computing department a utility-based qos model for emerging multimedia applications mu mu, andreas...
TRANSCRIPT
Computing Department
A Utility-based QoS Model for Emerging Multimedia Applications
Mu Mu, Andreas MautheComputing Department, Lancaster University
Lancaster, UK
Francisco GarciaAgilent Laboratories, UK
21 April 2023
International Workshop on Future Multimedia Networking - FMN’08
Computing Department
2
Challenges of QoS on Emerging Multimedia Application
Utility-based QoS Model
Simulation
Conclusion and Future Work
Agenda
Computing Department
3
Challenges of QoS on Emerging Multimedia Application
Utility-based QoS Model
Simulation
Conclusion and Future Work
Agenda
Computing Department
Multimedia Applications:
Emerging Multimedia Application
Requirements on QoS model:
• QoS on Application/System/User Level (QoE)
• Real-time Diagnosis
4
Application 1
Flow 1 Flow 2 Flow N
Application N
End System
QoS Requirements
QoS Requirements
QoS Requirements
System
Computing Department
5
Challenges of QoS on Emerging Multimedia Application
Utility-based QoS Model
Simulation
Conclusion and Future Work
Agenda
Computing Department
NetworkSender Receiver
Direct Impairments
Delay, Jitter, LossIndirect Impairments
Encoding Loss
Quality
Evaluation
Quality
Evaluation
DelayDelay
JitterJitter
LossLoss
BandwidthBandwidth
QoEQoE
Utility-based QoS Model
6
Computing Department
Constructing Utility Functions
System Utility Functions
Application Utility Functions
Flow Utility Functions
Impairment Utility Functions
From bottom to top
Application 1
Flow 1
Application N
End System
D J L B
Flow 2
D J L B
Flow N
D J L B
7
System
Computing Department
Constructing Utility FunctionsImpairments Utility Functions
8
Empirical FunctionsEmpirical Functions
Experiment Results
Experiment Results
Application 1
Flow 1
Application N
End System
D J L B
Flow 2
D J L B
Flow N
D J L B
Computing Department
Constructing Utility Functions
System Utility Functions
Application Utility Functions
Flow Utility Functions
Impairment Utility Functions
From bottom to top
Application 1
Flow 1
Application N
End System
D J L B
Flow 2
D J L B
Flow N
D J L B
9
System
Computing Department
Constructing Utility FunctionsFlow Utility Functions
Weighted Sum:
Minimum:
User Preference, Flow CharacteristicUser Preference, Flow Characteristic
Product:
10
Application 1
Flow 1
Application N
End System
D J L B
Flow 2
D J L B
Flow N
D J L B
Computing Department
Constructing Utility Functions
System Utility Functions
Application Utility Functions
Flow Utility Functions
Impairment Utility Functions
From bottom to top
Application 1
Flow 1
Application N
End System
D J L B
Flow 2
D J L B
Flow N
D J L B
11
System
Computing Department
Constructing Utility Functions Application Utility Functions
User Preference, Application CharacteristicUser Preference, Application Characteristic
The advantage factor “A”: additional human perception factorThe advantage factor “A”: additional human perception factor
12
Application 1
Flow 1
Application N
End System
D J L B
Flow 2
D J L B
Flow N
D J L B
Computing Department
Constructing Utility Functions
System Utility Functions
Application Utility Functions
Flow Utility Functions
Impairment Utility Functions
From bottom to top
Application 1
Flow 1
Application N
End System
D J L B
Flow 2
D J L B
Flow N
D J L B
13
System
Computing Department
Constructing Utility Functions System Utility Function
System utility function is introduced to evaluate the efficiency of resource utilization by user experience.
14
Application 1
Flow 1
Application N
End System
D J L B
Flow 2
D J L B
Flow N
D J L B
Computing Department
QoS Control SingalingControl and
ManagementSummarized Flow Specification and Evaluation Results
Oth
er C
MS
Constructing Utility Functions
15
Application 1
Flow 1
Application N
End System
D J L B
Flow 2
D J L B
Flow N
D J L B
Monitoring and Evaluation
Control and Management
Application 1
Flow 1
Application N
End System
D J L B
Flow 2
D J L B
Flow N
D J L B
System Utility Functions
Application Utility Functions
Flow Utility Functions
Impairment Utility Functions
Computing Department
16
Challenges of QoS on Emerging Multimedia Application
Utility-based QoS Model
Simulation
Conclusion and Future Work
Agenda
Computing Department
Simulation
• Goals:
Demonstrate the behavior of the single-end monitoring and evaluation system.Demonstrate the way utility-based method can be used to support QoS management.
• Simulation Tools:
Network Simulator 2
• Traffic Generator
17
VoIPPareto On/Off.
FTPOne way FTP agent.
IPTV“Traffic Trace”
Jurassic Park I H.263 VBR http://www.tkn.tu-berlin.de/research/trace/trace.html
Online GamingCBR stream.
