"Those who ignore history are doomed to repeat it."
Well, it can also be said, "Those who ignore testing are
doomed to repeated mistakes." Within the telecom and
datacom worlds, testing is an absolute requirement in
order to ensure product quality, performance, and
interoperability. Empirix (formerly Hammer Technologies)
makes several testing products specifically focused on
the telecom world, but are also now crossing into the
datacom world with their support for VoIP testing as
well as network load/traffic generation via their
PacketSphere product.
"Our customers who design and deploy VoIP network
equipment are discovering that they can't ensure good
voice quality without also understanding how real
networks affect performance," said Gordon Eddy,
Marketing Manager at Empirix. "With the new level
PacketSphere integration in the VoIP 2.7 release, we've
added that crucial dimension. It's easy to create
realistic network conditions in the lab and flush out
problems before they become very expensive to fix. The
many other new test capabilities in the VoIP 2.7 toolkit
provide our customers with an even more powerful
solution for verifying their VoIP components and
networks."
TMC Labs examined Empirix's new Transport product,
which we affectionately call the "Mini-Me Hammer" due
its small portable self-contained enclosure (Dolch-like
box) with an embedded flat-screen display and with the
same feature-set as the original Hammer IT system. The
Empirix Transport features real-world testing
capabilities that include testing voice (T1/E1 and
analog interfaces), fax, modem, CAS, ISDN, VoIP, and SS7
signaling. It also features QoS metrics and can measure
connection and speech latency, and analyze voice quality
utilizing the PSQM and PAMS standards. Also, we received
a pre-release beta of the Hammer VoIP 2.7 software
running on the Transport system, which includes a new
set of voice quality test capabilities, including front
end clipping measurement for VADs, echo cancellation
testing and the PAMS voice quality scoring.
INSTALLATION
There was no installation required for the Transport. It
comes pre-installed with the appropriate software and
hardware taking any installation complexities away from
the end-user.
DOCUMENTATION
We were quite pleased with the documentation. The online
help was pretty good, though the help file for the
reports could have a bit more descriptive. We especially
liked the step-by-step instructions in the Hammer
TestBuilder Getting Started Guide. It explains from
start to finish how to use the very complex and
comprehensive test system in just a very short number of
pages. Similarly, the Hammer VoIP Test System Getting
Started Guide was also well-organized and included
step-by-step instructions. For such a complex test
system, we were quite surprised that the documentation
so easily explained how to use the system. Indeed, the
documentation has improved immensely from prior
documentation we've seen from past Empirix releases.
FEATURES
One of the first features of the Hammer Transport we
should point out is that that it is a lightweight,
portable test platform configurable with four T1/E1 and
eight analog interfaces for testing. The Transport is
housed in a portable, rugged chassis with carrying case
and measures just 10.75" H, 8.5" D, and
16" W and weighs only 24 pounds. The default specs
include an 866MHz CPU with 256 MB RAM, 20GB hard drive,
CDROM, 14.1-inch high-resolution (XGA) active matrix
color display, 10/100 network card, and a floppy drive.
Creating customized test scripts, scheduling test
scripts, and monitoring running scripts are the core
components of the Transport system. Using scripts, you
can perform real world testing with voice, fax, modem,
and tone traffic, CAS, ISDN, and SS7 signaling. Testing
of VoIP is available as is the ability to analyze voice
quality via the PAMS and PSQM standards. An
oscilloscope-like application called VQScope enables
visual inspection of the sound wave forms for in-depth
voice quality analysis.
Other features include:
- Monitor activity on all channels simultaneously or
drill down to view detailed activity on a single
channel for error diagnosis;
- Ability to listen to channel activity;
- Real-time monitoring of run time statistics;
- Pre-defined test reports for simplified test
analysis;
- High-level summaries and granular
channel-by-channel reporting;
- Isolate and view results from specific time
periods;
- Graphical summary reports;
- Metrics for measuring the performance of gateway
Voice Activity Detection and Echo Cancellation
features;
- Saves test speech files above user definable PSQM
score threshold for further analysis; and
- Optional Hammer Visual Basic Programming
Environment.
OPERATIONAL TESTING
Using the Hammer Transport comprises five main phases:
- Setting up the telephony boards to use a specific
protocol (T1/E1, analog) to match the system under
test (SUT) (setup from the Hammer Configurator);
- Creating your test scripts (Using TestBuilder, an
app-gen like application);
- Making calls through the test scripts (Can execute
scripts via the Call Profiler, the Scheduler, or
execute manually);
- Real-time monitoring and analysis of the calls
(utilizing System Monitor, Call Summary Monitor, and
the Quality of Experience applications); and
- Post-analysis reporting (Reports plus VQSCOPE).
