
441 possible interaction patterns leading to different performance rates. Therefore, in
this simple scenario, agent coordination is already complex enough to make it suitable
for evaluating the emergence of communication structures.
For this purpose, the expectation-based agents are configured as follows. The set
of possible orders to be sent by a customer is given by the possible combinations of
their receivers, their performatives, and their content. As there is only one type and a
fixed amount of units to order per customer, there is only one possible content. The
same holds for the performative, as an order is always a REQUEST message. Thus,
the set of possible orders is determined by the possible combinations of a message's
receivers (the power set of the set of possible suppliers). For the replies, on the other
hand, the receiver as well as their contents are preassigned by the incoming orders.
Hence, a supplier's only choice is between the message performatives according to the
FIPA-REQUEST interaction protocol.
For generating the results presented in the following subsection, the constant values
are based on those used by Dittrich et al. [3]:
c
M
=
(2
m
·
−
2and
c
f
=
0
.
02. The agent memory
size is set to
n
=
25 for both
MEM
ego
and
MEM
alter
, the balance between EC and AE
to
3. All agent memories are
initially populated with randomly chosen messages in order to reflect the agents not
having any specific prior information about promising interaction channels.
In order to validate the approach to expectation-based self-organization, it is com-
pared with an optimal configuration as outlined above. The performance is measured
with regard to the number of receivers per order, the final consumers' customer satisfac-
tion rate (i.e., the number of fulfilled orders), and the utilization of the final consumers'
product consumption. The first two criteria directly reflect the customers' utility func-
tion. They give information about the communication effort needed to operate the net-
work (message receivers) as well as about the reliability of the emerging relationships
between the agents (customer satisfaction). Thus, these measures reflect the extend of
stability of the evolving network structures. The consumers' utilization, on the other
hand, is an additional logistics performance measure that allows for validating the sup-
ply network's overall operating efficiency in terms of product throughput rates.
α =
0
.
5, and the customers' selection value gain to
γ =
5.2
Results and Discussion
The results depicted in Figures 4-6 show the number of receivers, the customer satisfac-
tion, and the consumer utilization as average values over 200 simulation runs. Each run
consists of 1000 production and/or consumption operations. For the calculation of the
order fulfillment rate, the last ten messages received are considered for each time slice
2
There are
m
agents at a tier with 2
m
−
1 possible interaction partners, each. The potential paths
through the network are given by the combination of those options over all
n
−
1 links between
two tiers.
