Entertainment Networking II

1
Entertainment Networking - II

• NMP, P2P Collaborative Streaming
• Presentation by Hari Govindharaj. Date
  01/22/2007

2
Paper 1 Enabling Network-Centric Music
Performance in WANs

• By Zefir Kurtisi, Xiaoyuan Gu and Lars Wolf.
• Technische Universitat Braunschweig, Germany.

3
Network Centric Music Performance (NMP)

• Ubiquitous broadband connectivity
• Musicians collaborate and compose music in
  real-time over internet
• NMP system delivers music within 30 millisecond
  delay
• Similar works are Stanford SoundWire Project and
  Conductor Driven Scheme

4
NMP System Setup

• Client captures music
• Server acts as the mixer
• Server returns the mixed packets back to all
  clients for playback

5
NMP Latency

• Total Delay
• Signal Propagation Delay
• Link Capacity related Delay
• Processing Delay at Routers
• Processing Delay at End Systems (NMP Client
  Server)
• Processing Delay at End Systems
• Computational Latencies
• Buffering Latencies

6
Buffering Latencies

• Major buffering latency at the client Sound card
  introduces minimum delay of 4 buffering units
• With sampling rate of 48kHz with 128 samples per
  buffer unit, the packet buffer delay is 2.667msec
• Server buffers to compensate for network jitter
• Total end system delay under ideal network
  conditions is 10.7msec

7
Delay Scenarios
8
Future Work

• Can the 30msec bound be relaxed further?
• Error concealment and correction schemes to deal
  with packet loss during transmission
• Improved user interface and user evaluation for
  commercial launch

9
Paper 2 Anyone can Broadcast Video over the
Internet

• By Reza Rejaie.
• University of Oregon, Eugene, OR.

10
P2P Streaming

• Availability of broadband connections and
  increasing ability of users to capture and view
  multimedia
• IP multicast didnt take off in 1990s
• P2P streaming has potential to provide TV like
  distribution service
• P2P streaming needs no special support from
  network or high bandwidth connections

11
P2P Streaming Key Features

• Scalability of available resources
• Individual peers contribute outgoing bandwidth
• Aggregate bandwidth scales organically with
  number of participating peers
• Ease of deployment
• Groups of participating peers easily deploy the
  system
• For example, applications are run on the desktop
  computers with no special network support

12
P2P Streaming Challenges

• Bandwidth heterogeneity and asymmetry
• Bandwidth of peers limited by access link
  bandwidth
• Dynamics of peer participation (churn)
• Ongoing change to overlay topology can disrupt
  content delivery
• Utilization of available resources
• Structure of overlay can prevent a peer from
  contributing. Example Tree overlay

13
Mesh based P2P Streaming

• Recent success BitTorrent
• Participating peers form randomly connected
  mesh-shaped overlay
• Swarm like content delivery
• Overcomes all three key challenges bandwidth
  heterogeneity, peer dynamics and resource
  utilization
• Good quality stream delivery to large number of
  users

14
Swarm-like Content Delivery

• Multiple neighbors enables a peer to gracefully
  cope with churn
• Multiple Description Coding (MDC) can address
  bandwidth heterogeneity
• Contents are split into segments
• Swarming couples push-content reporting with
  pull-content requesting
• A packet scheduling algorithm determines the
  segment delivery to each peer based on the
  available content and bandwidth from its neighbor

15
Design Issues to Explore

• Effect of overlay properties on performance
• Peer degree
• Number of neighbors
• Directed vs. Undirected pair-wise connections
• Pattern of content delivery
• Coding issues (MDC Codecs are still only research
  prototypes)
• Security
• Incentive to contribute resources

16
Related Work

• www.wwitv.com
• www.sopcast.com
• Based on the popularity of BitTorrent and other
  systems, the paper concludes that mesh based P2P
  streaming is a research topic worth exploring.

17
Paper 3 Collaborative Streaming in Heterogeneous
and Dynamic Scenarios

• By Reza Rejaie.
• University of Oregon, Eugene, OR.

18
Collaborative Media Streaming

• A group of users participate
• Presentations are often streamed from a remote
  media server
• Number of heterogeneous client devices
• Streaming sessions are established and its state
  controlled as a group

19
Individual vs. Collaborative Media Streaming
20
Co-Stream Architecture
21
Collaboration Control

• Defining a common session state
• Signaling session transfer and late joining of a
  streaming session
• Controlling collaborative sessions (conflicting
  requests)
• Supporting heterogeneous user preferences and
  device requirements
• Providing user, service and session
  identification and location

22
Primary Services of Co-Stream Architecture

• Session Transfer
• Session Initiation Protocol (SIP) conferencing
  feature for session sharing
• Session Control
• Association service manages the common session
  state
• Streaming control through Real-time Streaming
  Protocol (RTSP) proxy methods
• Common group management

23
Related Work

• Mobile IP based environments implement session
  transfer features
• Network Integrated Multimedia Middleware (NMM)
  supports media session sharing
• RTP/I media packets carry session state and events

24
Future Work

• Support for mobility
• Service composition

25
Questions

• ?