Video Streaming Over Wireless: Where TCP is Not Enough

1
Video Streaming Over Wireless Where TCP is Not
Enough

• Xiaoqing Zhu, Jatinder Pal Singh and Bernd Girod

Information Systems Laboratory Stanford University
2
Wireless Home Networking
6 Mbps
24 Mbps
54 Mbps
12 Mbps
3
Outline

• TCP throughput over wireless
• Demo video streaming vs. file transfer
• Comparison with ideal solution
• Whats missing in TCP?

4
Heterogeneity in Wireless Link Speeds
CN
C1
Cl

Channel Time
5
UDP Throughput over Wireless
Simulation in NS2, for 802.11a network

30
Stream 1, alone
25
Stream 2, alone
20
Stream 1, shared
Throughput (Mbps)
15
10
Stream 2, shared
5
0

10
20
30
40
50
Nominal Speed of Second Link (Mbps)
6
TCP Throughput over Wireless
Simulation in NS2, for 802.11a network

20
Stream 1, alone
15
Stream 2, alone
Throughput (Mbps)
Stream 1, shared
10
Stream 2, shared
5
0

10
20
30
40
50
Nominal Speed of Second Link (Mbps)
7
Overhead of TCP ACK
8
Demo Two Nodes
Scenario A
Video Source _at_ 2Mbps
Link Speed 11 Mbps
Throughput 4.4 Mbps
Shared 1.0 Mbps ( 20 channel time)
Shared 1.0 Mbps ( 70 channel time)
File Transfer Source 3.7MB
Link Speed 2 Mbps
Throughput 1.4 Mbps
9
TCP Performance
Video Streaming _at_ 2 Mbps

30 s
Rate
File Transfer _at_ 1.0 Mbps
Time
10
What Could Have Happened
Video Streaming _at_ 2 Mbps

Rate
File Transfer _at_ 0.7 Mbps
42 s
Time
11
Scenario B
Link Speed 2 Mbps
File Transfer Source 3.7MB
Throughput 1.4 Mbps
Shared 1.2 Mbps ( 85 Channel Time)
Shared 1.2 Mbps ( 6 Channel Time)
Link Speed 54 Mbps
Throughput 20 Mbps
Video Source _at_ 3 Mbps
12
TCP Performance
Video Streaming _at_ 3 Mbps

Rate
25 s
File Transfer _at_ 1.2 Mbps
Time
13
What Could Have Happened
Video Streaming _at_ 3 Mbps

Rate
27 s
File Transfer _at_ 1.2 Mbps
Time
14
Whats Missing in TCP?

• Awareness of applications utility function
• For file transfer, aggregate rate matters
• For video streaming, instantaneous rate matters
• Video streams differ in their rate-quality
  tradeoffs

Utility function only needed at the source

• Knowledge of wireless link heterogeneity
• Channel time shared among competing links
• Congestion due to neighboring transmissions
• High rate over a fast link vs. low rate over a
  slow link

End-to-end measurement no longer suffices
Notion of fairness should be revisited
15
Clean Slate Design or Evolution?
16
TCP Throughput over Wireless

• Per-packet fairness at the MAC layer
• Similar end-to-end observations of p, and RTT for
  competing wireless links
• Approximately equal rate, regardless of link speed

packet size
data rate
Heusse et al. 2003
packet loss rate
round trip time
Mahdavi, Floyd 1997 Floyd et al. 2000
17
TCP Throughput over Wireless
Simulation in NS2, for 802.11a network
) ) ) ) )
54Mbps
6 54 Mbps
) ) ) ) )