[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Discuss-gnuradio] CFO in OFDM

From: David Halls
Subject: Re: [Discuss-gnuradio] CFO in OFDM
Date: Wed, 29 Oct 2014 21:19:16 +0000

Thanks Martin,

I dug out the S&C paper and think I'm clear on that, thanks. 

I have a currently uncoded system with superimposed QPSKs from multiple sources so it's pretty sensitive! its fine using the Octoclock, though giving 0 BER end-to-end. I also have 32 symbol packets; so fairly long. I've implemented some basic CFO tracking using the pilots which helps.

I think it's residual CFO error causing the issue,  manifesting as rotation of frequency over time, but as it's high SNR this should be small. With tracking and correction the constellation *looks* OK...

I'm going to keep digging around...

-------- Original message --------
From: Martin Braun
Date:2014/10/29 20:38 (GMT+00:00)
To: address@hidden
Subject: Re: [Discuss-gnuradio] CFO in OFDM

On 10/29/2014 01:34 PM, David Halls wrote:
> I am suffering from issues related to residual CFO error manifesting
> itself as a slight rotation over time. This can be partially corrected
> in the FD using tracking pilots which I have implemented but there is
> still a slight problem which may be caused by ICI.
> In the OFDM example code, how is the value for the Frequency Mod
> sensitivity (-2/FFT_len) derived? Has anyone implemented other CFO
> estimation blocks for OFDM systems, other than the Schmidl and Cox
> technique?

I recommend the S&C paper for the derivation, but the maths is not that
hard. The very short version is that 1/fft_len is the frequency offset
of carriers in normalized frequency, and the fine frequency offset is
unambiguous between \pm 1 carrier. The result of the estimator is thus
in 'fraction of carriers in normalized frequency'.

I've played around with other techniques, but not in GNU Radio. The
reason being, a tracking CFO estimator (i.e. one that changes the
estimate over time during a packet) is much harder to implement (read, I
was lazy). Now, if you don't actually want to track, but only estimate
once at the beginning, you won't get much better than S&C.

Note that you'll always have ICI as soon as you have a Doppler spectrum.
With decent equalization, it's not *that* big of a deal as long as it's
not too much ICI.


Discuss-gnuradio mailing list

NOTE: The information in this email and any attachments may be confidential and/or legally privileged. This message may be read, copied and used only by the intended recipient. If you are not the intended recipient, please destroy this message, delete any copies held on your system and notify the sender immediately.

Toshiba Research Europe Limited, registered in England and Wales (2519556). Registered Office 208 Cambridge Science Park, Milton Road, Cambridge CB4 0GZ, England. Web: www.toshiba.eu/research/trl

This email has been scanned for email related threats and delivered safely by Mimecast.
For more information please visit http://www.mimecast.com

reply via email to

[Prev in Thread] Current Thread [Next in Thread]