Network Group

Some Kosta/Dr. J Q/A Dialogue is available.

In Spring 2004, Laurel and Mike produced some preliminary chat client code: SecondChat.icn

Some networking resources we may use include:

• Robert Parlett's Unicon IPC classes
• Blender.org has a protocol called Verse which might be useful. At the least, its discussion of networking protocols is useful.
• Mike Kemp claims SCTP may be an appropriate protocol for this type of project. Some more links on SCTP are available.
• Bandwidth Monitor NG is a network bandwidth monitor that Jayanthan has had good luck with.

We will have to think about network architecture and design. Probably a good place to start would be comparing what we want with the Alice architecture (other Alice docs). Alice is a system for rapid prototyping of VR's, and it uses a two-process model in which one process is the graphics renderer and the other is the VR simulation.

We need a similar client-server model, in which the VR world is running in some central location, with which clients communicate. One way we could start to achieve this would be to separate out Ray's PLEASE lab demo into a graphical client (most of the code) and a server which stores dynamic information such as avatar movements, drawing on the whiteboard, etc. Without worrying about n-person chat integration yet, can we do a one-person client-server system in which the client GUI Events() generate server requests for state changes, and the server replies with the state changes which cause the graphics to be updated?

We may need, for the sake of interactive performance, to handle avatar motion specially, such as: allowing the view update to be immediate, and batching many movements together into single packets, say, every half-second or second.

Our preliminary designs for the TCP-based client protocol are available.

We expect to add UDP down the road; the UDP protocol would be the same as the TCP above, except it would only be used for transient commands such as \move, and would include a serial # to handle (drop) out of order packets.

We probably want to think about adding Peer-2-Peer client transmission of UDP messages when we can establish that (a) P2P communication is possible, and (b) clients can see each other, i.e. they are in the same room. Akshay reports having written a checker for (a).

CVE Protocols

MCP

The MUD (or MOO) Client Protocol is a widely used standard for textual virtual environments. However, it is not clear how useful it is by itself, it seems to be a meta- protocol without any underlying command set.

VRTP

Virtual Reality Transfer Protocol. An early stage of development. May be related to X3D which is low-level and XML-based.

Kessler and Hodges

Proposed a protocol for networked VE's in 1996. E. Drew Kessler and Larry F. Hodges. A network communication protocol for distributed virtual environment systems. IEEE Virtual Reality Annual International Symposium (VRAIS) '96, March 1996.

Broll's DWTP

appeared in VRML98 Symposium, possibly WET-ICE 97.

ISTP

Interactive Sharing Transport Protocol. Mitsubishi Electronics Research Lab, Cambridge MA. 84 page TR available. See also slides 51-65 of: http://internetng.dit.upm.es/ponencias-jing/2004/Isabel2004_Georganas.pdf

pLVE

Bouras et al's protocol for educational virtual environments. Bouras' group also wrote an article: architectures and protocols for educational virtual environments, citing several protocols: DIS, DWTP, USTP, VSCP, VIP...

VIP

VRML Interchange Protocol, http://www.csclub.uwaterloo.ca/~sfwhite/vnet/VIP.html

VSCP

Virtual Society Client Protocol, for Sony's Virtual Society project. Probably not an open standard.

DIS

Distributed Interactive Protocol, a military standard. Messages oriented to simulations of tanks or planes; not likely to be of use for our purposes.

ZMP

Zenith MUD Protocol proposd for an open source MUD system. Sort of a meta-protocol, or else it is under construction.

Ports