Sergei N. Kozlov, [e-mail]

1
Streaming MPEG video over wireless link

• QoS in Digital In-home Networks
• PROGRESS project EES.5653

2
Agenda

• I-frame delay (summary on the approach)
• Progress since May 2004
• Future plans
• Demo

3
Video over wireless link

• Typical scenario video transmission from a
  set-top box to the (mutiple) screens

Between 5 and 13 seconds a microwave oven is on
?t50ms

• Wireless link
• Low and highly variable throughput

4
Sender/receiver communication (RTP-based)
video source
video sink
Packets get lost here
sender buffer
receiver buffer
MAC retransmission mechanism
wireless interface (sender)
wireless inerface (receiver)
5
MPEG encoding

• GOP (group-of-pictures)
• Frame types I, P, B
• Typical GOP structure and dependences

I BBPBBPBB(I)
6
Importance of I and P frames

• Missing of I/P frames causes video artefacts

• A complete stream only missing B frames has
• no artefacts

In worst case you only get 8.3fps from originally
25fps
7
Cumulative weight of B-frames

• An example of a 5Mbps stream (LOTR)
• B-frames make up more than 50 of the whole
  bitrate

8
Link layer approach
Sender
MAC-retransmissions
Application/ encoder
Selectively drop frames here
video stream
IP packets
OS network stack
scheduler buffer
9
I-frame delay (IFD)
Stream generated by application
Transmitting under low and variable throughput
some frames take longer time
Frames displayed at receiver NN 1, 4, 5, 8 are
skipped
10
Pros and cons of link layer approach

• We only need to modify sending part
• it will work with any terminals supporting RTP
  reception and equiped with a general MPEG decoder
• due to RTP (UDP-based), it can be used for
  broadcasting
• It is very reactive against fast network
  fluctuations

• Requires access to wireless interface
• should be implemented at every sending device

11
Deployment of IFD

• ASL, SC (Ewout Brandsma, Eric Persoon)
• CE Connected Planet project (Tom Suters,
  Daniel Meirsman)
• SLx00 products (Streamium)
• CES at Las Vegas in June 2005 (with CE)
• CRE in June 2005 (with ASL)

12
Progress since May 2004

• A publication submitted to WoWMoM 2005
• A demo set-up created
• a number of successful demonstrations given
• Deployment of IFD into real projects started (SC
  CE)
• Future work made concrete (coming slides)
• Yesterday IFD proved to work with a Linksys AP
• IFD and wired wireless network
• streaming to wireless CE devices (such as
  HotMan-2)

new
13
Further work (I) Evaluation of IFD models,
simulations and optimizations

• Simulator (based on ns2)
• variations of network topologies
• variations of algorithms/buffer sizes for IFD
• variations of video stream (GOP pattern
  variations, bit-rates, coding standards etc)

Optimal settings are the simulation goal
14
Future work (II) IFD above transport layer

• If...
• we dont have direct access to the wireless
  interface
• we dont want to modify it
• we want to use it with both wired/wireless
  networks
• we want it over a reliable transport protocol
• ... Then we would like to look at the tranport
  layer
• implement the same idea in the application layer
  above TCP (master-student working on that)

15
Future work (III) IFD SNR scalability
(together with Dmitry)

• The goal is to handle even higher variations in
  the link bandwidth
• KISS project at NatLab is implementing it

16
Welcome to the demo!(given after the meeting)

• Questions?