SlideShare a Scribd company logo

Improving Quality in a Distributed IP Telephony System by the use of Multiplexing Techniques

Jose Saldana, profile picture
Jose Saldana

El documento presenta un sistema de telefonía IP distribuido con el objetivo de mejorar la calidad y reducir costos, utilizando soluciones basadas en software y gestión centralizada de recursos. Se discuten técnicas para garantizar la calidad del servicio (QoS) y la probabilidad de admisión en conferencias múltiples entre oficinas. Además, se establecen pruebas en un entorno virtualizado para analizar el rendimiento del sistema y las ventajas de protocolos de multiplexión como el TCRTP.

EducationTechnologyBusiness
1 of 50
Downloaded 28 times
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
SPECTS 2011 Presentación <> June 27-30, 2011, The Hague, Netherlands Jenifer Murillo Jose Saldana Julián Fernández Navajas G RUPO DE T ECNOLOGÍAS DE LAS COMUNICACIONES CPS - University of Zaragoza, Spain José Ruiz Mas José I. Aznar Eduardo Viruete Navarro
Index INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM TESTS AND RESULTS CONCLUSIONS
Index INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM TESTS AND RESULTS CONCLUSIONS
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Introduction - VoIP is replacing traditional telephony systems. - Objective: Cost saving, taking advantage of Internet connections. - Decentralized resources: each office is independent. SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 4
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Introduction - Making a global management of the resources, we can save costs and increase admission probability Zone 2 Country 4 Country 3 IP network Zone 3 PSTN Country 5 Country 2 Country 1 Zone 1 SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 5
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Introduction - Software-based solutions allow a simple PC to assume the role of the PBX, making it possible to - Build a central-managed system - Sharing gateways of different offices Country 1 Gateway Country 2 Gateway PSTN Office 2 Office 1 IP network IP call SPECTS, Jun 27-30 2011. The Hague Local call Improving Quality in a Distributed IP Telephony System 6
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Introduction - VoIP is a real-time service which has to work in a best-effort network. - Users demand a QoS similar to the one they used to have with traditional telephony. - Solutions for adding QoS: - CAC: Call Access Control SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 7
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS R-factor - Defined by ITU G.107 (E-Model) Ranges from 0 (bad quality) to 100 (good) Acceptable for R > 70 Dependence on delay and packet loss Widely accepted quality estimator for VoIP services SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 8
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Introduction - In a business with many offices, each one with many users, a number of conferences can be established between the same pair of offices. SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 9
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Introduction - In a business with many offices, each one with many users, a number of conferences can be established between the same pair of offices. SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 10
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Introduction - Advantage of grouping traffic: - Overhead reduction - Disadvantage: - Added delays IP network MUX/DEMUX RTP SPECTS, Jun 27-30 2011. The Hague MUX/DEMUX TCRTP RTP Improving Quality in a Distributed IP Telephony System 11
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Introduction - We will study the advantages of adding a multiplexing protocol to the central managed IP telephony system. - Admission Probability - R-Factor IP network MUX/DEMUX RTP SPECTS, Jun 27-30 2011. The Hague MUX/DEMUX TCRTP RTP Improving Quality in a Distributed IP Telephony System 12
Index INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM TESTS AND RESULTS CONCLUSIONS
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS CAC systems - Measurement-based: Use the state of the network to take admission decisions - Parameter-based: Some measurements are carried out during system’s set up, and a maximum number of calls is set. SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 14
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Buffer size and buffer policies - «Rule of the thumb»: Bandwidth-delay product. - «Stanford model»: Division by sqrt(N) (N:number of TCP flows). - Other proposal: time-limited buffer. Interesting for this work. Limits OWD. But penalizes big packets. SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 15
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Header compressing techniques - IPHC: Only compresses TCP, UDP and IP headers - CRTP: Also RTP - ECRTP: Enhanced CRTP - ROHCv2: Uses more recent compressing algorithms - TCRTP uses ECRTP SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 16
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS TCRTP samples samples ... ECRTP RTP ECRTP UDP IP PPP Mux PPP L2TP IP CH IP header L2TP PPP Header header MH RH PPP Mux header Reduced header SPECTS, Jun 27-30 2011. The Hague S MH RH S Samples PPP Mux header Reduced header Samples ... Improving Quality in a Distributed IP Telephony System 17
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS TCRTP Bandwidth saving - As bandwidth is saved, parameter-based CAC can support a bigger number of simultaneous conferences. Bandwidth saving X 0.9 0.8 0.8-0.9 0.7 X 0.7-0.8 0.6-0.7 0.6 0.5-0.6 0.4-0.5 0.5 0.3-0.4 0.4 0.3 1 2 3 4 5 6 7 1 8 9 k SPECTS, Jun 27-30 2011. The Hague 10 11 0.9 12 13 14 15 0.8 16 17 18 19 0.7 prob. of reduced header 20 Improving Quality in a Distributed IP Telephony System 18
Index INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM TESTS AND RESULTS CONCLUSIONS