We tested each of these five phases extensively and
our results are as follows:
Phase 1 -- Set up telephony boards and connect
to System Under Test (SUT)
From the Hammer Configurator we selected the T1
protocol with ESF framing and B8ZS line encoding to
match our system under test, namely a Gordon Gapes T1
simulator. We decided to test the Transport's
functionality by utilizing the Gordon Kapes T1 simulator
for a couple of reasons. First, it happened to be a T1
product we had in the lab at the time. Secondly, we've
done speech quality measurements (PAMS and PSQM) on the
Gordon Kapes using Agilent VQT (Voice Quality Tester) in
the past so it would give us a good benchmark
comparison. In any event, after choosing the T1
protocol, framing, line encoding, and clock source, we
saved the configuration and applied the settings, which
then downloaded the proper protocol settings into the
Natural MicroSystems boards. Also, since the Transport
comes with four T1/E1 spans and the Gordon Kapes only
has two spans, we only applied the T1 protocol to just
two of the four Transport spans. We should also mention
that you could quickly test a variety of protocols since
the Transport supports mixing and matching of different
protocols across the various spans. We connected a T1
cable from Span1 on the Transport to Port1 on the Gordon
Kapes, and then we ran a T1 cable from Port2 on the
Gordon Kapes back to the Hammer Transport, connecting it
to Span2. We configured the Gordon Kapes simulator to
redirect calls from its Port1 to Port2 so calls were
looped back to the Hammer Transport system, which is
required for latency and speech quality measurements.
Phase 2 -- Creating Test Scripts
The Transport comes with several test scripts, including
one designed to test speech quality and score the
results based on PSQM or PAMS measurements. From the
Hammer TestBuilder we designed our scripts utilizing a
graphical app-gen like interface. TestBuilder utilizes a
simple ladder diagram with the left side (A on diagram)
acting as the originating side, the middle (N on
diagram) acting as the network, and the right side (B on
diagram) acting as the terminating side. Four tabs next
to the ladder diagram contain different icons with
actions that can be placed on the ladder diagram, such
as play prompt, calculate voice quality, send digit, and
more. Creating scripts was an easy and straightforward
process. The Transport also comes with several useful
sample scripts that run without any modification.
Phase 3 -- Making Test Calls
Just like the original Hammer IT system, the Empirix
Transport supports several means of initiating test
calls to your scripts. One way is to use the Scheduler.
We were able to quickly schedule a script by
right-clicking on a script within the Library Manager,
and then choosing "Edit & Run." The Scheduler window
pops up where you can then set the start time, channels,
maximum number of connections, maximum test time, and
other parameters.
We did have some difficult using the right-click "Edit
& Run" feature if the script was modified in any
way. Once you try and execute the script using the "Edit
& Run" command, a message is displayed saying that
we must save the script first. We then clicked "ok" and
then clicked on the floppy disk icon on the toolbar to
save the script. We then tried to run the script and
again it gave us the same message saying we had to save
the script first. We just did that! After several
frustrating attempts, we finally realized what was
happening. The message saying we must save the script
first was changing the focus of the screen to the
Library pane instead of the script pane.
Thus, when we tried to save the script we were in
fact saving the Library instead of the script. Once we
clicked on the script pane and then clicked the floppy
disk icon, the script was successfully saved and we
could execute the script. Once we knew this fact, it
wasn't that hard to save the script. We believe Empirix
copied the model utilized by several development
programs such as Visual Basic where it has "project
files" and then individual modules (forms, header files,
etc.) which are saved independently according to where
the "focus" is located, i.e., which screen is in the
foreground. However, these development programs at least
have separate "save" menu options, which allow you to
save individual modules or the entire project. In
addition, they also give you the option of saving the
entire project if a module is modified. The Empirix
software does not provide these options.
In addition to using the Scheduler to initiate calls,
we also tested using the Call Profiler, which allows you
to apply a traffic pattern to your test calls. For
example, you can perform call patterns such as saw tooth
(ramps call up then down), blast, Poisson, and others.
We should mention that the Hammer also has a CLI
interface for executing scripts, which we looked at, but
didn't actually test.
We were most interested in the "Speech Quality"
script, so this was one of our first scripts we
executed. As can be seen in the ladder diagram, this
script not only "calculates" the VQ (voice quality), but
it also measures speech latency. Since we had connected
the Transport directly to the Gordon Kapes with two T1
cables just 10 feet long, we expected the measured
latency to be minimal and the voice quality score to be
very good and indeed our supposition was correct. PSQM
scores range from 0 to 5 with 0 being the best quality
and 5 the worst. In the multi-page report, the PSQM
scores for the system under test were below 1, ranging
from 0.28 to 0.68 for the boy/male prompts. On the
second page of the report (not shown), the girl/female
prompts also had superb voice quality scores. The PSQM
average for all the prompts was 0.53. The Transport
displayed the speech latency as averaging 0.71ms. The
low PSQM score is to be expected, as a low number
indicates better quality with 0 having no degradation.
Also, the very low latency (0.71ms) should be expected
due to the short T1 wires and digital T1 connection to
an internal test environment. Comparatively, the Agilent
VQT PSQM scores for the Gordon Kapes were slightly
different, averaging a 0.21 PSQM score and 0.87ms
latency.
