(GSoC) MIMO stuff in GNU Radio

GNU Radio
1

Hi all,

I’m Zhou, a graduate student now. I’m pretty interested in MIMO systems,
especially Multiuser-MIMO, since it can increase the total throughput of
the network significantly. But to the best of my knowledge, there seems
no
related standard blocks (gr-mimo or something) to achieve this in GNU
Radio, few tutorials or examples are available either.

Currently, I’m working on a project to build a Multiuser-MIMO system
based
on OFDM. The basic requirement of this project is to implement
Maximal-ratio combining (MRC) and Beamforming on both uplink and
downlink
sides. I notice that there existed an old GSoC project named “MIMO
stuff”
on the GNU Radio GSoC webpage a few months ago, but it is disappeared
now.
So I am wondering if “MIMO stuff” or some related work could be a part
of
GNU Radio GSoC 2014? If so, what kind of work it would be? Looking
forward
to get more details on it!

Best regards

Zhou

2

On 02/18/2014 05:14 PM, YiZiRui Z. wrote:

Maximal-ratio combining (MRC) and Beamforming on both uplink and
downlink sides. I notice that there existed an old GSoC project named
“MIMO stuff” on the GNU Radio GSoC webpage a few months ago, but it is
disappeared now. So I am wondering if “MIMO stuff” or some related work
could be a part of GNU Radio GSoC 2014? If so, what kind of work it
would be? Looking forward to get more details on it!

I could – depending on whether or not we find a mentor suitable. In
any case, it would be up to you to write a good project proposal. You
would have to detail what exactly you would be working on for 3 months,
how it benefits the GNU Radio project etc.

MB

3

On Wed, Feb 19, 2014 at 4:18 PM, Martin B.
[email protected]wrote:

based on OFDM. The basic requirement of this project is to implement
how it benefits the GNU Radio project etc.

MB

Discuss-gnuradio mailing list
[email protected]
Discuss-gnuradio Info Page

Hi Martin,

Thanks for your advice.

Although it is not so clear, I have some elementary ideas on this
project
and I’m on it now. On the other hand, I will keep an eye on the GSoC
2014.
Thus, I want to know is there any way to discuss these ideas with
potential
mentor(s)? How do I know if anyone is interested in it? Anyway, I’ll go
on
with this project no matter whether it is accepted or not. Of course, If
it
could be involved in GSoC, that would be great.

Regards

Zhou

4

On 02/20/2014 04:01 AM, YiZiRui Z. wrote:

Hi Martin,

Thanks for your advice.

Although it is not so clear, I have some elementary ideas on this
project and I’m on it now. On the other hand, I will keep an eye on the
GSoC 2014. Thus, I want to know is there any way to discuss these ideas
with potential mentor(s)? How do I know if anyone is interested in it?
Anyway, I’ll go on with this project no matter whether it is accepted or
not. Of course, If it could be involved in GSoC, that would be great.

Nice to hear!

DISCLAIMER: We have not been accepted (yet?) as mentoring org for GSoC.
This is all speculative!

To clarify, our mentors are all volunteers, which means we can’t assign
any mentor a project if he or she doesn’t like it. This means that
project proposals might get declined, even if they’re very good, simply
because we can’t find a mentor. Also, not everyone can sign up as
mentor, which means that the mentor will always be limited.

On the other side, I do not want to punish people like you, who are
being proactive and creative, which is a key feature we are looking for
in students.

That said, we have never, ever been in the situation where we had to
decline a student because of the proposal content. Last year, the top 5
proposals were all for projects that already had a mentor. A really
great proposal would probably be easy to sell to a mentor, because it
reduces the risk of having to micromanage a lot (read: we prefer good
proposals for new ideas rather than bad proposals for or own ideas).

You’re in a bit of a chicken-and-egg situation, here: You need a mentor
to discuss the proposal with, but you first need to get someone excited
enough to potentially be a mentor.

Here’s my suggestion: Read all there is on GSoC wiki pages on GNU
Radio, including old proposals (but don’t follow them to the letter,
just use them as inspiration). Whip up something, and post it on the
list. You’ve already got something going for you: You’re early.

Now, as with all list traffic, there’s no guarantee for you that someone
will help you. But sticking to list netiquette helps a lot.

Martin

5

Hi Zhou - Can you post some links to relevant papers / research, for
folks like me to do a quick look at? I studied MIMO just a little bit,
mostly from an IT perspective, not a practical one. But, I’ve read
through a lot of papers (on many subjects) and would be interested in
knowing more about practical MIMO. I encourage you to get your project
up on the GSoC14 GR page if it is not already there … who knows, I
might be willing to mentor :slight_smile: But, I’d like to see some papers /
research first, as I’m sure others would too. - MLD

6

On Thu, Feb 20, 2014 at 3:24 PM, Martin B.
[email protected]wrote:

not. Of course, If it could be involved in GSoC, that would be great.
mentor, which means that the mentor will always be limited.
proposals for new ideas rather than bad proposals for or own ideas).
Now, as with all list traffic, there’s no guarantee for you that someone
will help you. But sticking to list netiquette helps a lot.

