home *** CD-ROM | disk | FTP | other *** search
- Path: dispatch.news.demon.net!demon!bram-sys.demon.co.uk
- From: pete@bram-sys.demon.co.uk (Pete Jones)
- Newsgroups: comp.dcom.modems
- Subject: Re: What triggers retrains and fall-forwards?
- Date: Fri, 08 Mar 1996 23:57:59 GMT
- Organization: Bramble Software
- Message-ID: <3140bd05.2634257@news.demon.co.uk>
- References: <Dnw50n.7n3@abs.net>
- Reply-To: pete@bram-sys.demon.co.uk
- NNTP-Posting-Host: bram-sys.demon.co.uk
- X-NNTP-Posting-Host: bram-sys.demon.co.uk
- X-Newsreader: Forte Agent .99d/16.182
-
- dbryan@mail.bcpl.lib.md.us (J. David Bryan) wrote:
-
- >Dear ListFolks,
- >
- >I understand that line conditions may cause retrains and fall-backs in
- >response to degradation (determined by errors in transmission?), but what
- >causes the modems to fall-forward? That is, once the modems are operating at
- >a slower speed and transmission errors are no longer occurring, what prompts
- >the modems to attempt to speed up the communication?
- The error rate of the link. Modem designers use statistics gathered
- by the protocol to continually optimise the modem throughput for the
- conditions.
- > Is it a function of time, or some number of packets received without error?
- Both; One good packet does not indicate an error free link!
-
- The number of error free packets is taken into consideration, together
- with the nature of any errors received.
-
- If a link is suffering from random bit errors reducing the speed of
- the link may reduce the errors. If the link has suffered a burst
- error, then the data rate will have little effect on the number of bad
- packets on the link. To avoid problems caused by burst errors, bits
- are often transmitted by the modems out of sequence to ensure that a
- burst error will hit a number of bits in a number of packets rather
- than totally obliterating a single packet. This gives the error
- correction a chance to recover! (I think that this is known as
- trellis encoding??).
-
- > Are there user settings which can alter the criteria used?
- I am not familiar with the configuration of off the shelf modems, but
- beware. The majority of protocols have a large number of
- "configurable" items which allow the operation of the protocol to be
- "tuned". We are, at present, involved in such an exercise with a
- radio modem, and it is an engineering nightmare. By all means play
- with parameters and learn about what is going on. I would be
- interested to hear if you improve things. I would advise that you
- keep a record of the factory settings so that you are able to
- "de-optimise" your kit.
-
- >I have connected initally at a reduced speed, worked for a while, and then
- >have had a retrain improve the transmission speed. What triggered the modem
- >to attempt the retrain?
-
- >Thanks for any insight into this phenomenon.
- >--
- > -- Dave Bryan
- > dbryan@mail.bcpl.lib.md.us
- >Pete Jones
-