Sergei N. Kozlov, [e-mail]

1
Streaming digital video over wireless link

• PROGRESS project EES.5653

2
Agenda

• General issues with streaming over a wireless
  link
• Streaming MPEG video - link layer approach
• Briefly - transport layer approach
• Future work

3
Video over wireless link

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

wireless
4
Simplified sender/receiver communication
video source
video sink
sender buffer
receiver buffer
wireless interface (sender)
wireless inerface (receiver)
5
Video over wireless link

• Wireless link properties
• High jitter vs high losses
• because of interferences on physical level
• jitter comes up to 0.25s with default settings on
  CISCO 350 WiFi card
• Low troughput (lt 5Mbps for 802.11b)
• can be higher with 802.11a/g, but never as high
  as wire/optical
• Variable throughput
• Typical perceived quality issues
• Artefacts
• Hick-ups

6
Layers we look at
OSI stack
application layer
MPEG en-/decoder
MPEG-packetizer
transport layer
TCP, UDP/RTP
network layer
IP (Internet Protocol)
packet scheduler
802.11 driver
link layer
physical layer
802.11 WLAN
7
Link layer approach
Sender
Application/ encoder
MAC-level retries
video stream
driver/scheduler buffer
IP packets
TCP/IP stack
8
MPEG encoding

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

I BBPBBPBB(I)

• Importance of a frame decreases in following
  order I, P, B

9
Importance of I and P frames

• Missing of I/P frames causes a lot of artefacts
• A complete stream only missing B frames has no
  artefacts
• in worst case you only get then 8.3 fps instead
  of 25

10
Example of an mpeg stream

• 5Mpbs stream (LOTR)
• B-frames make up more than a half of the whole
  bitrate (i.e. gt2.5Mbps)

11
Scheduling of frames (I)

• Packetizing
• Assign prioritites to packets according to frame
  types
• P(B) 1
• P(P) 2
• P(I) 3
• Scheduling
• Packets with higher priorities should get a
  better chance to be sent

Involved layers
12
Scheduling of frames (II)

• Scheduling algorithm

buffered frame
sent frame
incoming frame
Fi
Fb
Fs
if P(Fi) gt P(Fs) and P(Fb) gt P(Fs) Fs Fb, Fb
Fi (discard Fs)
else if P(Fi) gt P(Fb) Fb Fi (discard
Fb) else Fi 0 (discard Fi)
13
Advantages of link layer approach

• We only need to modify the sending part
• it will work with any terminals supporting RTP
  reception
• it can be used for broadcasting
• it is very reactive against fast network
  fluctuations

14
Transport layer approach

• If...
• we dont have direct access to the wireless
  interface
• we dont want to modify it
• we want more reliability, than UDP/RTP does
• or we want to combine link layer and transport
  layer approaches
• Then we would like to look at the tranport layer
• Our main interest Transport Control Protocol
  (TCP)
• Can we make it suitable for real-time?
• Can we make it suitable for wireless?

15
Further work

• Link-layer transport-layer
• creating math. model
• evaluation of the approach
• gathering statistics
• evaluation of perceived quality

16
Questions?