Computing Department
SimulationScenario
182 50 100200 300 500 700 800
Home 2 VoIP
Home 1 VoIP
Home 1 Online Gaming
Home 2 Online Gaming
Home 1 IPTV
Home 2 IPTV
Home 1 FTP
NS2 Simulation Time (s)
Backbone
2 5
TCP
FTP
UDP
VoIP
4
UDP
Game
3
UDP
IPTV
Home 1 1
a0
5
TCP
FTP
4
UDP
Game
3
UDP
IPTV
ISP 1
b1
1
2
UDP
VoIP
Home 4
1
b0
a1
100G 5ms Queue 40
1000G 30ms Queue 20
2.5Mb 15ms Queue 10
6Mb 5ms Queue 20
1.5Mb 35ms Queue 15
2
UDP
VoIP
4
UDP
Game
3
UDP
IPTV
Home 2
1
100M 5ms Queue 20
25M 10ms Queue 10
Computing Department
Simulation
Backbone
2 5
TCP
FTP
UDP
VoIP
4
UDP
Game
3
UDP
IPTV
Home 1 1
a0
5
TCP
FTP
4
UDP
Game
3
UDP
IPTV
ISP 1
b1
1
2
UDP
VoIP
Home 4
1
b0
a1
100G 5ms Queue 40
1000G 30ms Queue 20
2.5Mb 15ms Queue 10
6Mb 5ms Queue 20
1.5Mb 35ms Queue 15
2
UDP
VoIP
4
UDP
Game
3
UDP
IPTV
Home 2
1
100M 5ms Queue 20
25M 10ms Queue 10
19
Impairment Value Utility Value
Application Utility
Delay
Jitter
Loss
Computing Department Control and Management
Backbone
2 5
TCP
FTP
UDP
VoIP
4
UDP
Game
3
UDP
IPTV
Home 1 1
a0
5
TCP
FTP
4
UDP
Game
3
UDP
IPTV
ISP 1
b1
1
2
UDP
VoIP
Home 4
1
b0
a1
100G 5ms Queue 40
1000G 30ms Queue 20
2.5Mb 15ms Queue 10
6Mb 5ms Queue 20
1.5Mb 35ms Queue 15
2
UDP
VoIP
4
UDP
Game
3
UDP
IPTV
Home 2
1
100M 5ms Queue 20
25M 10ms Queue 10
20
Monitoring and Evaluation
Simulation
Computing Department
21
Challenges of QoS on Emerging Multimedia Application
Utility-based QoS Model
Simulation
Conclusion and Future Work
Agenda
Computing Department
Overview and Future Works
• Things that we’ve achieved:
1. Analysis of emerging applications on IP Network
2. A Design of a QoS Model
3. Modeling user experience on hierarchical application with utility functions
4. Outlook of utility-based QoS management
• Things for future study
1. Utility-based QoS management
2. Improve application QoS model
3. Integrate NG QoS model into existing/future network architecture
22
Computing Department23
Computing Department
24
Framework for the Integrated Video Quality Assessment
Quality Estimator
Network Analyzer
Packet Inspector
Artifacts Measurement
Artifacts Prediction
PacketDecoderVideo
QualityScore
qual
ity o
f del
iver
y
artif
acts
mea
sure
men
t
Use Case:If (PLR<=threshold_1) { QE simply announces the quality as “perfect” so no more assessment actions are required.}elseif (threshold_1<PLR<threshold_2){• QE can use PLR in relation to video entropy, content characteristics, encoding schemes and other additional information for extended assessment.• QE uses artifact prediction or artifact measurement results for more precisely prediction of video quality.}else {QE Instantly determine the quality as “out-of-services” or “bad” and notify the service provider.}
Computing Department
Simulation
21.04.23 25
Computing Department
SimulationTraffic Generator
• VoIP• We choose Pareto On/Off object to
simulate traffic of VoIP application.
• packetSize_ 160burst_time_ 500msidle_time_ 50msrate_ 68kbshape_ 1.5
• FTP• We choose one way agent because in this
simulation we only analyze the traffics toward Home 1 and Home 2.
• Agent/TCP set window_ 20Agent/TCP set windowInit_ 2Agent/TCP set packetSize_ 1000Agent/TCPSink set packetSize_ 40
IPTVNS-2 offers “Traffic Trace” as a type of application. Traffic Trace reads the binary trace file and generate packets as defined in it. set tfile [new Tracefile]$tfile filename example-traceset t1 [new Application/Traffic/Trace]$t1 attach-tracefile $tfile
_Jurassic Park I H.263 VBR
http://www.tkn.tu-berlin.de/research/trace/trace.html
Online GamingCBR stream with sending rate of 15kbit/s and packet size of 80 Bytes to simulate game traffic.
21.04.23 26
Computing Department
Challenges of Traditional QoS Model
Overprovisioning + always an solution - bandwidth != service assurance
Integrated Services+ end-to-end - too much state information - RSVP
Differentiated Services+ easy to implement - number of class for ToS - cross ISP
27