We should state that the PAMS (Perceptual Analysis
Measurement System) voice quality scoring algorithm is a
new feature in the 2.7 release. The software includes
features to test the effectiveness of echo cancellers
and VADs (Voice Activity Detectors). We ran a script to
measure and score PAMS and it performed flawlessly.
Phase 4 -- Perform real-time monitoring and
analysis of the calls
While the Hammer was making test calls into the Gordon
Kapes system, we had the capability to listen in on the
calls using the Hammer System Monitor application along
with PC speakers connected to the Transport. Simply
clicking on the channel allowed you to instantly "tune
in" to the test call being made on that channel. The
Hammer System Monitor also displays log information on
the channel selected. Test engineers have the option of
selecting various logging levels, including None,
Failure, Normal, and Debug. When a script is running, if
logging is set to Normal or Debug, you will be able to
see events on the screen as they happen, such as "Place
Call," "Listening for voice," "Transmitting digits," and
so on. With the speakers enabled and tuned to that
channel you can both listen and see exactly what is
taking place while the test script is running.
In addition, the Transport also features two other
real-time monitoring applications: Call Summary Monitor
and the Quality of Experience applications. The Call
Summary Monitor displays a graph with number of current
connected calls, calls per hour, successful calls,
failed calls, and other statistics. The Quality of
Experience application includes real-time statistics,
including a graph displaying whichever variables are
selected on the top half of the screen. Our only
complaint was that we couldn't seem to change the zoom
level or adjust the height of the Y-axis within VQScope.
According to the online help, this feature is available,
but we weren't able to perform this feature as stated in
the online help. In fact, the Zoom option was nowhere to
be found.
Phase 5 -- Post-analysis reports
The Transport comes with several built-in reports,
including Call Summary, Call Detail, Error Detail, VoIP
Summary, VoIP Detailed, and more. We ran several reports
and were pleased with the quality and readability of the
reports.
As part of the post-analysis, the VQScope application
is utilized to compare two waveforms -- that of the
original canned prompts and that of the canned prompts
after they've passed through the system under test to
the terminating end of the Transport. Obviously the
quality of the canned prompts only slightly degraded
since we were testing with the Gordon Kapes T1 simulator
as opposed to a VoIP gateway. We were able to select the
threshold whereby the system would automatically capture
and store any calls that were below a certain level of
quality and then view the before and after waveforms
using VQScope's oscilloscope-like interface.
When we performed a VoIP Detail report and a VoIP
Summary report, we noticed that the "average speech
latency" numbers in the two reports did not "jibe" with
one another. They should have been exactly the same. A
quick call to Empirix revealed that we had discovered a
bug. In the VoIP Summary Report, zeros (0 milliseconds)
are used in the calculation of the average speech
latency. In the VoIP Detail Report, zeros are NOT used
in the calculation. We confirmed this by dividing the
speech latency average in the VoIP Detail report by the
number of tests performed and then the numbers for both
reports did indeed match up. Empirix is releasing a
patch to fix this minor bug.
ROOM FOR IMPROVEMENT
Unfortunately, you cannot perform PAMS and PSQM scoring
within the same running script. When we attempted to
have both of these within the same script we received an
error during the compile. We confirmed with Empirix that
this is by design. Although we could schedule two
separate speech quality scripts a few hours apart, this
would increase testing time. Thus, we would like to be
able to perform PAMS and PSQM scoring within a single
script.
Although you can schedule scripts, you cannot
schedule scripts using the advanced Call Profiler which
features calling patterns. So we would like to see this
in a future release. Another suggestion related to the
scripts is moving items on the ladder diagram. First,
you cannot drag and drop actions or move action items
around using the mouse. Second, in order to move items,
you need to highlight the arrow connecting the two
actions and then click the up or down icons. We kept
highlighting the action labels instead of the arrow and
then clicking the up or down icons, which unfortunately
doesn't work.
We had some minor usability complaints. Within the
Hammer TestBuilder you cannot rename the scripts. You
must do a File, Save As with a new filename, and then
delete the original script file. Also, you cannot
highlight one or more arrows or items within the Ladder
diagram and just hit the Del key on the keyboard,
although you can right-click delete items one at a time.
Finally, we'd like to see a sound player included such
as CoolEdit for playing back recorded test calls
recorded in PCM format since the Windows Media Player
doesn't support this format.
CONCLUSION
This product covers all the bases including load/stress
testing, VoIP testing, speech quality analysis,
real-time monitoring, and post analysis comprehensive
reports. TMC Labs has always enjoyed using past versions
of the Hammer product line. Now, with the portability of
Empirix's Hammer Transport, lab technicians can use the
Hammer Transport in the lab or on the road. Whether it's
the full tower version of the Hammer or the portable
version, the testing feature set is very impressive and
in our opinion no telecom testing lab should be without
one.
[ Return
To The November 2001 Table Of Contents ]
|