Martin

Discuss-gnuradio mailing list
[email protected]
Discuss-gnuradio Info Page

Hi Martin,

Thanks for your patient reply. Your words really inspire me.

I am writing a block to implement MIMO transceiver based on “ofdm_rx”
and
“ofdm_tx” now, willing to build a MIMO WLAN with several USRPs.
Basically,
general MIMO encoding and decoding method will be included, some
techniques
like rate adaptation is also in consideration. But those ideas are not
so
complete now, next I will try to figure it out. As you said, posting the
idea on the list is a good way to make it known to all, I will do that
when
everything is ready. Also, I will go through the GNU Radio GSoC wiki
page,
maybe there are some other good things I can do.

I know that if there is no mentor to guide me, I would not be able to
take
part in the GSoC. But, maybe this is not the most important thing to me.
There is no doubt that I can benefit a lot from working on the project
and
learning from the mailing list, these are the things that shine.

Thank you again

Zhou

7

On Thu, Feb 20, 2014 at 9:15 PM, Michael D.
[email protected]wrote:

Thanks for your patient reply. Your words really inspire me.
I know that if there is no mentor to guide me, I would not be able to
take part in the GSoC. But, maybe this is not the most important thing to
me. There is no doubt that I can benefit a lot from working on the project
and learning from the mailing list, these are the things that shine.

Hi Michael,

Glad to hear from you.

To be frank, MIMO is not a novel thing, since it has been proposed for
many
years. In the past, most research on MIMO stay in theory due to the
limitation of signal processing technology. But things are different
now.
Since 2009, when IEEE passed the 802.11n draft, study on practical MIMO
surges. Now, it becomes more prevalent.

There are many research papers on MIMO, many of them are published in
Mobicom or Sigcomm, which are all top conferences on Computer Network.

Among them, I think “802.11 with multiple antennas for dummies” is a
good
guide for someone who are not so familiar with practical MIMO systems.
This
is the link.

A paper “Rate adaptation for 802.11 multiuser mimo networks” implement
MIMO
with GNU Radio and USRP, but it focus on uplink transmission only. Also,
the details on how to build MIMO systems with USRP is not illustrated,
maybe you will be interested on how it works too. This is the link.

Other works are not list here for brevity. By the way, some studies on
MIMO
are based on Sora, another SDR platform.

http://research.microsoft.com/en-us/projects/sora/

In my opinion, MIMO systems will be more ubiquitous in the future. But
it
seems that there is no relevant standard block (gr-mimo) to achieve this
in
GNU Radio yet. So I think maybe there is something I can do.

Best wishes.

Zhou

8

On Thu, Feb 20, 2014 at 11:10 PM, Michael D.
<[email protected]

wrote:

should be concise yet complete and realistic in what you think you can

Glad to hear from you.
Among them, I think “802.11 with multiple antennas for dummies” is a

Hi Michael,

As you said, although there is a pile of research papers on MIMO,
tutorials
and examples on how to build MIMO system are limited. Many researchers
are
not willing to release their source codes for the sake of maintaining
secrecy. But there do exist some works on building practical MIMO
systems,
I will try to get these ready.

When the preparation gets done, I will try to add my proposal to the GR
GSoC website in the following weeks and keep in touch with you / the
list.
Wish to get valuable advice from you.

Thanks for your encouragement.

Zhou

9

Hi Zhou - I think it would be great to see how far GNU Radio can be
pushed with respect to MIMO processing. With advances in both
algorithms and multi-core CPU raw processing capabilities, the added
computational requirements for MIMO might be realizable at some
reasonable level (IIRC, they were not [for commodity processors] back in
2007-2009 when I looked into this last). I agree that MIMO is in
wireless networking’s future; the big question is how much anyone can
accomplish in a few month’s work towards getting even a basic MIMO
system working (within GNU Radio, in this case; but, really, any
platform that does not already provide such capabilities).

Your paper / research links are fine, though there must be a few more
which are more seminal works on “how to do practical MIMO” which are
concise and basic.

SORA is a non-starter for many researchers; compare the number of papers
on or using SORA to those using GNU Radio …

The GR GSoC14 website is <
http://gnuradio.org/redmine/projects/gnuradio/wiki/GSoC > and I
encourage you to add your proposal to the “Signal-Processing” section.
I think you should be concise yet complete and realistic in what you
think you can accomplish in a few months of work on this topic – for
example, you’re not going to be able to implement 802.11n but you can
probably come up with some basic MIMO using OFDM PHY layer concepts
which would prove useful and interesting (taking in multiple data
streams, then using MIMO to do encoding, then Tx them to a receiver,
which does Rx, decoding, and pushing out multiple data streams; with or
without feedback). MIMO can be very complicated, so I’d encourage you
to keep the scope reasonable (yet truly MIMO). Let me / the list know
when your part is up. - MLD