Re: [Nel] TCP vs. UDP

[John Cosby]

The US Dept. of Defense has been doing simulation for decades.  When I first
read this thread, I though of my years-ago experience in Distributed
Interactive Simulation (DIS) and its successor, the High-Level Artchitecture
for Simulation (HLA).  Although designed primarily for LANs, both have been
successfully been used over WANs.

DIS is a protocol defined atop UDP.  Derived from work in the 70s, it's a
bit-level definition of Ethernet packets, the most important of which is the
Entity State Protocol Data Unit (PDU), which describes in all detail
necessary for rendering one particular simulation entity and was rebroadcast
whenever something changed over a configurable threshold (direction change,
speed, etc.) or a max waiting time occurred (5 secs).  Members of the
simulation were responsible for receiving and filtering things they cared
about, dead-reckoning the motion based on PDU updates, and broadcasting the
ES PDUs they were responsible for whenever the dead-reckoning thresholds
were hit or the entities timed out.  Other PDUs describe weapons fire,
detonation, weather effects, etc.  DIS was originally broadcast UDP.

HLA was an OO approach to simulation where Entities published Events, and
subscribed to Events that other Entities published.  It's based on classes,
and some reference object models are available.

There's a lot of stuff published about these topics, and an experience base
that's studied many of the concerns brought up.  Check the US Defense
Modeling and SImulation office (www.dmso.mil) and their HLA site
(www.dmso.mil/hla).  Also check the Simulation Interoperability Workshop
(http://www.sisostds.org/siw/) for lots of interesting topics and papers -
these people have been working on creating worlds for a long time.

-John

----- Original Message -----
From: Daniel Miller <address@hidden>
To: <address@hidden>
Sent: Sunday, July 08, 2001 12:44 PM
Subject: RE: [Nel] TCP vs. UDP

> These two points combined mean that, whichever way one looks at it, a lot
of
> scene information will have to be filtered out or sent at low frequency.
If
> one isn't careful low frequency sends can obviously be very sensitive to
> packet loss - which leads me to the point of this posting: In order to
work
> the problem we need to have a good understanding of true internet
behaviour
> and to have a good set of test data for simulating it.
>
> The trouble right now is that I have no hard data to use to model packet
> loss or delivery latency over time. If anybody knows of any studies that
> have been done or has any of their own data I'd be very interested. In
> particular I'm interested in moderately bad connections that exhibit both
> good behaviour and bad behaviour over time.