Transcript 01_intro

MV-3500
DoD Modeling & Simulation Network
Protocols
Admin
This class is intended to introduce you to the
simulation network protocols used in DoD
M&S applications
This will require a little background on
networking first, then we can look at two of
the major protocols, DIS and HLA.
Admin
Grading
• Two midterms (50%)
• Final project (25%)
• Programming assignments (25%)
Contact:
Watkins 281 (stop by any time)
Phone (don’t) x-7605
Email:mcgredo at nps edu; put MV3500 in the subject
Web page:
https://www.movesinstitute.org/~mcgredo/MV3500
Textbook
Networked Graphics, Steed & Oliveira
Read chapters 1 & 2
Chapter 2 will probably be over your head a bit;
just roll with it.
Other Sources
“I want to know everything about TCP/IP networking”
Stevens: Unix Network Programming and TCP/IP Illustrated (V 1,
2, 3)
Zyda: Networked Virtual Environments
Gaffer on Games: http://gafferongames.com/
Game Developer’s Conference
Some commercial techniques are encumbered with intellectual property
restrictions
Neyland: Virtual Combat: A Guide to DIS
Nicol, Fundamentals of Real-Time Distributed Simulation
SISO: DIS Plain and Simple
Alexander, Massively Multiplayer Game Development
Epic Games, Unreal Networking Architecture
What Do We Want to Do?
Live, Virtual, Constructive
The military wants to do a variety of
simulations, including but not limited to
Live: Real people, real systems
Virtual: Real people, simulated systems
Constructive: simulated people, simulated
systems
Live
An E-5 is driving a Stryker at NTC. The Stryker is
generating Blue Force Tracker data.
This is a real person driving a real vehicle. Perhaps he
sees other live vehicles on his BFT.
To increase the training experience, we might want to
display simulated enemy forces as well.
The training exercise is now no longer completely live—
the BFT may be displaying computer-generated
forces as well, a constructive element
Virtual
An F-18 pilot is in a 6-axis dome simulator. This
is a simulated weapon controlled by a real
person.
We can inject the Stryker BFT data into the F18 simulator so that the pilot sees Strykers on
the ground on the simulated NTC terrain
The training now has live and virtual
components
Constructive
Suppose we don’t have all the live simulation red force
necessary to achieve the training objectives.
We can create computer-generated, AI-controlled tanks
and inject them into the training environment
The E-5 on the Stryker may see these on his situational
display
The F-18 pilot can see the computer-generated forces
in his dome terrain
The training is live, virtual, and constructive
How Do We Get There?
Notice that these all involve sending data between computers,
sometimes computers that are a long ways apart geographically
What data do we want to send?
What semantics do we need?
What about practical issues, like latency and reliability?
What if the computers are running different operating systems on
different CPU architectures?
What if the Stryker is using BFT and the dome simulator something
else?
Can you come up with some examples of the state data that we
need to exchange?
Data
(X,Y,Z), orientation, what entity type, terrain,
velocity, unique entity identifiers, ….
Semantics
Semantics are “meaning” associated with the values we send.
Three floating point numbers (X,Y,Z) isn’t enough—also implied is a
coordinate system with which those values are associated.
These are the semantics associated with the three floating point
numbers
Are they referring to lat/lon/altitude? MGRS? What units? Where
is the origin? Spherical or rectilinear?
How do we determine a position message is referring to a
particular M1A2 tank?
Semantics is extremely tricky to do in the general case. It is
basically all the assumptions you make about data values
Practical Issues
The speed of light limits network traffic speed, and the
reality is often much worse. Human reaction time is
around 250ms, and long haul network latency can
often approach or exceed this.
~10ms speed of light latency per time zone, ~250ms
for geosynchronous orbit, minimum
This can make “twitch” applications tricky; air to air
combat, first person shooter, etc
What if we send a message and it doesn’t arrive?
What Do We Want to Do?
