On 17 Jun 2004 at 6:46, Dave Tweed wrote:

> Mark Jordan <mark@CPOVO.NET> wrote:
> > I have to decode a 100-bit serial data stream received via radio.
> > The original data is FEC encoded, but passes through a /XOR with
> > itself before going to the radio modulator. The /XOR is between the
> > current bit and the previous bit in the stream.
> > That /XOR at the transmitter seems to be stupid. A NRZI conversion
> > would do lots better there.
>
> Or a proper data scrambler, depending on what the intent is.
>
> Usually, long-distance communication systems (both wired and wireless) want
> to randomize the statistics of the data bits so that timing recovery is
> more reliable. There are commonly-used data scrambling circuits (anywhere
> from 2 to 16 bits long) used for this purpose. The corresponding
> descrambling logic at the receiver end has well-known characteristics with
> regard to error propogation, and the FEC algorithms are designed to take
> this into account.
>
> The /XOR gate basically produces a pulse whenever the input data changes
> state, so if a pulse is created or lost because of noise in the channel,
> bits from there to the end of the frame at the output of the decoder will
> be flipped.
>
> > So, I can decode the frame correctly at the receiver side, but when
> > an error occurs, just one wrong bit is sufficient to trash the whole
> > frame due to that /XOR at the transmitter. In case two bits are wrong,
> > the FEC sometimes can correct the error.
> >
> > As I'm not allowed to modify the transmitter side, I'm looking for a
> > clever way to decode such a frame.
>
> No, with such a terrible transmitter implementation, you'll have to do an
> exhaustive search in order to decode the frame. For each bit position in
> the frame, try flipping the bits from there to the end and see whether the
> FEC can make sense out of it. If not, try the next position.
>
> Obviously, you need to have an FEC implementation on the receive side that
> can handle up to 100x the bandwidth of the channel. That cheap /XOR gate is
> basically negating most of the benefit of all that expensive FEC logic at
> both ends of the link. You really need to argue strenuously for a change.
> If you can't delete the /XOR gate itself (is it embedded in the radio
> module?), perhaps you can insert some logic upstream of it to cancel out
> its effect.
>
> Your only other option is to hope that the frames are transmitted
> redundantly and that it's OK to simply drop the bad ones altogether.
>
> -- Dave Tweed
>

        Would you believe if I tell you that this system was designed by
Motorola?  It is the old MDC1200.

        Some details here:  http://www.batlabs.com/mdc1200.html

        Thanks,
        Mark Jordan

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email listserv@mitvma.mit.edu with SET PICList DIGEST in the body