A Protocol For Reliable Decentralized Conferencing: Oct. 8. 2005 Sungmin Park

  • Uploaded by: Nicole Mercado
  • 0
  • 0
  • July 2020
  • PDF

This document was uploaded by user and they confirmed that they have the permission to share it. If you are author or own the copyright of this book, please report to us by using this DMCA report form. Report DMCA


Overview

Download & View A Protocol For Reliable Decentralized Conferencing: Oct. 8. 2005 Sungmin Park as PDF for free.

More details

  • Words: 464
  • Pages: 11
A Protocol for Reliable Decentralized Conferencing

Oct. 8. 2005 Sungmin Park [email protected] DPNM Lab., Dept. of CSE, POSTECH - 1/11-

DP&NM Lab. CSE, POSTECH

Goal  Multicast – A decentralized network that is useful to handle multiple parties in a conference with rather heavy traffic such as multimedia communication.

 Mixing – Reduces the network traffic amount, but can be a burden to the nodes in charge

- 2/11-

DP&NM Lab. CSE, POSTECH

Purpose  Reduce of network bandwidth – Due to the enormous increase of network traffic, multicast function is required for the application such as internet broadcasting, network game, in which data should be simultaneously delivered to the participants

 Decentralized network – Withstand to the unexpected failure – Reduce the processing burden on the server

 New approach to p2p networks – New approach needed to the network in which every participants is a server and a client

- 3/11-

DP&NM Lab. CSE, POSTECH

Existing Conferencing Models  Existing Conferencing : Multicast – One or more multicast addresses are allocated to the conference – Each participant joins the multicast groups – Sends their media to the groups

- 4/11-

DP&NM Lab. CSE, POSTECH

Existing Conferencing Models  Existing Conferencing: Mixing – This model is probably the most common way of doing SIP conferencing

Conferencing: end system mixing

Conferencing: conference server mixing - 5/11-

DP&NM Lab. CSE, POSTECH

Proposed Model  Data Mixing – Reduces traffic of sound data

 Topology Management – Minimum-spanning-tree problem • Minimized total cost (latency) • Constraint of Maximum outbound degree in each node – Adapt to the dynamic change in the network • Join • leaving

- 6/11-

DP&NM Lab. CSE, POSTECH

Proposed Model  Minimized Spanning Tree with Mixing

A

B ∀i ∈ neighbors outbound stream for i

A+B+D

= ∑inbound stream from j ( j ≠ i ) j

+ live stream

C

- 7/11-

D DP&NM Lab. CSE, POSTECH

Conference Server  Minimum Spanning Tree with degree bound – Every nodes is a server and a client as well – The spanning tree is directed

– Each node has degree bound 1

100

1 1

1

A 40

100 C

40(?)

B

– The tree should be balanced 1 1 100

- 8/11-

DP&NM Lab. CSE, POSTECH

Node  Node – Each node has following information • Route to the neighbors • Synchronized state with conference server

 Operation

Mixer

Mixer





A

B

A

C - 9/11-

B

C DP&NM Lab. CSE, POSTECH

Schedule – – – – – – –

information gathering ~10.09 requirements analysis ~10.14 high-level design ~10.14 detailed design ~11.05 prototype implementation 10.22~11.12 testing and debugging 10.30~12.04 demo and submitting final report ~12.08

- 10/11-

DP&NM Lab. CSE, POSTECH

Question and Discussion

- 11/11-

DP&NM Lab. CSE, POSTECH

Related Documents


More Documents from "Gaurav Kumar"