Ladies and Gentlemen, we are DOOMed
Geoff White
The first issue of WiReD has an article by Bruce Sterling on
virtual battlefields, he described something called the Defense
Simulation Internet, he also talked about some emerging protocol
that the military types were creating, it seemed, to be able to
connect their simulation facilities (spread out over the globe)
via the Internet and to carry out virtual manuvers.
Anybody have that WIRED on hand?
Craig Partridge
> Anybody have that WIRED on hand?
No, but I know a little bit about the Defense Simulation Internet (DSI).
Essentially folks sit in simulators (tanks, airplanes, trucks, etc.)
and direct their simulators around the simulated battlefield. The
simulators send out periodic packets (about 1,000 bits long on
average) describing the simulator's condition (i.e., I'm a tank moving at
10 MPH in a given direction and my gun is raised X degrees). Each
simulator uses the information it hears from other simulators to predict
what other objects nearby are doing. In other words, based on what it hears
from your simulator, my tank computes how your tank will likely behave for
the next few seconds.
The simulated vehicle also computes its predicted performance, and it is
instructed by its driver to do something unexpected (a sharp turn, a gun
firing) it spits out a new update. [There were fun glitches in some
early simulators in which the updates weren't done quite right -- by
speeding up and then stomping on the brakes you could cause your tank to
appear to jump backwards as the predictive algorithm overpredicted your
tank's location and the image was then corrected -- really fooled enemy
gunners :-)].
Currently, I believe (I've misplaced my notes from a recent meeting),
everyone hears everyone else (broadcast). In the future, what a simulator
hears will depend on the part of the simulated terrain it is in (i.e., one
listens to, say, a multicast group representing a grid on a map, and listens
to only those grids representing folks in one's line of sight, or within
shooting distance).
The DSI folks are currently working on how to scale the DSI and its
algorithms to very large simulations (many 10's of thousands of entities).
My personal sense it that it all looks feasible with current technology
(perhaps tweaked slightly) but the jury is still out.
Craig
PS: For those interested, the simulation protocol is now an IEEE standard
(IEEE 1278-1993). A new version is to be balloted in 1994.
Matt Crawford
> [...] some emerging protocol [...]
I haven't read WIRED but the DSI is run by Loral Western Development
Labs and the protocol you refer to is probably DIS 2.0 (Distributed
Interactive Simulation) format, carried in UDP/IP multicast
datagrams.
_________________________________________________________
Matt Crawford cra...@fnal.gov Fermilab
Michael A. Patton
From: Geoff White <geo...@nexsys.net>
The first issue of WiReD has an article by Bruce Sterling on
virtual battlefields, he described something called the Defense
Simulation Internet
Greetings. I'm one of the engineers for the Defense Simulation
Internet here at BBN. If you want to ask specific questions, I'd be
glad to help (technical questions I can answer myself, others I may
have to refer you to the right person).
he also talked about some emerging protocol that the military types
were creating, it seemed, to be able to connect their simulation
facilities (spread out over the globe) via the Internet and to
carry out virtual manuvers.
We already do this, the original simulation protocol was _not_ designed
with wide-area use in mind and we have had to employ some engineering
compromises to get it to work. Newer protocols are in development
which will improve this situation. The DSI today connects sites all
over the US and in Europe and Korea for simulation, video
conferencing, and other uses.
Anybody have that WIRED on hand?
WIRED is on the net... Follow this URL to see the article in question:
http://wired.com/0h/1.1/features/virthell
I do recommend you get the magazine if you can, the pictures are both
interesting and helpful but don't seem to be in the on-line copy.
__
/| /| /| \ Michael A. Patton, Network Analyst
/ | / | /_|__/ Defense Simulation Internet
/ |/ |/ |atton Bolt Beranek and Newman
Paul Mockapetris
ARPA through various contractors. My understanding is that DIS
specifies standards for objects to describe what's going on in the
simulated world via PDUs, with reference to the laws of motion.
The scheme uses IP multicast in the local environment, but not in the
long haul. That role can only be performed by ST multicast (right
now) although certain rebels have argued for IP multicast as the wave
of the future. In any case, DIS isn't the only simulation protocol in
use on the DSI, and all engineering decisions are much more
complicated than a normal person can stand given crypto and backwards
compatibility issues.
In today's DSI the visions of multiple dynamic multicast groups are
mostly that, and the resource allocation issue is simplified by what
amounts to a private net. While a bunch of the engineering is
painful, the challenges are extremely interesting.
paul
ka...@gropo.tgv.com
> > Ever hear of something called "dogfight"? A multiplayer game in
> > which participants "fly" flight simulators around and "shoot" at
> > each other? [ ] Its traffic is forwarded in real life, non-trivial
> > internets using DVMRP as well as "tunneling."
>
> I have heard two different figures on DOOM. One said 150 packets per
> second, another said 1500 pps. I chose to take the lower figure because it
> sounded more reasonable.
This dialog sounds as if we were all isolated from the day-to-day
major traffic generators which are found on many, if not most networks
-- among these are NetWare and DEC's LAVC.
By comparision, the TCP/IP suite of protocols is very well behaved.
Even NFS has a hard time generating as many packets as NetWare.
The network here generally runs 2500+ packets/second. And only a
small part is IP.
--karl--
Geoff White
> Date: Wed, 29 Dec 93 11:18:40 PST
> From: Geoff White <geo...@nexsys.net>
>
> The first issue of WiReD has an article by Bruce Sterling on
> virtual battlefields, he described something called the Defense
> Simulation Internet
>
> Greetings. I'm one of the engineers for the Defense Simulation
> Internet here at BBN. If you want to ask specific questions, I'd be
> glad to help (technical questions I can answer myself, others I may
> have to refer you to the right person).
>
> he also talked about some emerging protocol that the military types
> were creating, it seemed, to be able to connect their simulation
> facilities (spread out over the globe) via the Internet and to
> carry out virtual manuvers.
>
> We already do this, the original simulation protocol was _not_ designed
> with wide-area use in mind and we have had to employ some engineering
> compromises to get it to work. Newer protocols are in development
> which will improve this situation. The DSI today connects sites all
> over the US and in Europe and Korea for simulation, video
> conferencing, and other uses.
>
The point of this whole thing is that DIS (or whatever it will be called
IS an emerging protocol that can be used by people writing "games" or other
forms of highly interactive multi -user software. Whether the likes
of Sega and Nitendo adopt DIS (or a subset) or develope their own is another
question but to say that the Internet is doomed is a little extreme.
> Anybody have that WIRED on hand?
>
> WIRED is on the net... Follow this URL to see the article in question:
> http://wired.com/0h/1.1/features/virthell
> I do recommend you get the magazine if you can, the pictures are both
> interesting and helpful but don't seem to be in the on-line copy.
I have read the article, (which is why I made the post in the first place)
Are there any public documents if DIS2.0 (or what ever its currently called?)
geoffw
Edward Feustel
Yes, they are available from the IEEE (DIS 1.0) and University of
Central Florida Institute for Simulation and Training, Orlando Fl. (DIS
2.0 and DIS 3.0 Draft.
: geoffw
--
Edward A. Feustel, Research Staff Member
Institute for Defense Analyses Phone: (703)-845-6657
1801 N. Beauregard Street Fax: (703)-845-6848
Alexandria, Va. 22311-1772 Email: efeu...@ida.org