Transcript Dolby VoIP? How close are we to better than PSTN Sound?

Analyze
Assure
Standard IP Network Model for
Comparing Voice Quality of IP
Telephony Devices
Dolby VoIP? How close are we to better than PSTN Sound?
Jack Douglass, Spirent
Chair TIA TR30.3
Accelerate
How close are we to better than
PSTN Sound?
• An objective method of measuring and
comparing the Audio Quality and Performance
of IP Telephony Equipment in the presence of
Network Impairments is needed
– A Standard Network Model for Evaluating
Multimedia Transmission Performance Over Internet
Protocol
– Audio Quality must be measured over time-varying
network impairments that are statistically accurate
– TIA TR30.3 is working with the ITU-T on a
Statistically Based IP Network Model
• Will become TIA-921when released
• Submit to ITU-T for consideration as a Recommendation
Converged IP Telephony Network is Very
Complex and has Many Impairments
Signaling Path
Network Topologies that need be considered when
testing Audio Quality over a Converged Network
Converged Network Reference Model Diagram
A
TE
D
Telco Switch
L
Gateway
IP Network
L
A
D
Telco Switch
Gateway
R,G,S*
Gateway
L
L
R,G,S*
R,G,S*
Gateway
TE
Parameters & Impairments that Need to be
Considered when Simulating an IP Network
•
•
•
•
•
•
•
•
•
•
•
Network Architecture
Types of Access Links
QoS controlled Edge Routing
Route Flapping
Link Failure
Load Sharing
Time Drift
MTU Size
Packet Loss
Background Traffic (Congestion, Bandwidth, Utilization,
Network Load)
One Way Delay
Sources of IP Network Impairments
IP Network Model
Source
Device A
LAN A
Local Access
B
Local Access
A
64 kbit/s
*128 kbit/s
256 kbit/s
1000BaseX
*384 kbit/s
* 100BaseT Switch
512 kbit/s
100BaseT Hub
*768 kbit/s
10BaseT
*T1 (1.536 kbit/s)
* WLAN (~4 Mbit/s) E1 (1.920 kbit/s)
---------------------E3 (34 Mbit/s)
Occupancy level
*T3 (44 Mbit/s)
Packet loss
ADSL (~256 kbit/s)
*Cable (~256 kbit/s)
Fiber (1-10 Gbit/s)
-------------------Occupancy level
QoS edge router
Core IP Network
Route flapping
One-way delay
Jitter
Packet loss
64 kbit/s
*128 kbit/s
256 kbit/s
*384 kbit/s
512 kbit/s
*768 kbit/s
*T1 (1.536 kbit/s)
E1 (1.920 kbit/s)
E3 (34 Mbit/s)
*T3 (44 Mbit/s)
ADSL (~2 Mbit/s)
*Cable (~3 Mbit/s)
Fiber (1-10 Gbit/s)
-------------------Occupancy level
QoS edge router
* Case used in impairment tables
LAN B
1000BaseX
* 100BaseT Switch
100BaseT Hub
10BaseT
* WLAN (~4 Mbit/s)
---------------------Occupancy level
Packet loss
Destination
Device B
Packet Delay
Example of Test Profile with Time
Variable Network Impairments
Time
Time Varying Statistically Based IP
Network Impairment Conditions (ICs)
IC1
LOO X%
IC2
LOO X%
IC100
LOO X%
Time
• Each Impairment Condition is assigned a Likely-hood of
Occurrence (LOO) based on real IP Network Statistics, Network
Architecture, Classes of Service
• The goal is to have approximately 100 test combinations so that
an automated run of the test suite completes in less than a day.
Test Profiles Based on QoS Classes
Test Profiles
QoS Class
(Y.1541)
Applications (Examples)
Node Mechanisms
A
(VoIP, MoIP,
FoIP, ToIP)
0
Real-Time, loss sensitive, Jitter
sensitive, high interaction (VoIP, VTC)
B
(VoIP, MoIP,
FoIP, ToIP)
1
Real-Time, Jitter sensitive, interactive
(VoIP, VTC).
C
(FoIP only)
2
Transaction Data, Highly Interactive
(Signaling)
3
Transaction Data, Interactive
4
Low Loss Only (Short Transactions,
Bulk Data, Video Streaming)
Long Queue, Drop priority
Any route/path
5
Traditional Applications of Default IP
Networks
Separate Queue (lowest priority)
Any route/path
Strict QoS. Guaranteed no over
subscription on links.
Separate Queue with preferential
servicing, Traffic grooming
Network Techniques
Constrained Routing and Distance
Less constrained Routing and
Distances
Constrained Routing and Distance
Separate Queue, Drop priority
Less constrained Routing and
Distances
• Statistically based models can be created for different
QoS Classes
PESQ, PSQM, MOS, R-Factor, Throughput, Connect Rate, etc.
Example of Network Model Coverage
(NMC) Curve
Parameter X Vs Network Model Coverage Percentage
Device A
Device B
Device C
0
0
10
20
30
40
50
60
70
Percentage of Network Coverage
80
90
100
Examples of Communication Equipment that can
be tested over the Converged Network Model
• Plain Old Telephone Service (POTS) and IP telephones
• PSTN Video H320 and H324
• IP Network Devices such as User Agents, Call Agents,
•
•
•
•
Media Servers, Media Gateway Controllers,
Gatekeepers, Application Servers, Edge Routers,
Gateways, IP Phones, IAF (Internet Aware Fax)
Voice-over-IP (VoIP) gateways
T.38 facsimile devices and gateways
V.150.1 and voiceband data (VBD) modem-over-IP
gateways
TIA-1001 (and V.toip) textphone-over-IP gateways
TR30.3 Test Standards
• TIA/EIA 496A-1989: Interface Between Data Circuit Terminating
Equipment (DCE) and the Public Switched Telephone Network
– Included Network Model for Evaluating Modem Performance
• TIA/EIA TSB 37A-1994: Telephone Network Transmission Model
for Evaluating Analog Modem Performance, which became ITU-T
Recommendation V.56bis-1995
• EIA/TIA TSB 38-1994: Test Procedures for Evaluation of 2-Wire 4
Kilohertz Voice Band Duplex Modems, which became ITU-T
Recommendation V.56ter-1996
• ANSI/TIA/EIA 3700-1999: Telephone Network Transmission
Model for Evaluating Analog Modem Performance
• ANSI/TIA/EIA 793 -2000: North American Telephone Network
Transmission Model for Evaluating Analog Client and Digitally
Connected Server Modems
• ANSI/TIA 876 – 2002: North American Network Access
Transmission Model for Evaluating xDSL Modem Performance
Value of Converged Network Model
• Predicts product performance under statistically based
network conditions
•
•
•
•
•
Finds design weaknesses
Find compatibility issues between network equipment
Facilitates isolating and resolving field problems
Assists in evaluating different technologies
Competition for better performance over the Network
Model drives the industry to create better quality
products
Target Audience for Converged Network Model
•
•
•
•
•
•
Operating Companies
Service Providers
Manufacturers
Design Engineers
Test houses
Magazines and product reviewers
How will TIA-921 help the Industry
Achieve Better than PSTN sound?
• PSTN quality sound on the IP Network has not yet
been achieved
• The Network Model will provide a standard test bed
– objectively compare the Audio Quality of various IP Telephony
devices
– Competition to perform better on the Network Model will help
drive the Quality of IP Telephony sound to be better that PSTN
sound