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS IP Telephony system - Offices in different countries Dial plan only at the PBX Internet used for VoIP traffic The system does not use any reservation protocol - VoIP traffic is the only real-time one we are going to take care of in a special way - A parameter-based CAC is used SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 20
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS IP Telephony system - Each office has a gateway and an Internet access Data Center IP network PBX CAC limit Country 1 Office 1 GW Country 3 Country 2 Office 2 GW Office 3 GW Office 4 GW Area 2 Area 1 Office 5 GW Office 6 GW Country 4 Office 7 GW Area 3 Gateway’s lines limit PSTN SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 21
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS IP Telephony system - Each office has a gateway and an Internet access Data Center IP network Signaling (SIP) goes to the PBX CAC limit Country 1 Office 1 GW PBX Country 3 Country 2 Office 2 GW Office 3 GW Office 4 GW Area 2 Area 1 Office 5 GW Office 6 GW Country 4 Office 7 GW Area 3 Gateway’s lines limit PSTN SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 22
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS IP Telephony system - Each office has a gateway and an Internet access Data Center IP network RTP uses a star topology: tunnels between offices Country 1 Office 1 GW PBX CAC limit Country 3 Country 2 Office 2 GW Office 3 GW Office 4 GW Area 2 Area 1 Office 5 GW Office 6 GW Country 4 Office 7 GW Area 3 Gateway’s lines limit PSTN SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 23
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Local Agent (SIP proxy) - All signaling messages pass through it - Counts the number of calls - In charge of admission decissions database IP phone Local agent PBX Local agent IP phone INVITE 100 Trying 183 Session Progress rarily 480 Tempo e Unavailabl Office 1 SPECTS, Jun 27-30 2011. The Hague INVITE 100 Trying 183 Session Progress rarily 480 Tempo e Unavailabl INVITE raril 480 Tempo e Unavailabl y Office i Improving Quality in a Distributed IP Telephony System 24
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Local Agent (SIP proxy) - All signaling messages pass through it - Counts the number of calls - In charge of admission decissions database IP phone Local agent PBX Local agent IP phone INVITE 100 Trying 183 Session Progress rarily 480 Tempo e Unavailabl Office 1 SPECTS, Jun 27-30 2011. The Hague There is no place for this call INVITE 100 Trying 183 Session Progress rarily 480 Tempo e Unavailabl INVITE raril 480 Tempo e Unavailabl y Office i Improving Quality in a Distributed IP Telephony System 25
Index INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM TESTS AND RESULTS CONCLUSIONS
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Scheme of tests Real Traffic in a testbed Offline post-processing Buffer policies VoIP Network delays + Dejitter buffer Background Router Traffic Generation Traffic Capture Traffic Trace QoS parameters Final results Scenario parameters Conferences/ office Simulated scenario Final trace Simulation in Matlab SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 27
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Scheme of tests Emulation in a testbed Real Traffic in a testbed Offline post-processing Buffer policies VoIP Network delays + Dejitter buffer Background Router Traffic Generation Traffic Capture Traffic Trace QoS parameters Final results Scenario parameters Conferences/ office Simulated scenario Final trace Simulation in Matlab SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 28
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Testbed - Xen Virtualization-based testbed - Each computer is translated into a VM - Bandwidth of office’s routers emulated with Linux tc (Traffic Control) - Codec used: G.729a with 2 samples per packet Public IP address xenbr0 Physical machine Virtual network SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 29
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Scheme of tests Addition of delays Real Traffic in a testbed Offline post-processing Buffer policies VoIP Network delays + Dejitter buffer Background Router Traffic Generation Traffic Capture Traffic Trace QoS parameters Final results Scenario parameters Conferences/ office Simulated scenario Final trace Simulation in Matlab SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 30
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS System delays Added offline Tpacketization Tretention Tprocess Tqueue Tnetwork Tprocess IP network . . . MUX SPECTS, Jun 27-30 2011. The Hague Tdejitter . . . DEMUX Improving Quality in a Distributed IP Telephony System 31
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Scheme of tests Real Traffic in a testbed Offline post-processing Buffer policies VoIP Network delays + Dejitter buffer Background Router Traffic Generation Traffic Capture Traffic Trace QoS parameters Final results Scenario parameters Conferences/ office Simulated scenario Final trace Simulation in Matlab Simulation of the whole scenario SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 32
Index INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM TESTS AND RESULTS CONCLUSIONS
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Admission probability - Each office has 25 users Gateways have 6 lines CAC limit=6 (variable in some tests) Different values of l and number of offices ... ... M2 IP network AI21 AP2 N2 ... ... PSTN ... N1 M1 ... AI12 AP1 AO1 Office 1 SPECTS, Jun 27-30 2011. The Hague PSTN AO2 Office 2 Improving Quality in a Distributed IP Telephony System 34
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Admission probability Admission probability 3 offices 5 offices 7 offices 10 offices 15 offices Percentage of admitted conferences (%) 100 95 90 85 80 3 SPECTS, Jun 27-30 2011. The Hague 3,5 4 4,5 λ (conferences per hour per user) 5 Improving Quality in a Distributed IP Telephony System 35
