Re: GsoC Proposal Feedback

[Hamza Mohammed]

No worries, I am glad you were able to reply. I just want to ask a few
questions before editing my proposal.
1. So the idea was to explore multiple open source error correcting
libraries and use only one of them in the gr-fec module as an optional
dependency or OOT module?
2. The goal of the project will be to develop a test bench for the
existing blocks in GNU Radio in comparison to other open source libraries ?
3. Is the project size Large or Medium ?
4. Do I need to explain the code in detail of how I will integrate the
chosen library to the gr-fec package?
5. What should my proposal mainly focus on ? The integration of the library
or How the test bench is set up?
>From my understanding of the previous mail the project's main objectives
are: 1) Introduce modern FEC codes to the gr-fec module. 2) Provide a test
bench for error correction codes in the gr-fec package 3) Provide library
based (optional) additional implementations for code blocks.

Sorry for my poor spelling errors in my previous proposal, I will get them
fixed asap. It's GNU Radio noted!!

Re: GsoC Proposal Feedback

[Andrej Rode]

Hi Hamza, 

sorry for the late reply/not giving you feedback earlier. I’ve been travelling 
in the last week(s) and it slipped my attention. 

Your proposal seems to already go a lot into details how to include the AFF3CT 
library in GNU Radio in terms of pointers etc. 
It seems to me that your plan would be to fully replace existing 
implementations with AFF3CT. I would recommend to either add them
in a GNU Radio OOT first or add them side-by-side to existing 
Encoders/Decoders, as overall we might opt to make AFF3CT an optional
dependency and also it would aid in benchmarking (not only for you, but also 
other community members).
With this idea on the ideas’ list we also meant to give some opportunity to 
find other error correcting libraries, e.g. https://github.com/xdsopl/LDPC 
which show good performance and may be directly integrated in GNU Radio. 

In terms of timeline I see a small discrepancy, since you say that you will 
spend 30 hours each week, but you can “only" dedicate 3-4 hours per day. 
If you would work 7 days a week (which I do not recommend), will work out at 
21-28 hours a week. 
For the scheduling of the different tasks, I feel that in the beginning you 
left yourself a little bit of room, but then you put the Delivery 1 with moving 
LDPC,TCP,Polar in a timeframe of two weeks, which seems to be quite challenging 
to do in this short time. I guess in case you finish earlier with the 
repetition en/decoders you can already start on the blocks mentioned in this 
delivery.

In terms of reporting, in the past our students wrote short but concise weekly 
blog posts which gave a good turnaround time for feedback from the community. I 
would recommend
the same to you. You can publish blog posts directly on our webpage, which is 
created with the static website generator “hugo” here: 
https://github.com/gnuradio/hugo-website

Some small comments on the form: 
 - You have a lot of typos/transposition errors in your proposal, I recommend 
running it through some error checking
 - It is spelled “GNU Radio” - not gnu-radio, or Gnu radio.

Cheers,
Andrej


> On 28. Mar 2024, at 17:50, Hamza Mohammed  
> wrote:
> 
> Hi all,
> Please take a look at my proposal for the project "Forward Error Correction", 
> that I attached below. Do you think the way I am planning to integrate the 
> AFF3CT library in the gr-fec package is efficient ? I think it is because my 
> implementation will not change the overall FEC API.
>