group 1.7 denos, khalid, chen, zhou, peng 217, 991, 037, 337 , 641

17
Group 1.7 Denos, Khalid, Chen, Zhou, Peng 217, 991, 037, 337 , 641 age from: https://www.tpg.com.au 1

Upload: sharne

Post on 05-Feb-2016

25 views

Category:

Documents


0 download

DESCRIPTION

Group 1.7 Denos, Khalid, Chen, Zhou, Peng 217, 991, 037, 337 , 641. Image from: https:// www.tpg.com.au. Outline. Introduction System architecture System implementation Used cases Conclusion. Internet Protocol Television (IPTV). Voice Service. IP Network. TV Service. Data Service. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Group 1.7 Denos, Khalid, Chen, Zhou, Peng 217, 991, 037, 337 , 641

Group 1.7Denos, Khalid, Chen, Zhou, Peng

217, 991, 037, 337 , 641

Image from: https://www.tpg.com.au 1

Page 2: Group 1.7 Denos, Khalid, Chen, Zhou, Peng 217, 991, 037, 337 , 641

Outline

• Introduction

• System architecture

• System implementation

• Used cases

• Conclusion

2

Page 3: Group 1.7 Denos, Khalid, Chen, Zhou, Peng 217, 991, 037, 337 , 641

Internet Protocol Television (IPTV)

Image from: http://joannekraft.com

IP Network

Data Service

TV Service

Voice Service

3

Page 4: Group 1.7 Denos, Khalid, Chen, Zhou, Peng 217, 991, 037, 337 , 641

IPTV Monitoring

4

STB SNMP Agent

Server

SNMPTrap

Server

STB SNMP Agent

• Collect data• Queue management• Filter and Parsing• Store

• Data Source• Periodical• Triggered by user (Channel Zapping)

Set Top Box

Page 5: Group 1.7 Denos, Khalid, Chen, Zhou, Peng 217, 991, 037, 337 , 641

System Architecture

5

Page 6: Group 1.7 Denos, Khalid, Chen, Zhou, Peng 217, 991, 037, 337 , 641

System Implementation

• Data Source (SNMP Traps)

Network Processing Network levelVideo decoding Application level

• Control using SNMP

6

Page 7: Group 1.7 Denos, Khalid, Chen, Zhou, Peng 217, 991, 037, 337 , 641

System Implementation

• Data Volume– 180 Bytes/msg, 100,000 subscribers

½ Min

4.8Mb/s

288 M msg 48 GB/day !!

7

Page 8: Group 1.7 Denos, Khalid, Chen, Zhou, Peng 217, 991, 037, 337 , 641

System Implementation

• Data Volume– 180 Bytes/msg, 100,000 subscribers

5 Minonly 5%

12 Kbps

1.44 M msg 370 MB/day !!

8

Page 9: Group 1.7 Denos, Khalid, Chen, Zhou, Peng 217, 991, 037, 337 , 641

System Implementation (server side)

9

Page 10: Group 1.7 Denos, Khalid, Chen, Zhou, Peng 217, 991, 037, 337 , 641

System Implementation (server side)

• Data Analysis– Diagram for historical data

– Diagram for real time data

10

Database APP

Queries Data

Page 11: Group 1.7 Denos, Khalid, Chen, Zhou, Peng 217, 991, 037, 337 , 641

Use Cases

• Application-Level IPTV Quality Monitoring

• Integration with Customer Support

• Network Topology Mapping

• Error Localization

• Correlation with Weather Phenomena

11

Page 12: Group 1.7 Denos, Khalid, Chen, Zhou, Peng 217, 991, 037, 337 , 641

App-level IPTV Quality Monitoring

12

Establish a baseline level of application-level metrics and network-related metrics

Detect any significant increase in errors = Experience of low quality

Page 13: Group 1.7 Denos, Khalid, Chen, Zhou, Peng 217, 991, 037, 337 , 641

Integration with Customer Support

Customer Customer Service Field Team

13

Cost long time to describe the problems, and very likely in a wrong way!!

Usual Case

Customer Customer Service

Data collected

Field Team

Advantage:

1.Guide further decisions to mediate the problem

2.Shorten the delay between a decision and its results

After Integration

Page 14: Group 1.7 Denos, Khalid, Chen, Zhou, Peng 217, 991, 037, 337 , 641

Network Topology Mapping

14

Unavailable precise network topology map create network graph using IP addressing hierarchy

It allows visual exploration of network hierarchy and quick identification of problematic nodes by their color.

Page 15: Group 1.7 Denos, Khalid, Chen, Zhou, Peng 217, 991, 037, 337 , 641

Error Localization

15

Heat map of error severityvisualize the percentage of errors well suited for visual analytics allow the patterns to be discovered quickly

Horizontal streaks Long running underperformance of an individual BNG

Vertical streaksA connection between independent BNGsOr a similar usage pattern

Page 16: Group 1.7 Denos, Khalid, Chen, Zhou, Peng 217, 991, 037, 337 , 641

Correlation With Weather Phenomena

16

Natural causes

Lightning strikesCreate a large amount of impulse noise

IPTV systems without FEC are especially susceptible to such disturbance

Highly localized and little can be done

Weather Radar MapExplain away the unavoidable and focus on the preventable

Page 17: Group 1.7 Denos, Khalid, Chen, Zhou, Peng 217, 991, 037, 337 , 641

Conclusion

• Other cases Conveys information about how the subscribers use and interact with the

IPTV system Rate for individual TV shows & Imply undesirable contents

• Future work Automation

The personal TV activity data could in the future be stored without anonymization.

17