Great reporting, JEM, appreciate it. Of course, any meaningful tests would need to be conducted in a hostile environment. But whether hostile conditions actually existed, and to what extent, could still be a subjective assertion, unless real numbers are published along with those claims. Hey! I get paid to be skeptical. What can I tell you? Besides, the devil makes me this way. <g>
For the sake of discussion, and after establishing a level of trust in your research, let's accept your findings for their face value, and declare this one yours, and a win for VoIP at the same time. I'm all *for* that, after all, despite my seemingly contrary stance, at times. Congrats!
When you say using QoS (still not a term that the IETF prefers to use, in deference to class of service <COS> and other real-time and near-real-time terminology), are you referring to real time protocol (RTP)? Resource reservation (RSVP). Is this done using UDP protocol? Please explain what they did in greater detail if you have the information at your disposal, and if you can do so without jeopardizing confidence. Better yet, is there an ftp site or web page that covers their tests?
In reply to your question about what test methods would be suitable, I have yet to formulate in my mind an approach that I think would be effective, using conventional metrics and existing paradigms as a reference. I suspect that the response time measurements will be easily established and standardized fairly quickly if they haven't already, but the signal quality at the analog level is another story.
From what I've seen, historically, and most of the measurement schemes that have been described in "public bakeoffs" (cough) in the trade press, many are good attempts, but in the end, they are laughable to anyone who has done time exploring the tradeoffs that usually go into the definition of test parameters. Downright pedestrian and bread-board level, thus far.
But good attempts, nonetheless, in the absence of any TIA, IEEE or ITU carefully-derived and universally-agreed-on test protocols. This area should lead to an interesting and undoubtedly heated series of debates, because there will be a lot at stake, since test instrumentation can be made very friendly to one vendor's product philosophies, while extremely hostile to another's. It's all in the numbers and the way the parameters are packaged.
A very interesting thing arises here, as a result of the resurgence in the interest of analog voice quality. Whereas
(1) the quality of voice services has been extremely good in the recent f-o digital years, and
(2) "analog data" has virtually disappeared (except for dialup modems and xdsl, if one 'must' consider xdsl analog),
...analog test sets have been, for the most part, relegated to power and noise levels, continuity testing and occasional nonlinear distortion tests for those remaining dedicated lines used for data (more often than not to overseas destinations due to the poorer quality of those lines in many instances).
The more sophisticated devices that we needed during the late Seventies and early Eighties (when "high-speed" 9.6 and 19.2 kbps lines first arrived) have gone by the wayside. With the arrival of VoIP, however, this interest will undergo a rebirth, and new equipment standards and products will again focus on the analog side of the equation once again. Keep your eyes open for analog and digital test set manufacturers who specialize in VoIP test and measurement!
Formulating a test and measurement paradigm will be almost as politically heated (maybe more so) as the selection of a compression algorithm was for the IMTC draft (and still is where old wounds don't go away easily).
Let me think about this some more, and I'll get back to you later or over the weekend. BTW, there's this excellent application I came across using wireless. Care to hear about it? <smile>
Regards, Frank Coluccio |
