Re: Viewing other signals in OFDM

[Marcus Müller]

Hi Sourya,

> However, I want to attempt to view signals in OFDM receiver which are being sent using 
a non-OFDM transmitter.

how is that supposed to work? An OFDM receiver necessarily needs to synchronize to the 
OFDM transmission.

Without an OFDM transmission, there's nothing to synchronize to.

A working OFDM receiver will hence simply *not* receive something outside of the 
possibility of mis-detecting something as OFDM that isn't, and in that rare and probably 
very intermittent case, what you then get would be totally random nonsense.

You can throw out the OFDM part and just use the DFT to estimate the spectrum, but that's 
just that: the spectrum of the reception, not giving you much to go on about the frequency 
responses of the channels observed:

> What I actually want to see is the frequency response of the signals on the subcarriers 
on the receiving side.

Without OFDM, no subcarriers exist. So, we can talk about the frequency response of the 
channel, if you like. Yes, you can do *blind* channel estimation (i.e., estimating the 
channel without knowing characteristics of the signal transmitted over said channel), but 
it's not going to be possible with an OFDM receiver.

> What would be the right way to move forward with this?

Honestly: forget about OFDM, and learn about *blind channel estimation*.

In multiple of your previous emails (including this) I had the feeling that, although I've 
pointed you to OFDM literature, you at best skimmed that and never developed a 
mathematical understanding for why you use it, how it works, and why it's implemented that 
way.¹

Channel estimation is a more mathematical problem, and you won't build something useful 
without having a solid understanding of the fundamentals; you need books at this point, 
not software:

I'd personally re-read the basics chapters of Kay, Steven: Statistical Signal Processing 
Vol 1: Estimation Theory, because that's what some of the material that I learned with was 
based on, then followed by the famous Kailath report on sampling time-variant filters 
(which fits the channel model you'd use OFDM in, anyways), but I think this might not fit 
you well. However,

D. Manolakis, V. Ingle, and S. Kogon, Statistical and Adaptive Signal 
Processing. 2005.

might be more of your way! You'd need to understand what an equalizer is, in detail, why 
you need one, and what the problems estimating the channel for one are. But these are the 
same basics you would have to learn to really understand why you're doing OFDM, so you 
need to read that, anyways.

---

So here comes the part were I advise you if we'd be working at the same institute. You 
seem to be doing a thesis on this, so:

Write an email to your advisor. Advisors are typically people that are busy, and have to, 
but also want to, supervise students. Their least favorite kind of student is the one that 
doesn't come and ask questions and delivers no results; their favorite kind of student is 
the one that comes, asks good questions that help the advisor understand their own 
research better and help refine the goal of the thesis the student is doing, and then in 
the end deliver interesting results. I've myself been the question-answering, 
does-more-on-other-student's-projects-than-he-should kind of student, back in the day; I 
leave the assessment of my results up to my former advisors. I learned a lot during my 
theses! It does mean reading stuff you don't understand the first time around, noting down 
questions, trying to answer these yourself, dropping by your advisor's office occasionally 
(or them dropping by at your desk), getting the rest answered or categorized as "that's 
not relevant to your thesis right now".

So, write an email to your advisor. Start it with a short description of what you think 
your current task is, and a sentence on why you think that. Just forward this email to 
them and say that while you were researching (and I swear your advisor will love that 
you've been researching and talking to experts on your own!) this grumpy guy on the 
internet told you that an OFDM receiver is not what you need, that a blind channel 
estimation / blind equalization approach is probably what you need, and had literature 
recommendations. Say you're attaching the email, because you want to discuss the right way 
to move forward.

I say this because your question,

> What would be the right way to move forward with this?

is one that makes advisors very happy. You've researched stuff, you hit an obstacle, you 
identified the obstacle, and you wonder where to go next. Helping you choose where to go 
next is *literally* the advisor's job. And it's also *literally* in their own interest, 
because they *want* you to succeed in your thesis because ideally, your results help them 
understand their own topics better.

Best regards,
Marcus

¹ per my mail from the 13th of March:

> If you have a good basic education in digital communications, Proakis' "Digital 
> Communications", I think starting from 4th or 5th edition. In the third edition (which 
> can be bought very cheaply used), it's called "An FFT-based Multicarrier System",  but 
> it's really OFDM, just without the name. Either way, that chapter would have explained 
> the same as I did above!
>
> If you need a slightly gentler entry to OFDM (in the lectures on digital communications 
> that I've been part of, we treated it as something rather advanced, because explaining 
> *why* it is like it is requires understanding of a lot of channel concepts), you would 
> probably *not* want to rely on Proakis.
>
> Instead, you could try the free "Software-Defined Radio for Engineers" by Travis Collins 
> et al.,
> https://www.analog.com/en/resources/technical-books/software-defined-radio-for-engineers.html

On 25.03.24 18:42, Sourya Saha wrote:
> Hi.
> I am working with OFDM. I know that OFDM receiver in GNURadio needs a lot of parameters 
> that are similar to the transmitter side in order to function. However, I want to 
> attempt to view signals in OFDM receiver which are being sent using a non-OFDM 
> transmitter. What I actually want to see is the frequency response of the signals on the 
> subcarriers on the receiving side. What would be the right way to move forward with this?