TECH Re: Distributed Interactive Simulation (DIS)

John W. Barrus (
Fri, 1 Jul 1994 07:54:55 -0400

>Discussion and evolution of a VRML would be incomplete
>without considering the work of DIS (and SIMNET). DIS is
>actually more along the lines of multi (10-10,000?) player
>interactive VR rather than VR based information navagation as VRML
>seems to be heading, but it's still worth looking at.

I wanted to clarify a few things about DIS.

DIS has no specification (to my knowledge) of an object file format.
Michael Zyda and his crew at the Naval Postgraduate School in Monterey, CA
have created a Simnet-style simulation platform called NPSNET. They have
their own data format, etc.

DIS specifies how different simulators on a common network can communicate
effectively with many computers at many different sites. Network bandwidth
was considered invaluable in the design and one of the basic premises of
the original SIMNET system (the origins of the DIS spec) was that every
simulator starts out with a complete model of the world. Therefore, in
DIS, there are no PDU's (Protocol Data Units - used to communicate
information to other simulators) that handle the transport of models
between simulators. This also means that there is no need for a common
object format between simulators. In fact, a radar simulator needs no
graphical representation of most objects (blips on the screen) where a
high-end, texture mapping, out-the-window simulator needs textures,
levels-of-detail, and all the polygons.

In other words, we could look to DIS to determine how to communicate
dynamic information over the network in real-time (which is what it is good
at). Or, we could look to the Naval Postgraduate School, the Institute for
Simulation and Training in FL, Loral (where the BBN group who started
Simnet ended up), SGI (who makes performer) and find out about all of their
data formats and then try to pick the best.

A note: SGI's Performer has an internal data format that is useful for
high-frame-rate, out-the-window simulations. However, they have "loaders"
that convert from many different input formats to the correct internal
format. This might suggest to us that we consider the 3D file format as a
separate issue from the VRML format. The VRML file could tell us the
location of the 3D object files and how the objects should be positioned in
the world. I don't necessarily think this is the right thing to do, but
that is what is done with images and other file types in the WWW today.

John B.


John Barrus Research Scientist

Mitsubishi Electric Research Laboratories, Inc. voice 1.617.621.7535
201 Broadway fax 1.617.621.7550
Cambridge, MA 02139