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Admission probability Admission probability 3 offices 5 offices 7 offices 10 offices 15 offices Percentage of admitted conferences (%) 100 95 90 Increasing the number of offices is beneficial 85 80 3 SPECTS, Jun 27-30 2011. The Hague 3,5 4 4,5 λ (conferences per hour per user) 5 Improving Quality in a Distributed IP Telephony System 36
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Admission probability - Influence of CAC limit 2 offices and isolated mode 2 offices and sharing mode 4 offices and isolated mode 4 offices and sharing mode 6 offices and isolated mode 6 offices and sharing mode 100 Admission probability Admission probability (%) 95 90 85 80 1 2 3 SPECTS, Jun 27-30 2011. The Hague 4 5 6 7 CAC limit 8 9 10 11 12 13 Improving Quality in a Distributed IP Telephony System 37
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Admission probability - Influence of CAC limit 2 offices and isolated mode 2 offices and sharing mode 4 offices and isolated mode 4 offices and sharing mode 6 offices and isolated mode 6 offices and sharing mode 100 Admission probability Sharing mode is better than isolated mode Admission probability (%) 95 90 85 80 1 2 3 SPECTS, Jun 27-30 2011. The Hague 4 5 6 7 CAC limit 8 9 10 11 12 13 Improving Quality in a Distributed IP Telephony System 38
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Admission probability - Influence of CAC limit 2 offices and isolated mode 2 offices and sharing mode 4 offices and isolated mode 4 offices and sharing mode 6 offices and isolated mode 6 offices and sharing mode 100 Admission probability Increasing the number of offices is beneficial Admission probability (%) 95 90 85 80 1 2 3 SPECTS, Jun 27-30 2011. The Hague 4 5 6 7 CAC limit 8 9 10 11 12 13 Improving Quality in a Distributed IP Telephony System 39
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Admission probability - Influence of CAC limit 2 offices and isolated mode 2 offices and sharing mode 4 offices and isolated mode 4 offices and sharing mode 6 offices and isolated mode 6 offices and sharing mode 100 Admission probability In Sharing mode, the increase of CAC limit improves the Admission probability Admission probability (%) 95 90 85 80 1 2 3 SPECTS, Jun 27-30 2011. The Hague 4 5 6 7 CAC limit 8 9 10 11 12 13 Improving Quality in a Distributed IP Telephony System 40
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS R-factor - Each office has 25 users - Background traffic: 80% of the capacity R-factor average 80 78 76 74 R-factor 72 70 68 66 64 62 60 0 200 400 600 800 1000 1200 Conferences SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 41
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS R-factor R-factor average 80 78 76 74 Better values for sharing mode R-factor 72 70 68 66 64 62 60 0 200 400 600 800 1000 1200 Conferences SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 42
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS R-factor R-factor average 80 78 76 74 Less variance for sharing mode R-factor 72 70 68 66 64 62 60 0 200 400 600 800 1000 1200 Conferences SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 43
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS R-factor - Original vs Multiplexing as a function of CAC limit R-factor 82 80 Simple RTP, λ = 2 Simple RTP, λ = 4 78 Simple RTP, λ = 6 R-factor 76 Simple RTP, λ = 8 Simple RTP, λ = 10 74 TCRTP, λ = 2 72 TCRTP, λ = 4 TCRTP, λ = 6 70 TCRTP, λ = 8 68 TCRTP, λ = 10 66 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 CAC limit SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 44
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS R-factor - Original vs Multiplexing as a function of CAC limit R-factor 82 80 Simple RTP, λ = 2 Simple RTP, λ = 4 78 Increasing CAC limit makes average quality fall R-factor 76 74 Simple RTP, λ = 6 Simple RTP, λ = 8 Simple RTP, λ = 10 TCRTP, λ = 2 72 TCRTP, λ = 4 TCRTP, λ = 6 70 TCRTP, λ = 8 68 TCRTP, λ = 10 66 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 CAC limit SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 45
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS R-factor - Original vs Multiplexing as a function of CAC limit R-factor 82 80 Simple RTP, λ = 2 Simple RTP, λ = 4 78 Simple RTP, λ = 6 Increase of CAC limit for a desired R-factor R-factor 76 Simple RTP, λ = 8 Simple RTP, λ = 10 74 TCRTP, λ = 2 72 TCRTP, λ = 4 TCRTP, λ = 6 70 TCRTP, λ = 8 68 TCRTP, λ = 10 66 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 CAC limit SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 46
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS R-factor - Tradeoff Quality vs Admission Probability Less quality butR-factoradmission probability more 82 80 Simple RTP, λ = 2 Simple RTP, λ = 4 78 Simple RTP, λ = 6 R-factor 76 Simple RTP, λ = 8 Simple RTP, λ = 10 74 TCRTP, λ = 2 72 TCRTP, λ = 4 TCRTP, λ = 6 70 TCRTP, λ = 8 68 TCRTP, λ = 10 66 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 CAC limit SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 47
Index INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM TESTS AND RESULTS CONCLUSIONS
INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Conclusions - An SIP-based IP telephony system using a CAC has been tested. The scenario mathches an enterprise with several offices. - QoS parameters are obtained in a testbed. - Sharing the gateways increases admission probability. - TCRTP saves bandwidth so R-factor can be improved. - Tradeoff admission probability vs QoS SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 49
SPECTS 2011 Presentación Presentación <> June 27-30, 2011, The Hague, Netherlands Jenifer Murillo Jose Saldana Julián Fernández Navajas G RUPO DE T ECNOLOGÍAS DE LAS COMUNICACIONES CPS - University of Zaragoza, Spain José Ruiz Mas José I. Aznar Eduardo Viruete Navarro