“Get relevant data to other computers in a
format that can be understood, fast enough
to be useful”
This problem statement also happens to
coincide with a lot of other commercial
applications. The DoD uses the standards
adopted by industry for “basic plumbing”, and
adds its own “protocols” for its applications
Commercial Applications
First Person Shooter Games: Modern Warfare, Call of Duty et al
Massively Multiplayer Online Games (MMOG). Large scale, with
thousands or tens of thousands of participants. World of Tanks
Online worlds--live simulations and role playing in a virtual world;
less emphasis on physics, more on human interaction. Second
Life, Open Simulator
To an extent there has been a convergence between the
capabilities of military & commercial, but commercial is heavily
focused on games
What’s the difference between training and games, anyway?
More of an issue of the goals than the technology
Protocols
What do network protocols do?
Basically, they are an agreed-upon way for
applications to talk to each other
If you have a Boeing flight simulator and a
General Dynamics tank simulator and a
Hughes helicopter trainer, it would be helpful
if all of them could cooperate in the same
networked virtual environment
Networked Virtual Environment
(NVE)
GD Tank
Simulator
Sikorsky Helo
Simulator
“NVE?”
Huh? What’s a “Networked Virtual Environment?”
An illusion of a virtual environment that is shared
between participating users. This is typically done
across a network. In the tanks & helos example, we
have a virtual battlefield, users controlling vehicles,
and each user can view the virtual world from their
own point of view.
Protocols
So how do simulations talk to each other?
Generally this happens across the network (“distributed”) as
multiple computers cooperate on the problem.
The substrate used by all modern simulations is TCP/IP sockets.
This is a protocol, but DoD simulation protocols are built on top
of TCP/IP
Simulation protocols started with ad-hoc protocols, made up for
each task
Moved on to Distributed Interactive Simulation (DIS) and High
Level Architecture (HLA)
Web-based protocols are emerging
Simplified Diagram
Host 1
The Application:
OneSAF, VBS2, etc
Host 2
The Application:
OneSAF, VBS2, etc
“Semantics”: Simulation
Protocol (DIS, HLA, etc)
“Semantics”: Simulation
Protocol (DIS, HLA, etc)
“Basic Plumbing”:
TCP/IP
“Basic Plumbing”:
TCP/IP
The Network:State information messages sent
Protocols
What if Strykers are generating BFT data with (X, Y, Z) in
lat/lon/altitude, and the dismounted infantry simulator is using a
simulation protocol coordinate system that is local and
rectilinear?
The contents of the protocols are slightly different, in different
order, floats vs. doubles, different semantics, etc.
In our simplified diagram, the protocols are in the second row of
boxes
What happens if an application sends a DIS protocol message to
an HLA application?
What happens if VBS2 sends a DIS message to another DIS
application?
Protocols (another look)
Application/Simulation
Ad-Hoc
Protocols
DIS
HLA
TCP/IP Sockets
Web
Services
Installed Base vs Emerging
From the 90’s until very recently the classic thing to do was run
desktop simulation applications written in compiled C++ or Java
applications. This has meant a lot of conventional programming
and network programming
I think the world is moving towards web-based and cloud-based
applications, which require some different technology
The economics are compelling. See the lack of desktop applications
when compared to web applications. 10 years ago you installed
a desktop mail application; now you use a web-based mail
application
Good integration with mobile devices
Installed Base Technology
Compiled
App
Compiled
App
Compiled
App
Compiled Apps Communicating Over Network
Emerging Architectures
Javascript
Web App
Javascript
Web App
Web Server
Network
Brave New World
The transition makes teaching the class a bit
difficult. Most of the installed base of
simulations use the compiled applications on
workstations model. Much of the future is
oriented towards the web. The technologies
used are very different, and requires some
context switching.
We’ll try to give you some of both, but
remember to keep clear which is which
Background
During this class, we’ll look at each of these
elements:
• TCP/IP sockets
• Ad-hoc protocols
• DIS
• HLA
• Web Technology