Hi guys,
First, let me introduce myself. My name is Mark Priddy and I've recently
graduated with a bachelors degree in engineering. During my time at the
university, I took a class on communication theory. For the project in
this class, my classmates and I worked on looking at how we could
improve fldigi while we were learning about communication systems.
This brings me to why I'm sending this email. My professor in the class,
Rob Frohne, recently forwarded me an email from this mailing list about
interest in improving the RTTY code. I thought this might be a good time
to mention my (and Frohne's) work that we've done in improving the RTTY
code.
So far, in a rather old version now, we've created a matched filter for
the RTTY code and created a maximum a posteriori detector which uses
statistical analysis to determine whether the received letter was really
what was sent. We've also worked out a way to determine when the most
optimal time to to sample the bits in the signal, using statistics based
off of the current transmission. This is currently what I'm working on
implementing now.
The matched filter is the biggest improvement on output quality while
the maximum a posteriori detector makes roughly a 10% improvement
correcting a few more letters than it incorrectly "corrected." While I
don't remember the specifics of a few test cases we did I do remember
that the improvement was quite impressive; running the same signal
through both the improved and the original fldigi we found that the
original code made the signal completely unreadable while the matched
filter code was still readable. (Frohne, do you remember the specifics?)
While a lot of the code is done, I still have yet to polish the code and
integrate it with the interface and with the latest version of fldigi.
Since it has been almost a year since I started work on this, I haven't
looked at the latest fldigi code. Has any of this been implemented already?
Anyway, I'm nearing completion on this and I hope to be able to submit
my code to the main developers soon! I thought I should let people know
what I was working on to minimize duplication of work. Hopefully someone
will find the code to be useful.
Mark
KE7UFF
_______________________________________________
fldigi-alpha mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/fldigi-alpha