More Related Content

PDF
Evaluation of Multiplexing and Buffer Policies Influence on VoIP Conversation...
Jose Saldana
 
PPT
Ch1
JohnKloty
 
PDF
Mplswc2006 white paper-v1.1
Sean Andersen
 
PPT
Chp. 2 simulation examples
Pravesh Negi
 
PPTX
Heinz-Werner Schuelting - IPV6 Piloting
IPv6 Conference
 
PPTX
ETE405-lec4.pptx
mashiur
 
PDF
10 fn s17
Scott Foster
 
PPT
Efficient VoIP Solution for env. of Technical University of Kosice
Jozef Janitor
 
Evaluation of Multiplexing and Buffer Policies Influence on VoIP Conversation...
Jose Saldana
 
Ch1
JohnKloty
 
Mplswc2006 white paper-v1.1
Sean Andersen
 
Chp. 2 simulation examples
Pravesh Negi
 
Heinz-Werner Schuelting - IPV6 Piloting
IPv6 Conference
 
ETE405-lec4.pptx
mashiur
 
10 fn s17
Scott Foster
 
Efficient VoIP Solution for env. of Technical University of Kosice
Jozef Janitor
 

Similar to Improving Quality in a Distributed IP Telephony System by the use of Multiplexing Techniques (20)

PPT
Fep Master
aomran3
 
PDF
Osc ftth solutions v1.0 20160922
Guisun Han
 
PDF
Technology Update: Future of Broadcast according to EBU 2017
Media Perspectives
 
PPT
Download It
Videoguy
 
PDF
A Model Of An Integrated Unified Communication Network Using Public Switched ...
Becky Gilbert
 
PPTX
Ofelia open calls
FIA2010
 
PPT
TCP Performance analysis Wireless Multihop Networks
Abhishek Kona
 
PPT
B&B channel partner event ethernet intro - shane duffy
Shane Duffy
 
PDF
Extended QA testing to enhance productivity of a leading networking solution ...
Veryx Technologies
 
PDF
PP_PCN
Robert Rivenbark
 
PDF
ETE405-lec4.pdf
mashiur
 
PDF
Profinet implementation and testing - Dave Tomlin and Xing Ye
PROFIBUS and PROFINET InternationaI - PI UK
 
PPTX
China Telecom Americas - General Overview
Brian Trentacost
 
PDF
PI and the basics of profibus and profinet at e+h june 2018 mark freeman
PROFIBUS and PROFINET InternationaI - PI UK
 
PDF
IRJET- Campus-Wide Internet Telephony Design and Simulation using Voice over ...
IRJET Journal
 
PDF
Gaolei cv 2013 til2k
Lei Gao
 
PPTX
1 introduction
Jesusizdo
 
PDF
N017147679
IOSR Journals
 
PDF
Performance Evaluation of Ipv4, Ipv6 Migration Techniques
IOSR Journals
 
PDF
Improving Network Efficiency with Simplemux
Jose Saldana
 
Fep Master
aomran3
 
Osc ftth solutions v1.0 20160922
Guisun Han
 
Technology Update: Future of Broadcast according to EBU 2017
Media Perspectives
 
Download It
Videoguy
 
A Model Of An Integrated Unified Communication Network Using Public Switched ...
Becky Gilbert
 
Ofelia open calls
FIA2010
 
TCP Performance analysis Wireless Multihop Networks
Abhishek Kona
 
B&B channel partner event ethernet intro - shane duffy
Shane Duffy
 
Extended QA testing to enhance productivity of a leading networking solution ...
Veryx Technologies
 
PP_PCN
Robert Rivenbark
 
ETE405-lec4.pdf
mashiur
 
Profinet implementation and testing - Dave Tomlin and Xing Ye
PROFIBUS and PROFINET InternationaI - PI UK
 
China Telecom Americas - General Overview
Brian Trentacost
 
PI and the basics of profibus and profinet at e+h june 2018 mark freeman
PROFIBUS and PROFINET InternationaI - PI UK
 
IRJET- Campus-Wide Internet Telephony Design and Simulation using Voice over ...
IRJET Journal
 
Gaolei cv 2013 til2k
Lei Gao
 
1 introduction
Jesusizdo
 
N017147679
IOSR Journals
 
Performance Evaluation of Ipv4, Ipv6 Migration Techniques
IOSR Journals
 
Improving Network Efficiency with Simplemux
Jose Saldana
 
Ad

More from Jose Saldana (20)

PPTX
Pint of science Patinete as a Service
Jose Saldana
 
PDF
Mejorar tu empleabilidad como ingeniero
Jose Saldana
 
PDF
POUZ Universidad de Zaragoza - Telecomunicación 2º y 3º
Jose Saldana
 
PDF
Cómo se conectan los ordenadores y los móviles
Jose Saldana
 
PDF
La bala que dobló la esquina: el problema de los videojuegos online
Jose Saldana
 
PDF
Entretenimiento online. Una perspectiva cristiana
Jose Saldana
 
PDF
¿Por qué el WiFi se va y se viene?
Jose Saldana
 
PDF
Wi-5: Advanced Features for Low-cost Wi-Fi APs
Jose Saldana
 
PDF
Header compression and multiplexing in LISP
Jose Saldana
 
PDF
Simplemux traffic optimization
Jose Saldana
 
PPTX
Online games: a real-time problem for the network
Jose Saldana
 
PDF
GAIA and Alternative Networks
Jose Saldana
 
PDF
Bar-BoF session about Simplemux at IETF93, Prague
Jose Saldana
 
PDF
Alternative Network Deployments
Jose Saldana
 
PDF
Simplemux: a generic multiplexing protocol
Jose Saldana
 
PDF
Optimization of Low-efficiency Traffic in OpenFlow Software Defined Networks
Jose Saldana
 
