View
215
Download
1
Tags:
Embed Size (px)
Citation preview
Better-Behaved Multimedia Networking
Keith Barber
Jason Ingalsbe
Joel Thibault
Prof. Mark Claypool (Advisor)
April 19, 2001
Issues with Multimedia Traffic
• Basic Internet protocols inappropriate for multimedia– TCP does not provide smooth transmission– UDP takes up too much bandwidth
• Multimedia protocols must be TCP-friendly
Existing Alternatives
• TCP-Friendly Rate Control (TFRC)
• TCP Emulation At Receivers (TEAR)
• MM-Flow
Re-Engineering MM-Flow
• Location of flow control logic– “True” application and transport layers– Receiver decides whether scale value should
change– Sender converts scale values to transmission
rate
• Number of scale values• Weighted average scale
Evaluation Techniques
• Simulation Scenarios– Standard Bottleneck Layout– Standard Delay Layout– Standard Fragile Layout
Standard Bottleneck Layout
Standard Delay Layout
Standard Fragile Layout
Effects of Re-Engineering MM-Flow
• Responsiveness to Congestion
• Fair Share of Bandwidth
• Relative Smoothness
MM-App-Old vs. TCP
0
0.2
0.4
0.6
0.8
1
1.2
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 Time (sec)
Percent Utilization
TCP MM-App-Old Total
MM-App-Old Average Utilization = 66.0% TCP Average Utilizaton = 32.7%
MM-App-New vs. TCP
0
0.2
0.4
0.6
0.8
1
1.2
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 Time (sec)
Percent Utilization
TCP MM-App-New Total
MM-App-New Average Utilization = 51.8% TCP Average Utilizaton = 44.5%
Further Evaluation of MM-App-New
• Packet Size
• Number of Scale Values
• Delay
• Fragile Flows
• Weighted Scale Values
TCP-Friendly Flows
• Possible definitions:– Flow uses fair share of bandwidth– Flow responds to congestion– Flow transmission rate less than or equal to
TCP flow transmission rate
• Quantitative measurement is desired
TCP-Friendly Equation
• Implications and Assumptions:– Packet Drop Rate– Measurement Interval Size
pR
BT
*
*3/25.1
TCP-Friendly Results (1 sec.)
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
Time (sec)
Ban
dw
idth
(M
bp
s)
Actual Bandwidth
TCP-Friendly Bandwidth
TCP-Friendly Results (3 sec.)
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
0 6 12 18 24 30 36 42 48 54 60 66 72 78 84 90
Time (sec)
Ban
dw
idth
(M
bp
s)
Actual Bandwidth
TCP-Friendly Bandwidth
TCP-Friendliness of MM-App-New
0
0.5
1
1.5
2
2.5
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
Time (sec)
Ba
nd
wid
th (
Mb
ps
)
Actual Bandwidth
TCP-Friendly Bandwidth
Fair Bandwidth
TCP-Friendliness of TFRC
0
0.5
1
1.5
2
2.5
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
Time (sec)
Ba
nd
wid
th (
Mb
ps
)
Actual Bandwidth
TCP-Friendly Bandwidth
Fair Bandwidth
Multi-Protocol Layout
Multi-Protocol Env. (TCP)
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 63 66 69 72 75 78 81 84 87 90 93
Time (sec)
Ba
nd
wid
th (
Mb
ps
)
Actual Bandwidth
TCP-Friendly Bandwidth
Fair Bandwidth
Multi-Protocol Env. (TFRC)
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 63 66 69 72 75 78 81 84 87 90 93
Time (sec)
Ba
nd
wid
th (
Mb
ps
)
Actual Bandwidth
TCP-Friendly Bandwidth
Fair Bandwidth
Multi-Protocol Env. (MM-App-New)
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 63 66 69 72 75 78 81 84 87 90 93
Time (sec)
Ba
nd
wid
th (
Mb
ps
)
Actual Bandwidth
TCP-Friendly Bandwidth
Fair Bandwidth
Conclusion and Future Work
• MM-Flow is an improvement over original
• Areas for future research:– Increase transmission rate quickly at startup– Decrease fluctuation in transmission rate– Continue examination of TCP-Friendliness
Questions?
http://www.wpi.edu/~jmi725/mqp
http://perform.wpi.edu