home *** CD-ROM | disk | FTP | other *** search
- Path: in2.uu.net!insync!usenet
- From: bretting@insync.net (Greg Bretting)
- Newsgroups: comp.dcom.modems
- Subject: Re: Does anyone know about "flat-line" performance graphs?
- Date: Sun, 14 Jan 1996 12:29:39 -0800
- Organization: - not one of my strong points, really...
- Message-ID: <4dbkda$a9e@drencrom.insync.net>
- References: <DL2yC0.Gwp@txnews.amd.com>
- NNTP-Posting-Host: line-213.insync.net
- X-Newsreader: Forte Agent .99b.112
-
- On Fri, 12 Jan 1996 18:07:11 GMT, Your Name <first.last@amd.com> wrote:
-
- >Hello,
- >
- > My name is Sharif Sazzad and I am trying to find out
- >about how the performance of modems had improved during its
- >evolution (V.21, V.22, V.32, V.34)
- > To this end I have been told that "flat-line" graphs
- >are useful in comparing modem performance. Unfortunately I
- >don't know where such data can be found.
-
- Hmmm... I'm not certain, but I'm assuming that you are referring to the
- so-called "waterfall" curves that were once pretty popular. This is
- typically a BER or BLER measurement where the modems are forced to connect
- at a given modulation/speed (such as V.34/28800) and the SNR
- (signal-to-noise ratio) is swept from a very low value (lots of noise) to a
- higher value, with a call being made and data passed at each SNR step. The
- modems are set up in such a way that EC protocols and compression are
- turned off, retrains and renegotiations are disabled, and the modems are
- only allowed to connect at the target speed and no lower. The test is
- usually run on a flat, unimpaired phone line with only the target
- impairment (such as impulse noise level or 1kHz tone to total distortion
- (T/TD)) being injected and varied as the test is run. The graph results
- from plotting the BER/BLER (bit/block error rate) as the modems go from a
- no-connect to a no error state. The SNR is plotted along the X axis, and
- the BER along the Y axis of the graph. This will result in a
- waterfall-type curve.
-
- This is useful for some things, but comparing modem performance isn't
- really one of them. This test is typically used to characterize the
- performance of the modem DAA and the datapump, and only gives you part of
- the performance picture. The impairments used in this type of testing do
- little to predict how a modem will perform on real POTS lines. This test
- is best suited to things like component selection and characterization
- during DAA design, measuring modem sensitivity and testing of the datapump.
- I've seen modems that did well in this type of testing that fared poorly on
- POTS lines (real or simulated) due to problems associated with the error
- correction protocol layer other other firmware bugs that caused them to
- have problems passing data or connecting at the expected rate under certain
- line conditions.
-
- If you are interested in a standardized test for modem performance
- comparisons, you might want to look into the TIA TSB-38 test suite, and the
- line characteristics defined in TIA TSB-37A and ITU-T V.56bis. This offers
- a more comprehensive test strategy that does a better job of predicting
- modem performance over POTS lines. It's not perfect either, but you'll
- learn a lot more about relative modem performance using this approach than
- you will with waterfall curves.
-
- This is especially true with V.34 modems, which are usually compared by
- performing the tests defined in the TIA TSB-38 spec.
-
-
-
-
- > If anyone out there knows where I can find such data
- >I would be very grateful if they could share their knowledge
- >with me.
- > Also please check out my homepage at
- > http://vcsun2.tamu.edu/~sharif/sharif.html
- >
- >
- >Peace
- >Sharif M. Sazzad
- >sharif@vanzandt.amd.com
- >
-
- --
- | Greg Bretting |"The whole problem with the world is that |
- | bretting@insync.net |fools and fanatics are always so certain of|
- | --==<< >>==-- |themselves, but wiser people are so full of|
- | |doubts." - Bertrand Russell |
-
-