Biomedical Engineering Reference
In-Depth Information
manager
component, and they are delivered to
corresponding destination objects according to
their expiration time, so that peers and agents
can operate concurrently along the simulation.
Moreover, the simulation environment exploits
the visual facilities and libraries offered by the
Swarm environment (Minar & al., 1996).
The simulator was integrated in the SO-Grid
portal through interactions with the SO-Grid
Controller and the file repository. In particular, the
Controller prepares the simulation configuration
file which contains the parameter values set by the
user, sends this file to the simulator and starts the
simulation. As the simulation proceeds, the partial
results are stored in the file repository, so as to let
the Controller access them. When the simulation
terminates, the simulator informs the Control-
ler, which in turn updates database information
pertaining to the running simulations.
the user before starting the simulation. A results
file contains the values of performance indices
(system entropy, number of replicas, discovery
response times, etc.) as they are periodically com-
puted by the simulator. A results file is organized
in rows and columns so as it can be given as input
to Gnuplot, a freeware software that offers batch
plotting utilities and is able to automatically build
graphical plots from data files. Gnuplot is used
both to visualize simulation results at run time
and compare results related to several simulations
performed by the same user.
The database stores two kinds of information.
The first is the account data (name, password,
e-mail address etc.) of registered users, which is
checked when a user requests to log on the portal.
Furthermore, the database maintains information
about the running simulations; in particular the
database associates the process id of each run-
ning simulation to the user which has requested
it and to the name of the file that will contain the
results. This helps the management of simula-
tions: for example, after the log in procedure of
a user, the portal checks if there is a simulation
started by this user which is still running, and in
this case the user is redirected to the portal form
which shows the results in real time. Moreover, the
database is used to control the number of ongoing
simulations: indeed, to maintain the performance
of the simulator at an acceptable level, each user
can execute only one simulation at a time and no
more than 5 simulations can be executed concur-
rently by different users.
File Repository and Database
Both the file repository and the database contain
information used by the Controller to manage
registered users, running simulations and simu-
lation results.
Specifically, the file repository is organized
in folders. Each folder is assigned to a registered
user, and essentially contains configuration files
and results files. A configuration file contains a
set of parameter-value associations and is given
as input to the simulator. Parameters can be re-
lated either to the simulated network (number of
nodes, average connection time of nodes, number
of resources, etc.) or to the bio-inspired protocols
(number of agents, values of probability function
constants, etc.). Each parameter has a default value
that can be modified by the user: the resulting
configuration file is saved in the user folder, in
order to maintain all information about simulation
settings, for future use.
Moreover, a user folder stores the files which
contain the results of the simulations performed
by the user. The name of a results file is chosen by
SO-GRID CONTROLLER
The SO-Grid Controller is the component that
receives, through the graphical interface, user
requests and interacts with other components to
satisfy them. It is built upon the J2EE technology
and its main functionalities are performed by Java
Servlets. Some of the tasks that the Controller
performs are the following:
Search WWH ::
Custom Search