PDF
TCM-TF 2014
Jose Saldana
 
PDF
Can We Multiplex ACKs without Harming the Performance of TCP?
Jose Saldana
 
PDF
The Effect of Multiplexing Delay on MMORPG TCP Traffic Flows
Jose Saldana
 
PDF
Online games traffic characterization and network support
Jose Saldana
 
Pint of science Patinete as a Service
Jose Saldana
 
Mejorar tu empleabilidad como ingeniero
Jose Saldana
 
POUZ Universidad de Zaragoza - Telecomunicación 2º y 3º
Jose Saldana
 
Cómo se conectan los ordenadores y los móviles
Jose Saldana
 
La bala que dobló la esquina: el problema de los videojuegos online
Jose Saldana
 
Entretenimiento online. Una perspectiva cristiana
Jose Saldana
 
¿Por qué el WiFi se va y se viene?
Jose Saldana
 
Wi-5: Advanced Features for Low-cost Wi-Fi APs
Jose Saldana
 
Header compression and multiplexing in LISP
Jose Saldana
 
Simplemux traffic optimization
Jose Saldana
 
Online games: a real-time problem for the network
Jose Saldana
 
GAIA and Alternative Networks
Jose Saldana
 
Bar-BoF session about Simplemux at IETF93, Prague
Jose Saldana
 
Alternative Network Deployments
Jose Saldana
 
Simplemux: a generic multiplexing protocol
Jose Saldana
 
Optimization of Low-efficiency Traffic in OpenFlow Software Defined Networks
Jose Saldana
 
TCM-TF 2014
Jose Saldana
 
Can We Multiplex ACKs without Harming the Performance of TCP?
Jose Saldana
 
The Effect of Multiplexing Delay on MMORPG TCP Traffic Flows
Jose Saldana
 
Online games traffic characterization and network support
Jose Saldana
 
Ad

Recently uploaded (20)

PPTX
COMPUTERS AS DATA ANALYSIS IN PRECLINICAL DEVELOPMENT.pptx
NIKITA BHUTE
 
PDF
What Is Coercive Control? Understanding and Recognizing Hidden Abuse
Robert Kaiser
 
PDF
Types of Literary Text: Poetry and Prose
kaelandreabibit
 
PPTX
Introduction to Child Health Nursing – Unit I | Child Health Nursing I | B.Sc...
RAKESH SAJJAN
 
PDF
Module 3: Health Systems Tutorial Slides S2 2025
Jonathan Hallett
 
PPTX
NOI Hackathon - Summer Edition - GreenThumber.pptx
MartinaBurlando1
 
PPTX
Skill Development Program For Physiotherapy Students by SRY.pptx
Prof.Dr.Y.SHANTHOSHRAJA MPT Orthopedic., MSc Microbiology
 
PDF
Phylum Arthropoda: Characteristics and Classification, Entomology Lecture
Miraj Khan
 
PPTX
vedic maths in python:unleasing ancient wisdom with modern code
mistrymuskan14
 
PPTX
Week 4 Term 3 Study Techniques revisited.pptx
mansk2
 
PDF
3.The-Rise-of-the-Marathas.pdfppt/pdf/8th class social science Exploring Soci...
Sandeep Swamy
 
PPTX
How to Manage Global Discount in Odoo 18 POS
Celine George
 
PPTX
UNDER FIVE CLINICS OR WELL BABY CLINICS.pptx
AneetaSharma15
 
PPTX
Congenital Hypothyroidism pptx
AneetaSharma15
 
PPTX
Open Quiz Monsoon Mind Game Final Set.pptx
Sourav Kr Podder
 
PPTX
ACUTE NASOPHARYNGITIS. pptx
AneetaSharma15
 
PPTX
IMMUNIZATION PROGRAMME pptx
AneetaSharma15
 
PDF
Mga Unang Hakbang Tungo Sa Tao by Joe Vibar Nero.pdf
MariellaTBesana
 
PPTX
Odoo 18 Sales_ Managing Quotation Validity
Celine George
 
PDF
2.Reshaping-Indias-Political-Map.ppt/pdf/8th class social science Exploring S...
Sandeep Swamy
 
COMPUTERS AS DATA ANALYSIS IN PRECLINICAL DEVELOPMENT.pptx
NIKITA BHUTE
 
What Is Coercive Control? Understanding and Recognizing Hidden Abuse
Robert Kaiser
 
Types of Literary Text: Poetry and Prose
kaelandreabibit
 
Introduction to Child Health Nursing – Unit I | Child Health Nursing I | B.Sc...
RAKESH SAJJAN
 
Module 3: Health Systems Tutorial Slides S2 2025
Jonathan Hallett
 
NOI Hackathon - Summer Edition - GreenThumber.pptx
MartinaBurlando1
 
Skill Development Program For Physiotherapy Students by SRY.pptx
Prof.Dr.Y.SHANTHOSHRAJA MPT Orthopedic., MSc Microbiology
 
Phylum Arthropoda: Characteristics and Classification, Entomology Lecture
Miraj Khan
 
vedic maths in python:unleasing ancient wisdom with modern code
mistrymuskan14
 
Week 4 Term 3 Study Techniques revisited.pptx
mansk2
 
3.The-Rise-of-the-Marathas.pdfppt/pdf/8th class social science Exploring Soci...
Sandeep Swamy
 
How to Manage Global Discount in Odoo 18 POS
Celine George
 
UNDER FIVE CLINICS OR WELL BABY CLINICS.pptx
AneetaSharma15
 
Congenital Hypothyroidism pptx
AneetaSharma15
 
Open Quiz Monsoon Mind Game Final Set.pptx
Sourav Kr Podder
 
ACUTE NASOPHARYNGITIS. pptx
AneetaSharma15
 
IMMUNIZATION PROGRAMME pptx
AneetaSharma15
 
Mga Unang Hakbang Tungo Sa Tao by Joe Vibar Nero.pdf
MariellaTBesana
 
Odoo 18 Sales_ Managing Quotation Validity
Celine George
 
2.Reshaping-Indias-Political-Map.ppt/pdf/8th class social science Exploring S...
Sandeep Swamy
 

