Re: [Discuss-gnuradio] Introduction for GSoC18 participation

[Wunsch, Felix (CEL)]

Hi Luca,

thanks for sharing your proposal draft! Here are some remarks:

1. You list a MIMO channel model as deliverable. For slow, flat flading you can use (as you correctly stated) a simple matrix multiplication block. That, however, already exists in GNU Radio. Or were you thinking of a wrapper around that functionality for some specific reason?

2. Without wanting to overload you with additional deliverables, I think a differential STBC such as [0] would be a nice addition. As it's very similar to the Alamouti STBC it shouldn't be much work but it dispenses with CSI estimation at the receiver, which simplifies implementation and potential applications quite a bit.

3. I'm not too familiar with the OFDM equalizer that is currently in the tree, but you should try to use the existing code where possible.

4. You write that you want to integrate the different MIMO schemes into gr-digitals OFDM. What exactly will that integration look like in the end? I would love to see some ready-to-use flow graphs as examples and potential building blocks for (even) more advanced applications. You would need and build them during development anyway, so this should also not really add to your workload, but immensely help GNU Radio users who want to use your work!

5. From your proposal, I'm not really sure if you are planning to do only OFDM-based examples or also narrowband ones (you write something about developing MIMO training sequences and channel estimation before you speak about OFDM). Personally, I think you should limit the scope of your proposal to OFDM. Narrowband systems require a  totally different synchronization and that is just overhead for you. Of course, you should still try to separate your, e.g., STBC encoding and decoding from the OFDM processing in a way that makes it usable for single-carrier schemes (Right now, I'm thinking of simply accepting different vector lengths for the respective blocks).

6. There are some minor typo / grammatical / wording issues, it would be great if you could also take care of that.

Despite of all the remarks I think you wrote a very nice proposal!

Cheers,
Felix

[0] https://pdfs.semanticscholar.org/ba1d/ba8af2e3cee3213794fe722d13e90de5baac.pdf

---

Von: Discuss-gnuradio <discuss-gnuradio-bounces+address@hidden> im Auftrag von Luca Schmid <address@hidden>
Gesendet: Samstag, 24. März 2018 22:24
An: Sebastian Müller; Nicolas Cuervo Benavides
Cc: GNURadio Discussion List
Betreff: Re: [Discuss-gnuradio] Introduction for GSoC18 participation

 

Hi everyone,

@Nico @Sebastian, thank you for commenting on my first thoughts about a MIMO project for GSoC.

I just uploaded a draft of my proposal on github.

It would be really nice, if you could read it, comment on it and share your ideas and thoughts, especially about my milestones and the timeline, with me.

Best and thank you in advance

Luca

2018-03-22 19:03 GMT+01:00 Sebastian Müller <address@hidden>:

Hi Luca,

great to hear from you again!

MIMO is definitely a topic that deserves to be tackled by GNU Radio. Your idea seems well thought of and sensible (and of course your last year’s work speaks for itself). If I understand correctly, your proposed approach would only cover the MIMO-OFDM case. I would like to point out that other modulation schemes than OFDM might be of interest as well for some users, and therefore I think it is sensible to provide an easy extensibility of your work. It would be great if you could cover that topic in your proposal!

As Nicolas pointed out, you should have a first draft of your proposal online in just a few days in order to get more feedback on the mailing list before the deadline on March 27! I’m looking forward to hear more from you soon.

Cheers,

Sebastian Müller

address@hidden

PGP ID DC2AA3EE

Am 22. März 2018 um 11:10:37, Nicolas Cuervo (address@hidden) schrieb:

Hello Luca!

Thank you for showing interest in GSoC (once again! :) ) 

Your idea sounds very cool and also very useful for users that want to have a first approach to MIMO, and you did your homework checking where this idea fits in the tree, and propose reuse of code, which is fantastic. As you mention, the clock is ticking and, although this idea is potentially well structured, it does not have a mentor assigned yet, and that puts a bit of pressure. Have a look at this old mailing-list thread [1], which was somewhat in the same position as you. If you have already contacted someone to mentor your project, that would be great! But if not, then writing a proposal ASAP would help us assess your expectations and probably make easier for any of us to hop on board as the mentor, should that person feel capable of providing a solid support. You know the drill :) so I would recommend you to work on your draft and continue on this open discussion to keep the conversation alive.

