Information Technology Reference
In-Depth Information
The mechanism for generating agents in the system is related to the necessity
of the proper selection of the number and types of generated agents, as well as the
selection of the appropriate way of inserting them into the system.
The mechanism for removing (liquidating) agents from the system is related to
the necessity of considering the problems of the termination of missions performed
by these agents.
The operation of the re-skilling of agents may be realized by the mechanism for
generating and liquidating the appropriate agents, as well as using the concept of
unemployed agents.
The above examples illustrate the possibilities of the agent systems in the field of
control of the type and the number of agents, and matching the system to the needs
of a task realized in real time, hence the ways of their scaling. It is noteworthy that
the concept of the unemployed agent may be used together with the mechanisms
presented above. If there is a need to reduce the number of agents, the unemployed
agents may be removed in the first place. This agent does not have any task to realize
which would be connected with the particular functionality of the system, hence
its liquidation will not disturb any functionality of the system. Similarly, if there
is a need to increase the number of agents, then generating unemployed agents is
the simplest solution. These agents should find a task to realize and automatically
contribute to the support of the realization of that functionality of the system which
is necessary at that moment.
It seems that the concept of the unemployed agent, i.e. the software searching for
tasks to realize may become the basic element of the large-scale systems operating
in cyberspace.
4.6 Illustrative Results of Research into Balancing
the Resources in the Multiprocessor Structure
Universal system for the simulation of the agent systems (Universal System for the
Simulation of the Systems of the Autonomous Agents), developed at the Department
of Computer Science of AGH University of Science and Technology, was used for
the research on the system.
With the use of this system the simulation of the process of balancing the resources
in the multi-processor structure was carried out. The structure, which was realized
in the form presented in Fig.
4.1
, consists of 400 processors constituting the nodes.
The processors (nodes) are identical and may perform the same calculations. Each
processor is connected 8 processors referred to as its neighbouring processors. Tasks
may be sent and agents may relocate through these connections between the neigh-
bouring processors. What is more, all processors are connected via the interface
trough which tasks may be sent between the two optionally chosen, and defined as
the sender and the receiver (Fig.
4.1
).
