Voice quality testing is critical to the advancement of the Internet
telephony industry to ensure that such elements as delay, clarity, and the
echo of VoIP calls are closely monitored. The latest version of Agilent's
Telegra Voice Quality Tester (VQT) measures these quantifiers and offers
other more traditional tests through a choice of analog, T1/PRI, or E1
interfaces. It uses a signal acquisition card for the DSP engine in a
portable Telegra "R" chassis and runs the software on Windows NT.
INSTALLATION
Since the Telegra VQT always comes with the hardware and software
preinstalled, there is not much to say about the installation process.
Essentially, it was unlatch the keyboard, plug in the power cord, and boot
up the computer. Of course, we also needed to connect either the T1 or
analog connections, depending on which we were using (both cards were
preinstalled as was an Ethernet card in case we wanted to connect to our
LAN). The system came with the software on a backup CD just in case there
was something wrong with the preinstalled software. We didn't have such a
problem, but went through the software installation to see how the process
worked. It turned out to be a typical software installation that only took a
few minutes.
Configuring the port settings for either the analog or T1/PRI interfaces
(an E1 connection uses yet another interface) was not difficult. To set the
analog ports, we only needed to accept and apply the default settings. If we
wanted, we could have selected different ports and configurations, such as
the signaling type and off hook to dial delay, but the default settings
suited us just fine. Setting the T1/PRI ports took a little knowledge of
what line coding, framing, and signaling type was used for our network.
Since we use the Gordon Kapes T1 simulator for the network, we kept its
settings and made sure that the Telegra VQT matched these settings except
for making the VQT set to the user side.
DOCUMENTATION
Both the printed materials and the help files effectively tell users about
the Telegra VQT. The Getting Started Guide gives the user the general idea
about all of the tests the product has to offer. It also displays plenty of
screenshots and a decent index. Accompanying this guide is a book that
focuses on IP telephony in general and a CD-ROM that includes more
information about Internet telephony, sample capture files of the major VoIP
protocols (H.323, SIP, MGCP), and live examples of VoIP traffic. The book
and the CD-ROM are certainly a welcome addition to the Telegra VQT
documentation.
While the help files are not quite context sensitive, they may be at least
as impressive as if they were. When you first access the Telegra VQT
interface, a Welcome screen pops up so that you can review the help files
before exploring the VQT interface. When using the interface, contents,
index, and search help files are also available. Perhaps more advantageous
is the information available on the bottom right corner of every screen,
which details the important information particular to the features of that
screen. It is an embedded help mode in which you can choose to display the
basic or advanced help information and click on the appropriate information
you need. You may also choose to hide these embedded files from appearing on
the screen. This method proves to be a very effective means of learning
about each individual voice quality test.
FEATURES
The primary features of the Telegra VQT involve testing VoIP packet systems
for voice quality information and the equipment (gateways, switches, etc.)
that makes up a VoIP network. These tests include Perceptual Speech Quality
Measure (PSQM), Perceptual Analysis Measurement System (PAMS), Echo-PACE,
and Echo-DTalk scores. Other tests includes DTMF tone detection and voice
activity (sometimes called silence suppressors) detection. Audio testing
tools for noise and tone generation can be applied to .wav files that can be
played back and recorded. Automated test sequences can also be implemented
for further analysis of VoIP calls.
OPERATIONAL TESTING
First, we examined and tested the T1/PRI interface. Once we hooked up the
VQT with the Gordon Kapes simulator and we synchronized those settings, we
set four digit numbers in the Call Control screen and placed a simulated
call. From here, we ran a test that measured the delay of the call as well
as a test for PSQM scores, which measures the clarity of the call. Because
of the nature of the T1 test we set up, the scores were extremely low, which
means that there was almost no delay and that the call was extremely clear.
This result is to be expected testing a regular T1 call from a simulated lab
environment.
Next, we wanted to test VoIP calls to obtain more genuine examples of how
the Telegra VQT would typically perform. We used two PUREDATA RT 201V
Internet Telephony Gateways (ITG), one containing a four-port FXO module and
the other containing a four-port FXS module. We set up the gateways and a
dialing plan so that we could call into the FXO module of the ITG through an
analog Teltone simulator. From the analog simulator, we connected a line to
the first FXO port on the VQT. From the VQT GUI, we entered
"5551000,,,,,201", which dialed into the first ITG VoIP gateway
and then paused the few seconds necessary to connect. Afterwards, the DTMF
digits "201" were transmitted, causing the ITG's dialing plan to
route the call to the second ITG gateway across the IP network. The second
ITG gateway received the IP call and then rang extension 201, which we
actually connected back to the VQT to complete the loop. With this loopback
in place, the VQT was capable of transmitting sounds as well as receiving
those sounds, and thus, it was able to calculate the difference in time
(latency) or sound quality degradation (PSQM or PAMS).
Through this method, we could test for delay, PSQM, PAMS, and other scores
that are associated with VoIP. Keep in mind that the lower the PSQM score,
the better. Below 3 for VoIP calls is considered very good. On the other
hand, the higher the PAMS scores, the better (with the highest score being a
5). Obtaining dial tone through the Teltone simulator, the PUREDATA ITG's
scores tested with an average delay of ~121ms, average PSQM of ~2.77, and an
average PAMS scores of ~4 for listening quality and ~4.3 for listening
effort. Over all, the scores for PUREDATA's gateways
indicate quality VoIP calls. For more information about the PUREDATA RT 201V
Internet Telephony Gateway, please read their review.
