Information Technology Reference
In-Depth Information
Fig. 2.
Histogram representation for the utilities of the coalitions
actors among only a few coalitions [1]. Single transfers have been shown to have the ten-
dency of diffusing the spread of actors into more evenly configurations. This is clearly
a desired state for our organization structure. Another important aspect achieved as a
result of this is as well, performing a minimization of the structural adaptation required.
Moreover, the organizational model proposed proves to be able to operate in open
environments and dynamically stabilize behavior while actors are being added or re-
moved to the system. In fact its adaptable features leverage on the inherent variations in
the system permitting it to escape local optima. In Figure 1(b) we plot a less usual in-
stance, where the system undergoes considerable variations during the adaptation phase,
as some of the actors of the coalitions are removed (possibly due to failures) and some
have joined the system. We consider a rather extreme situation where the proportion of
coalitions that experience such modifications is considered at 30%, while within each
coalition up to 10% of the actors have been disconnected or alternatively, have been
appended to the system. The system demonstrates a capability to reorganize and reach
a stationary configuration as the spikes injected into the system are flattened in a small
number of steps.
Subsequently we perform a series of experiments to draw more insight to the solu-
tion concept introduced. On one hand, our negotiation scheme implies that deviations
would only occur if a certain minimum gain
has been achieved. On the other hand,
the extent to which a coalition is willing to decrease its efficiency in detriment of the
gain in social welfare is represented by a satisfactory parameter
τ
. This represents in
effect a percentage, which defines what an acceptable performance would be and how
tolerant is one coalition towards suboptimal performance. For our simulations we chose
an initial value of 0.4 and considered a homogenous population of actors in the system.
Although this does not make the objective of our scenario, heterogeneity amongst the
actors involved may as well be introduced.
Following, we analyzed the implications of the dependency on this parameter for a
better understanding of its functionality. Thus, we have analyzed the stationary states
the system falls into as a function of
μ
μ
. For large values of
μ
, meaning that coalitions
