Hey GNU Radio’ers!
Another week has passed. Thus, it’s time for my next project update.
The week started with a fairly good event. I found a bug in the
decoder which led to incorrect results.
I studied various papers and algorithms for decoding performance
improvements. Especially Successive Cancellation List Decoding SCLD
caught my attention. Also, papers exist which describe vector
optimization for x86 CPUs and SIMD utilization. I will get back to
this once I’m working on VOLK optimization later in the project.
In order to have meaningful codes available, I started working on some
channel construction code. First of all, BEC channels can be
calculated efficiently, but I don’t consider them to be interesting
for GNU Radio. Let me know if you have a different opinion on this.
And why. I consider BSC as the most interesting case.
There exist algorithms which approximate Z-parameters or channel
capacities quite well. Also, Monte-Carlo simulations yield usable
results. I’ll stick with them for now and will work on more
sophisticated algorithms later. Just as outlined in my timeline.
Next week will bring some C++ coding. Encoders and decoders are
waiting to be implemented in C++ for FECAPI.
More info and current project progress can be found in [1] and [2].
Cheers
Johannes
[1] GitHub - jdemel/gnuradio: GNU Radio
[2] GitHub - jdemel/socis-2015-gr-results: ESA SOCIS 2015 files and information