Improving Quality in a Distributed IP Telephony System by the use of Multiplexing Techniques

  • 1. SPECTS 2011 Presentación <> June 27-30, 2011, The Hague, Netherlands Jenifer Murillo Jose Saldana Julián Fernández Navajas G RUPO DE T ECNOLOGÍAS DE LAS COMUNICACIONES CPS - University of Zaragoza, Spain José Ruiz Mas José I. Aznar Eduardo Viruete Navarro
  • 2. Index INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM TESTS AND RESULTS CONCLUSIONS
  • 3. Index INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM TESTS AND RESULTS CONCLUSIONS
  • 4. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Introduction - VoIP is replacing traditional telephony systems. - Objective: Cost saving, taking advantage of Internet connections. - Decentralized resources: each office is independent. SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 4
  • 5. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Introduction - Making a global management of the resources, we can save costs and increase admission probability Zone 2 Country 4 Country 3 IP network Zone 3 PSTN Country 5 Country 2 Country 1 Zone 1 SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 5
  • 6. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Introduction - Software-based solutions allow a simple PC to assume the role of the PBX, making it possible to - Build a central-managed system - Sharing gateways of different offices Country 1 Gateway Country 2 Gateway PSTN Office 2 Office 1 IP network IP call SPECTS, Jun 27-30 2011. The Hague Local call Improving Quality in a Distributed IP Telephony System 6
  • 7. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Introduction - VoIP is a real-time service which has to work in a best-effort network. - Users demand a QoS similar to the one they used to have with traditional telephony. - Solutions for adding QoS: - CAC: Call Access Control SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 7
  • 8. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS R-factor - Defined by ITU G.107 (E-Model) Ranges from 0 (bad quality) to 100 (good) Acceptable for R > 70 Dependence on delay and packet loss Widely accepted quality estimator for VoIP services SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 8
  • 9. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Introduction - In a business with many offices, each one with many users, a number of conferences can be established between the same pair of offices. SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 9
  • 10. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Introduction - In a business with many offices, each one with many users, a number of conferences can be established between the same pair of offices. SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 10
  • 11. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Introduction - Advantage of grouping traffic: - Overhead reduction - Disadvantage: - Added delays IP network MUX/DEMUX RTP SPECTS, Jun 27-30 2011. The Hague MUX/DEMUX TCRTP RTP Improving Quality in a Distributed IP Telephony System 11
  • 12. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Introduction - We will study the advantages of adding a multiplexing protocol to the central managed IP telephony system. - Admission Probability - R-Factor IP network MUX/DEMUX RTP SPECTS, Jun 27-30 2011. The Hague MUX/DEMUX TCRTP RTP Improving Quality in a Distributed IP Telephony System 12
  • 13. Index INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM TESTS AND RESULTS CONCLUSIONS
  • 14. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS CAC systems - Measurement-based: Use the state of the network to take admission decisions - Parameter-based: Some measurements are carried out during system’s set up, and a maximum number of calls is set. SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 14
  • 15. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Buffer size and buffer policies - «Rule of the thumb»: Bandwidth-delay product. - «Stanford model»: Division by sqrt(N) (N:number of TCP flows). - Other proposal: time-limited buffer. Interesting for this work. Limits OWD. But penalizes big packets. SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 15
  • 16. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Header compressing techniques - IPHC: Only compresses TCP, UDP and IP headers - CRTP: Also RTP - ECRTP: Enhanced CRTP - ROHCv2: Uses more recent compressing algorithms - TCRTP uses ECRTP SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 16
  • 17. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS TCRTP samples samples ... ECRTP RTP ECRTP UDP IP PPP Mux PPP L2TP IP CH IP header L2TP PPP Header header MH RH PPP Mux header Reduced header SPECTS, Jun 27-30 2011. The Hague S MH RH S Samples PPP Mux header Reduced header Samples ... Improving Quality in a Distributed IP Telephony System 17
  • 18. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS TCRTP Bandwidth saving - As bandwidth is saved, parameter-based CAC can support a bigger number of simultaneous conferences. Bandwidth saving X 0.9 0.8 0.8-0.9 0.7 X 0.7-0.8 0.6-0.7 0.6 0.5-0.6 0.4-0.5 0.5 0.3-0.4 0.4 0.3 1 2 3 4 5 6 7 1 8 9 k SPECTS, Jun 27-30 2011. The Hague 10 11 0.9 12 13 14 15 0.8 16 17 18 19 0.7 prob. of reduced header 20 Improving Quality in a Distributed IP Telephony System 18
  • 19. Index INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM TESTS AND RESULTS CONCLUSIONS
  • 20. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS IP Telephony system - Offices in different countries Dial plan only at the PBX Internet used for VoIP traffic The system does not use any reservation protocol - VoIP traffic is the only real-time one we are going to take care of in a special way - A parameter-based CAC is used SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 20
