DHPC Talk DHPCT-014
Jade: A Geographical Information System as an
application of Distributed Process Networks
Darren Webb
Archived: 21 May 1998
Seminar presented as part of the
Computer Science Honours Program Series, on 20 May 1998.
Abstract
The steps used to perform operations on satellite imagery can be thought of as a process
network. A process network is a directed acyclic graph where the nodes represent processes
and the edges represent dataflow between processes. An important property of process
networks is implicit parallelism. Parallelism in process networks is implicitly achieved
through parallel computation of independent processes or by overlapping communication
with computation. The process network is a high-level notation for compositions of more
primitive functions. It provides a well-defined semantic model facilitating construction of
higher-order functions. To derive parallelism, we extend the concept of a process network to
a distributed process network where processes are executed across distributed computers.
We have developed a Geographical Information System (GIS) as a proof of concept for
distributed process networks. A GIS is intended to aid domain experts make accurate and
quantitative decisions by providing tools for comparing, handling, and processing data
about geographic areas. The GIS we have developed provides the domain expert with a
sophisticated interface for manipulating GMS-5 satellite imagery with the Generic Mapping
Tools (GMT) mapping library.
PostScript version of the slides
[
DHPC Home |
Projects |
People |
Organisations |
Reports |
Talks |
Research |
Education ]
webmaster@dhpc.adelaide.edu.au