CING-YU CHU
2012.08.06
INFOCOM 2012
Outline
Introduction
 Measurement
 Measurement Results
 Modeling Skype Behaviors
 Analysis on TCP-friendly

Motivation
Skype VoIP service is well studied while
video service is not
 Skype video service consumes more
bandwidth

 Up to 950 kbps

Imperative for network providers and
network researchers
Key Questions
Q1: How does a Skype video call adapt
its sending rate, video rate and quality
under different network conditions?
 Q2: Are Skype video calls friendly to
TCP flows when they compete for
network resources?

Methodology

Measurement
 Black-box approach
 Different network setting with
○ configurable packet loss,
○ propagation delay
○ available bandwidth
 Aim to measure
○ sending rate
○ throughput
○ RTT
○ video bit rate
○ frame rate
Methodology

Modeling
 Rate control model
 FEC model
 Video quality model

Analysis
 User back-off
○ User-level rate control scheme
 TCP-friendliness
Contribution

Measures Skype’s stationary behaviors
of video calls
 Sending rate is insensitive to packet loss
when PLR < 10%
 Utilization of the available bandwidth is
around 80%
 Overly aggressive FEC scheme, 4.5 times
the PLR
Contribution

Derive various models to verify
 User back-offs react fast to congestion
 Skype video calls are TCP-friendly
○ Due to quality-driven user back-offs
Measurement Test-bed
Experiment Design

TV news video sequence “Akiyo”
 From Joint Video Team (JVT)
 Head and shoulder movements
Virtual video camera tool
 Data collection

 TCP-dump
 Skype technical reports
Skype Video Call

On2 video codec
 Video quantization step
 Video resolution
 Number of frames per seconds (FPS)
Measurements Results

Impact of
 Packet Loss
 Available Bandwidth
 Propagation Delay
Impact of Packet Loss
PLR varies from 0% to 12%
 Propagation delay: 50ms
 3 available bandwidth settings

 250 kbps
 750 kbps
 1000 kbps
Impact of Packet Loss
Forward Error Correction
 Two states

 PLR < 10% => NORM state
 PLR >= 10% => CONS state

Different from TCP congestion control
scheme
Impact of Available bandwidth
Available Bandwidth varies from 50 kbps
to 1000kbps
 Two PLRs: 2% and 10%
 Propagation delay: 50ms

Impact of Propagation Delay
Propagation delay varies from 50ms to
2000ms
 Available bandwidth: 500 kbps
 PLR: 0%

Modeling
Sending rate
 Video rate
 Video Quality

Sending Rate Model

NORM and CONS states

25 scenarios with
 PLR: 0% to 12%
 Available bandwidth: 50 kbps to 1000 kbps

γ = 0.77, μ = -10.8 and δ = 21
Video Rate Model

FEC ratio

FEC ratio model
 Ψ = 0.15 and ω = 4.5
Video Quality Model

ITU-T Recommendation G.1070
video rate
frame rate
Video Quality Model

Frame rate model
Video Quality Model

a = 1.431, b = 0.02228, c = 3.759, d =
184.1, e = 1.161, h = 1.446 and g =
0.03881
Model Validation

Co-current UDP traffic (from iPerf)
 0 kbps to 600 kbps
Link capacity: 700 kbps
 Propagation delay: 50ms
 Available bandwidth for Skype

Model Validation
Pearson Correlation Coefficient
Sending rate
Video rate
Frame rate
0.9898
0.9831
0.9545
Analysis
Q1: How Skype video call users respond
to quality degradation resulted from
network impairments? How effective
user back-offs are as a user-level rate
control scheme?
 Q2: What is the performance of a Skype
video call when it competes with other
Skype calls and TCP flows? Is Skype
video call TCP-friendly?

Network Model

LTE wireless network
 Multiple TCP and Skype users
 M/M/1/K queue using drop-tail
 Downlink: 100 Mbps, uplink: 50 Mbps
TCP Model

Reacts to packet loss and RTT
 p = pq + pc
 t = tq + tc
User-level Rate Control

Video drop-off probability

Number of active users

Effective traffic generated by all users
User-level Rate Control

Average traffic of each user
 Expected Skype sending rate
Responsiveness to Loss
Responsiveness to Delay
Competition with TCP
pc = 2% and tc = 50ms
 NT = # of TCP users, NS = # of Skype
users
 Aggregate traffic


For each Skype user
Competition with TCP

Scale-up factor: k
Conclusion

Measures Skype video traffic
 Shows that Skype is robust against mild
packet loss and propagation delay
 Skype can efficiently utilize available
bandwidth
Models Skype video behaviors
 Shows that Skype video is indeed TCPfriendly

 Based on user back-off rate control scheme
Q&A