Science at the Speed of Thought - Ambient Intelligence for Scientific Discovery - page 3

for a schematic). Examples of access points are the immersive visualization sys-
tem or a remotely connected workstation. For the purposes of this discussion,
data sources and analysis programs will be subsumed under computer experi-
ment. The interactions range from viewing the status of their currently running
experiment, or the results of a prior experiment, to providing feedback to the
experiment in order to alter or restart the experiment. To provide this range
of access requires a framework that enables communication between the user,
the experiment, the visualization, and possibly with other collaborators actively
interacting with the experiment.

Analysis
Tool s

Computer
Experiment

Laboratory
Experiment

Network available repository
of data and events

Immersive
Visualization

Fig. 2. Schematic of Distributed Computing Environment. Each component may also
access other data sources.

The VL uses what is generically referred to as a coordination framework to
provide a loose coupling between all participating entities of an experiment. The
coupling is loose in that participants, that is, laboratory experiment equipment,
computer simulations, visualization systems, user interfaces, and any other con-
nection to the VL, can choose whether or not to utilize this coupling and can
connect and disconnect from it at any time without disrupting the system. The
VL distributed computing framework is implemented using the Java technology
of Jini and JavaSpaces [9] [10]. Both of these packages, Jini and JavaSpaces,
are available as pure Java implementations and so are portable to any system
that supports a Java virtual machine, which includes systems running Linux,
Microsoft Windows, and most Unix based operating systems.
Jini provides support for a form of distributed computing that explicitly
handles common issues that arise in a distributed (networked) system, such as
intermittent network outages and server crashes. Support includes automatic
discovery of services available on the network, active leasing of services to help
maintain current service information and to purge services that no longer exist,
and distribution of events to remote applications to allow applications to com-
municate asynchronously and to react to expected or unexpected developments.