home *** CD-ROM | disk | FTP | other *** search
- Path: inforamp.net!ts49-12
- From: crs0794@inforamp.net (Geoffrey Welsh)
- Newsgroups: comp.dcom.modems
- Subject: Re: are 'byte-wide' modems possible
- Date: 18 Apr 1996 04:25:48 GMT
- Organization: InfoRamp Inc., Toronto, Ontario (416) 363-9100
- Message-ID: <4l4g8c$968@sam.inforamp.net>
- References: <Ken.Crossman-1104961348100001@ts1-port4.mas.ualberta.ca>
- NNTP-Posting-Host: ts49-12.tor.istar.ca
- X-Newsreader: News Xpress Version 1.0 Beta #4
-
- In article <Ken.Crossman-1104961348100001@ts1-port4.mas.ualberta.ca>,
- Ken.Crossman@ualberta.ca (Ken Crossman) wrote:
- >The latest word about modems has it that 33.6kb is about the limit. If I
-
- "About the limit" is a good description. I'm sure that we could go higher,
- but why bother putting research into slight gains? We certainly won't be
- doubling the bit rate again unless the entire telephone network is
- revolutionized.
-
- >understand modem transmission correctly, the modem encodes data onto a
- >phase-shifted audio carrier, with the speed of transmission governed
- >partly by the number of possible phase shifts per cycle. Could a modem be
- >designed such that each phase shift represents a complete 8-bit word, ie.
- >if there were 256 possible discrete phase angles then 256 (or 2**8) unique
- >digital words could be represented? The transmitting modem would encode
- >the 8-bit word received from the computer to one of the 256 possible phase
- >angles. The receiving modem would translate the phase angle back into the
- >8-bit word it represents. Or is this 'byte-wide' technology now in use?
-
- Actually, it's a combination of phase and amplitude. My
- understanding is that, neglecting echo cancellation, a modem works as follows:
-
- - Take raw bits, translate them to a phase/amplitude signal via a state table
-
- - Perform a reverse Fourier transform on the carrier frequency, phase shift, &
- amplitude to create a wave description
-
- - send that wave down the line
-
- - other end receives and digitizes that wave,
-
- - performs a Fourier transform on the wave data,
-
- - and obtains a result table showing frequencies, amplitudes, and relative
- phases of all component signals...
-
- - hopefully, the only significant one is the signal that the transmitting
- modem intended, which is then translated back into raw bit patterns
-
- However:
-
- Don't forget that the receiving end must be able to tell each of your 256
- phase shifts apart with some degree of certainty after the noise, frequency
- shift, phase jitter, etc. have done their worst. If you increase the
- resolution of the phase shifts, you reduce your margin to the point where you
- can't tell the difference above the noise.
-
- Also remember that most of the phone system except for the bit connecting you
- to the calling office is now digital, with a best case being an 8 kHz, 8-bit
- sample minus some supervsory overhead. Without getting into technical details
- that I'm not qualified to discuss with authority, I apply an information
- theory analogy of the law of conservation of matter... namely, that you don't
- get something for nothing.
-
- The digital connections talk binary to each other at 64 kbps. That means that
- no more than 2 ^ 64,000 distinct signals can be sent per second; there is no
- reliable way to get more than that through the channel and bring it out intact
- at the other end. Of course, with analog modems, we're trying to use digital
- technology to send and receive an analog signal which makes best use of the
- analog bandwidth of the typical phone line, the bandwidth of the telephone
- companies' A-D and D-A converters, and the digital sampling architecture... as
- a result, most modems fall somewhat short of the 64 kbps ideal.
-
- Improving technology has brought us closer to that point, but we'll never
- surpass it unless the sampling setup of the international telephone system is
- revised... and you can guess at the effort and cost of such a step.
-
- It looks like there's not much point developing analog modems anymore; we
- might as well accept that the next significant speed increase will be
- all-digital, be it ISDN arriving decades late, 'cable modems', or some new
- form of digital service from the phone company such as ASDL.
-
- --
- Geoffrey Welsh, Developer, InSystems Technologies Inc.
- Temporary: crs0794@inforamp.net; At work: insystem@pathcom.com
- At home: geoff@zswamp.uucp or [xenitec.on.ca|m2xenix.psg.com]!zswamp!geoff
- TYPING IN ALL CAPS IS GROUNDS FOR IMMEDIATE DISMISSAL.
-