Keep up the good work! Looking forward to that proposal draft ;)

Cheers,

Nicolas

[1] https://lists.gnu.org/archive/html/discuss-gnuradio/2014-02/msg00291.html

On Thu, Mar 22, 2018 at 1:19 AM, Luca Schmid <address@hidden> wrote:

Hi everyone,

I am Moritz Luca Schmid, a graduate student from Karlsruhe Institute of Technology (KIT). Since 2016 I am in touch with GNU Radio, mainly developing for the Communications Engineering Lab (CEL) in Karlsruhe, where I am working as an assistant researcher. In 2017, I successfully participated in GSoC'17 with GNU Radio, building a DAB/DAB+ Transceiver application[1][2], now also know as DABstep ;)

After GSoC'17, I have continued working on the DAB+ project, improving the code and introducing new features. As some examples, I added RTL-SDR devices as a possible signal source and currently I am implementing dynamic label and Media Object Transfer (MOT) support. I also rewrote the existing OFDM PHY layer and developed a more efficient and robust method for synchronization in my Bachelor thesis in fall 2017.

I am very excited about the idea of bringing a MIMO capability to the GNU Radio project. I read about the suggestion of a MIMO transceiver at the list of old ideas for GSOC, proposing to implement an OOT module (gr-mimo) that includes the basic encoding and decoding algorithms and therefore enriching GNU Radio with another basic telecommunication feature. To do so, it proposes an OFDM based phsical layer and the realization of MRC decoding and beamforming.

I like the idea of combining the proposed MIMO capability directly with OFDM. The combination of frequency and space diversity has shown very promising performance results and MIMO-OFDM is considered in a number of developing wireless standards. MIMO-OFDM in GR would therefore be a very valuable basis for all those, who are interested in these new techniques but don't want to build a whole communication system from scratch. 

The GNU Radio core module gr-digital already contains a solid OFDM implementation, including a complete transmitter and receiver with synchronization. (Actually I found multiple implementations (ofdm_receiver, ofdm_txrx, ofdm_mod/demod) including OFDM (de)modulation and synchronization (mainly with Schmidl & Cox but also other sync approaches). Some of them additionally include the digital (de)modulation/ symbol (de)mapping.)

In order to reuse this existing code for the MIMO idea, I propose the inclusion of a (of course optional) MIMO capability in gr-digital's OFDM transceiver. In GRC view, I am thinking of an additional parameter for the hierarchical blocks of the OFDM transmitter and receiver in form of a drop down menu, stating SISO transmission mode as default option, but also listing some MIMO possibilities. 

Talking of specific MIMO capabilities, I would mainly focus on the implementation of the most popular algorithms, e.g. an Alamouti 2xN and V-BLAST. Together with a loopback and over the air example (I would propose USRP B210s for the start) and a cool demo, this could be an attractive feature for many MIMO interested people.

Together with a unified interface and extensive qa tests, the MIMO feature fits well into gr-digital, instead of a separated OOT module.

I know, that I am very late in proposing this idea for GSoC. Please let me know, what you think of this approach for a possible GSoC'18 project and share your ideas and comments.

Best

Luca

[1] GSoC'17 blog about DAB/DAB+ Transceiver application

[2] gr-dab github repository

_______________________________________________
Discuss-gnuradio mailing list
address@hidden
https://lists.gnu.org/mailman/listinfo/discuss-gnuradio

_______________________________________________
Discuss-gnuradio mailing list
address@hidden
https://lists.gnu.org/mailman/listinfo/discuss-gnuradio

_______________________________________________
Discuss-gnuradio mailing list
address@hidden
https://lists.gnu.org/mailman/listinfo/discuss-gnuradio