|
From: | Marcus Müller |
Subject: | Re: [Discuss-gnuradio] Regarding GSoC'17 |
Date: | Sat, 17 Dec 2016 12:24:35 +0100 |
User-agent: | Mozilla/5.0 (X11; Linux x86_64; rv:45.0) Gecko/20100101 Thunderbird/45.3.0 |
Hi Kartik, exciting! So, packetized/bursty communication like bluetooth is really of
high interest for GNU Radio currently. In fact, in the upcoming
next release, there's been pretty much a complete overhaul of our
packetized communication ways. So if we throw that at you, you'd
have to learn two codebases – the current master/releases branch,
which works, and the next branch, which is still highly
fluctuating. I'd avoid that at first – it's probably better to
learn something that is already widely tested and then learn the
things that changed as soon as you've got a feeling for the stable
codebase. Since you've already worked with GRC and python, I'd recommend going through the Guided Tutorials on http://tutorials.gnuradio.org ; The first chapters will be a quick read (since you've been in touch with the python code GRC produces, it seems), but it's still a very worthy read – there's introductions of quite a few core concepts that you need to understand to fully make use of the later chapters, which explain how to write your own blocks in Python and C++. > I can get familiar with essential part of code base within 1-2 days I think you're overestimating the tidiness of the GNU Radio code
base :) I still think the idea of getting familiar with the code base is a great one, even if you didn't do it for the love of GSOC! If you haven't gone through the tutorials yet, I'd recommend you do that now.After that, you might have a clearer idea of what you'd love to do – after all, GSoC is a pretty mutual thing: you get the support and money, and GNU Radio gets your code, and what's often even more important, maybe a new lasting member of the developer community. Thus, whilst it's normally the job of the projects to come out with reasonable, realistic GSoC ideas that students can orient their applications on, I think it's just fair to allow you to pick anything that you feel like would be worth doing with your expertise – and help you even before GSoC starts. To be honest, it's always pretty hard to hold us back when someone enthusiastically announces whatever they're doing with GNU Radio on the mailing list, and asks good questions, and we surely won't exclude you from that willingness to help you with your "learning" project just because you want to do GSoC. In my head, there's ideas based on the fact that you know ns-3 (which I don't). I could imagine you like or don't like some type of projects, be it "adapters" between different frameworks (what if ns-3 could pass events to GNU Radio, which then generates a signal, which then is mixed on a real or simulated channel, to verify ns-3 event-based simulations? That'd be a huge effort, but also pretty rewarding. On the other hand, interfacing complex frameworks is often kinda menial work, so you would probably only do a small part of this during GSoC) or even just things like simulating the GNU Radio scheduler in ns-3, or maybe porting a feature that ns-3 has that you sorely miss in GNU Radio? Maybe you'd like to see yourself more as a personal area network (including BLE) expert – in which case I'm sure greatscottgadget's gr-bluetooth could use a reworking for the new packet radio structure (+ Bluetooth 4.0) I mentioned above, or you tackle gr-ieee802-15-4, which provides the PHY layer for IEEE802.15.4, and build things like 6LoWPAN atop (I think this would be possible – not sure, gotta ask Basti). So many opportunities! Generally, tinkering first is totally OK, and even necessary. Read the tutorial, make a block that does something simple (re-implement GMSK, bits in, signal out), find out what you like/don't like :) Best regards, Marcus On 17.12.2016 05:51, Kartik Patel
wrote:
|
[Prev in Thread] | Current Thread | [Next in Thread] |