(——
) layer
. GQP
ts, (3)
rrently
+.
ation).
inistra-
e same
verage
odule.
yed on
Access
es. Ar-
) value,
will be
cess 1S
actly to
ail with
ore 1:4
bits 1s
Jata re-
|y users
an only
lent re-
An ATM network at OC-3 (155 MBit/s) and two FORE
Systems switches (ASX-200, ASX-100) are currently in
place for initial trials. An ATM connection to Vienna via the
high-speed DQDB (E3, 34 MBit/sec) network is scheduled to
be available on demand.
4.2 GDSS Software
Software is being created by means of several projects.
An Interactive Venus Atlas provides users with a tool to
interactively browse the surface of planet Venus. This can be
done via a GUI in different levels of detail. An important
goal of the project is to guarantee a cross platform unified
software tool. JAVATM is used to implement the Venus Atlas
as an interactive WWW page. A JAVATM supporting browser
like Netscape 2.x must be used to interact with the system.
The Retrieval Client implemented as a special purpose soft-
ware gives the possibility to optimally address the needs of
the GDSS. This could raise serious problems. The GDSS is
designed as a large distributed system with WAN geographic
scope. Hence many heterogeneous hardware platforms and
operating systems at the user sites could be involved. The
development of a version for each operating system would be
necessary. Furthermore each time a new version is
introduced a time consuming distribution has to take place
and every single user site has to install the new software
manually.
These problems can also be solved using the JAVATM pro-
gramming language. JAVATM supports the development of
so called “applets“, software pieces that are sent via the net
from the CS to the user site and are executed locally.
COSMO, an integrated development environment for
JAVATM which is available from Silicon Graphics Inc. will
be used for implementation. Due to using the Netscape's GUI
the application always has the same appearance independent
of the underlying operating system. In addition, the update
problem is implicitly solved by storing the "applets* at the
CS. The Web-browser will always download the latest
version via the Local Server.
The Spatial Data Structure is based on an ongoing investi-
gation of modified R-trees (Samet, 1990) for handling
points/regions of interest and coverage queries. Results of
this work will show whether a special more expensive spatial
RDBMS must be used or a simple SQL capable RDBMS will
address the requirements of the GDSS.
The Data Processing Server supports parallel execution of
user requested image processing algorithms. Some computa-
tionally intensive image algorithms have already been paral-
lelized. A tool named HUGO has been developed to
distribute user defined jobs throughout a work station cluster,
to collect the results and finally to present them to the user
(Goller, 95).
4.3 Data Experiments
The validity and efficiency of the GDSS concepts will be
tested by means of several remote access data experiments.
Distribution of the Interactive Venus Atlas: As soon as the
Interactive Venus Atlas is working locally we will perform
between the two institutes in Graz. The browser will be at
331
one site and map tiles at the other. Interactivity via an ATM
network will be verified.
Retrieval of full resolution data serves to list the WAN scope
of an ATM based full resolution data retrieval from Vienna.
Distributed Image Processing is being built around a port of
HUGO to SUN platforms using the public domain Parallel
Virtual Machine (PVM) library for interprocess communica-
tion. This library is supported by several different platforms
including Paragon, Meiko CS-2, SUN, SGI. As a result of
this project a set of parallelized image processing algorithms
will be running on a heterogeneous workstation cluster be-
tween the two participating institutes in Graz. Communica-
tion will be accomplished via an OC-3 (155 MBit/sec) ATM
link and a FORE ASX-200 switch. The speed up in compari-
son to a standard Ethernet (IEEE 802.3) will be measured.
Annotation of remote sensing data: Scientific personnel as
well as commercial users often need expert advice on the
image data they want to order. Currently this is a tedious
process in terms of not having the desired possibility to
annotate the image data on-line during the browsing process.
Participation in the European Union's DIANE project as an
associated partner of the VCPC in Vienna results in a pilot
trial showing the benefits of a multimedia annotation tool in
connection with an on-line video conference.
4.4 Simulations
The efficiency and functionality of the suggested network
layout is being tested using the CACI COMNET-III network
simulator. A realistic implementation of the GDSS with an
ATM WAN geographic scope must include applications and
data transfer. ATM specific details such as Quality of
Service, support of stream oriented traffic, isochronous
services and the influence of different switch buffer sizes on
the TCP performance (Hongqing, 1996) can be investigated
in more detail by means of simulations. '
5. FROM VENUS TO EARTH
A recent Austrian remote sensing initiative is a project de-
noted as *Multi-Image Synergistic Satellite Information for
the Observation of Nature* (MISSION) (Leberl, 1995). This
project combines more than 16 partners in a collaborative
effort. The requirements for accessing images greatly re-
semble those of the GDSS. Data from multiple missions such
as SPOT, Landsat, ERS 1/2, KFA-photography and MOMS-
Priroda need to be made available to all participants, each
having special interests and geographic requirements.
The Austrian Post and Telecom has introduced ISDN with an
areal coverage of 100%. We want to investigate the usability
of the GDSS concepts using ISDN technology by developing
a prototype scenario within the MISSION project. This
project could be extended to medical applications such as
remote diagnostics and on-line annotation by multiple ex-
perts at locations all over Austria.
The stimulus for the GDSS project drives from the planetary
image processing requirements, as reflected in NASA's PDS.
However, we are optimistic that ideas, concepts and software
of the GDSS can also be applied to international Earth-
observation projects such as the European Union's Center for
Earth Observation (CEO) or NASA's Mission to Planet Earth
(MTPE), and national programs such as Austria's project
MISSION. And we hope that the ideas, software and experi-
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B2. Vienna 1996