  • 21. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS IP Telephony system - Each office has a gateway and an Internet access Data Center IP network PBX CAC limit Country 1 Office 1 GW Country 3 Country 2 Office 2 GW Office 3 GW Office 4 GW Area 2 Area 1 Office 5 GW Office 6 GW Country 4 Office 7 GW Area 3 Gateway’s lines limit PSTN SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 21
  • 22. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS IP Telephony system - Each office has a gateway and an Internet access Data Center IP network Signaling (SIP) goes to the PBX CAC limit Country 1 Office 1 GW PBX Country 3 Country 2 Office 2 GW Office 3 GW Office 4 GW Area 2 Area 1 Office 5 GW Office 6 GW Country 4 Office 7 GW Area 3 Gateway’s lines limit PSTN SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 22
  • 23. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS IP Telephony system - Each office has a gateway and an Internet access Data Center IP network RTP uses a star topology: tunnels between offices Country 1 Office 1 GW PBX CAC limit Country 3 Country 2 Office 2 GW Office 3 GW Office 4 GW Area 2 Area 1 Office 5 GW Office 6 GW Country 4 Office 7 GW Area 3 Gateway’s lines limit PSTN SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 23
  • 24. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Local Agent (SIP proxy) - All signaling messages pass through it - Counts the number of calls - In charge of admission decissions database IP phone Local agent PBX Local agent IP phone INVITE 100 Trying 183 Session Progress rarily 480 Tempo e Unavailabl Office 1 SPECTS, Jun 27-30 2011. The Hague INVITE 100 Trying 183 Session Progress rarily 480 Tempo e Unavailabl INVITE raril 480 Tempo e Unavailabl y Office i Improving Quality in a Distributed IP Telephony System 24
  • 25. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Local Agent (SIP proxy) - All signaling messages pass through it - Counts the number of calls - In charge of admission decissions database IP phone Local agent PBX Local agent IP phone INVITE 100 Trying 183 Session Progress rarily 480 Tempo e Unavailabl Office 1 SPECTS, Jun 27-30 2011. The Hague There is no place for this call INVITE 100 Trying 183 Session Progress rarily 480 Tempo e Unavailabl INVITE raril 480 Tempo e Unavailabl y Office i Improving Quality in a Distributed IP Telephony System 25
  • 26. Index INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM TESTS AND RESULTS CONCLUSIONS
  • 27. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Scheme of tests Real Traffic in a testbed Offline post-processing Buffer policies VoIP Network delays + Dejitter buffer Background Router Traffic Generation Traffic Capture Traffic Trace QoS parameters Final results Scenario parameters Conferences/ office Simulated scenario Final trace Simulation in Matlab SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 27
  • 28. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Scheme of tests Emulation in a testbed Real Traffic in a testbed Offline post-processing Buffer policies VoIP Network delays + Dejitter buffer Background Router Traffic Generation Traffic Capture Traffic Trace QoS parameters Final results Scenario parameters Conferences/ office Simulated scenario Final trace Simulation in Matlab SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 28
  • 29. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Testbed - Xen Virtualization-based testbed - Each computer is translated into a VM - Bandwidth of office’s routers emulated with Linux tc (Traffic Control) - Codec used: G.729a with 2 samples per packet Public IP address xenbr0 Physical machine Virtual network SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 29
  • 30. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Scheme of tests Addition of delays Real Traffic in a testbed Offline post-processing Buffer policies VoIP Network delays + Dejitter buffer Background Router Traffic Generation Traffic Capture Traffic Trace QoS parameters Final results Scenario parameters Conferences/ office Simulated scenario Final trace Simulation in Matlab SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 30
  • 31. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS System delays Added offline Tpacketization Tretention Tprocess Tqueue Tnetwork Tprocess IP network . . . MUX SPECTS, Jun 27-30 2011. The Hague Tdejitter . . . DEMUX Improving Quality in a Distributed IP Telephony System 31
  • 32. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Scheme of tests Real Traffic in a testbed Offline post-processing Buffer policies VoIP Network delays + Dejitter buffer Background Router Traffic Generation Traffic Capture Traffic Trace QoS parameters Final results Scenario parameters Conferences/ office Simulated scenario Final trace Simulation in Matlab Simulation of the whole scenario SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 32
  • 33. Index INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM TESTS AND RESULTS CONCLUSIONS
  • 34. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Admission probability - Each office has 25 users Gateways have 6 lines CAC limit=6 (variable in some tests) Different values of l and number of offices ... ... M2 IP network AI21 AP2 N2 ... ... PSTN ... N1 M1 ... AI12 AP1 AO1 Office 1 SPECTS, Jun 27-30 2011. The Hague PSTN AO2 Office 2 Improving Quality in a Distributed IP Telephony System 34
  • 35. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Admission probability Admission probability 3 offices 5 offices 7 offices 10 offices 15 offices Percentage of admitted conferences (%) 100 95 90 85 80 3 SPECTS, Jun 27-30 2011. The Hague 3,5 4 4,5 λ (conferences per hour per user) 5 Improving Quality in a Distributed IP Telephony System 35