There was latency introduced by the Teltone simulator, which we needed to
measure. However, we tested this simulator by hooking it up to the VQT
system and discovered that the latency introduced was virtually negligible
at a miniscule 0.125ms. As a reference, the Teltone simulator had a PSQM
score of ~1.05, which indicated excellent quality, and therefore, most
likely did not affect the VoIP scores. Remember, however, that the PSQM
scores are non-linear and are attained using complex algorithms. Trying to
"subtract" the Teltone's PSQM scores from a gateway's actual
scores would not give an accurate figure. Nevertheless, the PSQM figures
attained using a Teltone PSTN simulator is extremely close to the
"true" PSQM value of PUREDATA's ITG when testing on lines with
only small amounts of voice degradation or bandwidth issues.
We compared our test results for the analog Teltone simulator with the
Gordon Kapes T1 simulator. As expected, the T1 scores showed better quality
than the analog scores because of its digital connection. For instance, the
average analog PSQM score was 1.05 while the average T1 PSQM score was a
superb score of 0.21. Since the PAMS of the Teltone simulator do slightly
affect the overall PAMS score of PUREDATA's VoIP gateway, we also tested for
these scores. The simulator's excellent PAMS scores averaged a Listening
Quality Score of 4.45 and Listening Effort Score of 4.64.
In addition, we also used the VQT to test a "real world" PSTN call
using regular analog lines. While the delay was almost non-existent, scoring
the same as the Teltone simulator, the PSQM scores were actually a little
higher and thus slightly worse than the VoIP scores that used the Teltone
simulator. However, keep in mind that when we used regular analog lines to
call the local gateway in each direction and calling from gateway to gateway
via IP, these VoIP calls averaged more than a point higher than they do
testing through an internal simulator because of traffic and voice
degradation on the PSTN lines. The PAMS scores of typical PSTN calls were
slightly better than the PUREDATA ITG scores through the Teltone simulator
and averaged nearly a half-point better than the VoIP tests using PSTN lines
(Table 1).
Furthermore, we tested VoIP calls at ConvergeNET using the VQT. Since
ConvergeNET is an interoperability showcase of live VoIP calls from one
company's booth to another one, these tests were especially important. The
PSQM results averaged below 3, proving that quality VoIP calls can be made
even on a busy network. The tests were successful even when using two
different vendors' H.323 gateways. This certainly showed promise for both
VoIP testing and for VoIP interoperability.
Table
1 |
Delay |
PSQM |
PAMS
LQS |
PAMS
LES |
Teltone
simulator |
0.125ms |
1.05 |
4.45 |
4.64 |
Gordon
Kapes T1 simulator |
0.87ms |
0.21 |
Not
Tested |
Not Tested |
PSTN
call |
0.125ms |
3.08 |
4.03 |
4.39 |
Teltone-to-ITG
then VoIP to other ITG |
127ms |
2.77 |
4 |
4.3 |
PSTN-to-ITG
then VoIP to other ITG |
121ms |
3.95 |
3.66 |
3.93 |
ROOM FOR IMPROVEMENT
For testing Internet telephony gateways, the Telegra VQT is the top of the
line. Our suggestions involve testing other Internet telephony equipment
more easily. We could not test Microsoft NetMeeting or the IP phones we had
here in the lab because the VGT requires that the calls must be initiated
and terminated on one of VQT's telephony boards (T1/E1/PRI or analog) as
opposed to initiating or terminating a VoIP call from an IP connection. We
think that if Agilent provided a virtual H.323 gateway on the unit, it could
both initiate and terminate VoIP calls via an Ethernet connection, thereby
allowing virtually any Internet telephony equipment to be tested. This would
benefit the VQT and the entire Internet telephony industry. From talking
with Agilent representatives, there may very well be a product to support
the VQT in this way in the near future -- maybe even by the time this review
is in print.
While the graphical interface is admirable, we would like to see the analog,
T1, and E1 interfaces integrated. Currently, for example, we need to
disconnect a T1 test and the interface in order to open the analog interface
and conduct another test. It might be advantageous to be able to run both
tests concurrently or at least be able to schedule a specific time to run
the tests one after another. Integrating the GUIs would help towards that
end. On a more minor note, right-click functionality would also aid
usability.
CONCLUSION
The Telegra VQT is a sophisticated and intuitive piece of testing equipment.
Specifically catering to the Internet telephony market, it delivers accurate
scores for VoIP elements of gateways, routers, and switches. Any company in
this industry as well as the labs testing the equipment would find the VQT
as an extremely valuable resource. With the help of this resource, we will
see VoIP gateways and the like consistently improving so that the quality of
the calls will soon be of equal quality with close to the same rate of delay
to that of the PSTN
What's New In The Next
Version Of Agilent's Telegra VQT?
VQT version 4.0 began shipping March 1,
2001. It offers these features:
- PC client software for remote
control of VQT servers.
- Remote audio feed.
- Distributed one-way PSQM+ and PAMS
testing between two VQT servers at different sites.
- Signal Loss Measurement.
The VQT client allows the user to:
- Control VQT servers remotely from a
PC.
- Create and run common test scripts
for multiple VQT servers on a single PC.
- Compile test results from multiple
VQT servers on a single PC.
Coinciding with VQT 4.0 is the release
of the Agilent IP Telephony Reporter. Reporter can run on the same PC
as the VQT client. It produces graphical reports of VQT test results.
Reporter also compiles reports of VQT test results correlated with
test results from the Agilent IP Telephony Analyzer, which measures
VoIP packet loss and jitter performance. |
[ Return
To The May 2001 Table Of Contents ]
|