TCP-group 1995
[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
Re: FEC paper, implementation
- To: fgoldstein@bbn.com
- Subject: Re: FEC paper, implementation
- From: Phil Karn <karn@qualcomm.com>
- Date: Mon, 20 Mar 1995 16:03:15 -0800
- Cc: nelson@crynwr.com, pe1chl@wab-tis.rabobank.nl, pe1chl@wab-tis.rabobank.nl, geertj@ripe.net, henkp@paramount.nikhefk.nikhef.nl, tcp-group@ucsd.edu
- In-reply-to: <199503201852.KAA22041@UCSD.EDU> (fgoldstein@BBN.COM)
>There's no need to burden the Ethernet link and DSP box with TCP/IP or
>anything else like it, except for say management functions. Let the PC talk
>Ethernet. Let the DSP-based "modem" talk Ethernet, but as a dumb "bridge"
>or even "repeater". (Bridges can be shared among multiple 10-base-2 bus
True at least for TCP. For ISDN relay boxes, though, I'm a bit
uncomfortable with simple bridging because of the need to intelligently
filter broadcast traffic.
For the DSP application, you could just send raw ad-hoc packets over
the Ethernet and let the software on the PC do the rest. E.g., on
receive, the DSP box runs the correlator looking for an incoming sync
vector. When it finds it, it buffers up the physical packet as a set
of soft decision samples and sends it off to the PC in an Ethernet
frame for FEC decoding and upper layer processing.
Phil