  • 36. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Admission probability Admission probability 3 offices 5 offices 7 offices 10 offices 15 offices Percentage of admitted conferences (%) 100 95 90 Increasing the number of offices is beneficial 85 80 3 SPECTS, Jun 27-30 2011. The Hague 3,5 4 4,5 λ (conferences per hour per user) 5 Improving Quality in a Distributed IP Telephony System 36
  • 37. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Admission probability - Influence of CAC limit 2 offices and isolated mode 2 offices and sharing mode 4 offices and isolated mode 4 offices and sharing mode 6 offices and isolated mode 6 offices and sharing mode 100 Admission probability Admission probability (%) 95 90 85 80 1 2 3 SPECTS, Jun 27-30 2011. The Hague 4 5 6 7 CAC limit 8 9 10 11 12 13 Improving Quality in a Distributed IP Telephony System 37
  • 38. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Admission probability - Influence of CAC limit 2 offices and isolated mode 2 offices and sharing mode 4 offices and isolated mode 4 offices and sharing mode 6 offices and isolated mode 6 offices and sharing mode 100 Admission probability Sharing mode is better than isolated mode Admission probability (%) 95 90 85 80 1 2 3 SPECTS, Jun 27-30 2011. The Hague 4 5 6 7 CAC limit 8 9 10 11 12 13 Improving Quality in a Distributed IP Telephony System 38
  • 39. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Admission probability - Influence of CAC limit 2 offices and isolated mode 2 offices and sharing mode 4 offices and isolated mode 4 offices and sharing mode 6 offices and isolated mode 6 offices and sharing mode 100 Admission probability Increasing the number of offices is beneficial Admission probability (%) 95 90 85 80 1 2 3 SPECTS, Jun 27-30 2011. The Hague 4 5 6 7 CAC limit 8 9 10 11 12 13 Improving Quality in a Distributed IP Telephony System 39
  • 40. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Admission probability - Influence of CAC limit 2 offices and isolated mode 2 offices and sharing mode 4 offices and isolated mode 4 offices and sharing mode 6 offices and isolated mode 6 offices and sharing mode 100 Admission probability In Sharing mode, the increase of CAC limit improves the Admission probability Admission probability (%) 95 90 85 80 1 2 3 SPECTS, Jun 27-30 2011. The Hague 4 5 6 7 CAC limit 8 9 10 11 12 13 Improving Quality in a Distributed IP Telephony System 40
  • 41. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS R-factor - Each office has 25 users - Background traffic: 80% of the capacity R-factor average 80 78 76 74 R-factor 72 70 68 66 64 62 60 0 200 400 600 800 1000 1200 Conferences SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 41
  • 42. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS R-factor R-factor average 80 78 76 74 Better values for sharing mode R-factor 72 70 68 66 64 62 60 0 200 400 600 800 1000 1200 Conferences SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 42
  • 43. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS R-factor R-factor average 80 78 76 74 Less variance for sharing mode R-factor 72 70 68 66 64 62 60 0 200 400 600 800 1000 1200 Conferences SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 43
  • 44. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS R-factor - Original vs Multiplexing as a function of CAC limit R-factor 82 80 Simple RTP, λ = 2 Simple RTP, λ = 4 78 Simple RTP, λ = 6 R-factor 76 Simple RTP, λ = 8 Simple RTP, λ = 10 74 TCRTP, λ = 2 72 TCRTP, λ = 4 TCRTP, λ = 6 70 TCRTP, λ = 8 68 TCRTP, λ = 10 66 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 CAC limit SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 44
  • 45. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS R-factor - Original vs Multiplexing as a function of CAC limit R-factor 82 80 Simple RTP, λ = 2 Simple RTP, λ = 4 78 Increasing CAC limit makes average quality fall R-factor 76 74 Simple RTP, λ = 6 Simple RTP, λ = 8 Simple RTP, λ = 10 TCRTP, λ = 2 72 TCRTP, λ = 4 TCRTP, λ = 6 70 TCRTP, λ = 8 68 TCRTP, λ = 10 66 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 CAC limit SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 45
  • 46. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS R-factor - Original vs Multiplexing as a function of CAC limit R-factor 82 80 Simple RTP, λ = 2 Simple RTP, λ = 4 78 Simple RTP, λ = 6 Increase of CAC limit for a desired R-factor R-factor 76 Simple RTP, λ = 8 Simple RTP, λ = 10 74 TCRTP, λ = 2 72 TCRTP, λ = 4 TCRTP, λ = 6 70 TCRTP, λ = 8 68 TCRTP, λ = 10 66 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 CAC limit SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 46
  • 47. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS R-factor - Tradeoff Quality vs Admission Probability Less quality butR-factoradmission probability more 82 80 Simple RTP, λ = 2 Simple RTP, λ = 4 78 Simple RTP, λ = 6 R-factor 76 Simple RTP, λ = 8 Simple RTP, λ = 10 74 TCRTP, λ = 2 72 TCRTP, λ = 4 TCRTP, λ = 6 70 TCRTP, λ = 8 68 TCRTP, λ = 10 66 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 CAC limit SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 47
  • 48. Index INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM TESTS AND RESULTS CONCLUSIONS
  • 49. INTRODUCTION RELATED WORKS SYSTEM ARCHITECTURE TEST PLATFORM RESULTS CONCLUSIONS Conclusions - An SIP-based IP telephony system using a CAC has been tested. The scenario mathches an enterprise with several offices. - QoS parameters are obtained in a testbed. - Sharing the gateways increases admission probability. - TCRTP saves bandwidth so R-factor can be improved. - Tradeoff admission probability vs QoS SPECTS, Jun 27-30 2011. The Hague Improving Quality in a Distributed IP Telephony System 49
  • 50. SPECTS 2011 Presentación Presentación <> June 27-30, 2011, The Hague, Netherlands Jenifer Murillo Jose Saldana Julián Fernández Navajas G RUPO DE T ECNOLOGÍAS DE LAS COMUNICACIONES CPS - University of Zaragoza, Spain José Ruiz Mas José I. Aznar Eduardo Viruete Navarro