Re: [Discuss-gnuradio] FLL Band-Edge Detectors: Literature?

[Tom Rondeau]

On Tue, Oct 2, 2012 at 9:01 AM, Martin Braun (CEL) <address@hidden> wrote:
> On Mon, Oct 01, 2012 at 05:17:33PM -0400, Tom Rondeau wrote:
>> Yes, harris' book is the best to start with. There is another paper
>> from him called "Let's Assume the System is Synchronized" that also
>> goes over it. I'm not sure if he's published a paper that discusses
>> the specifics of the filter derivation, yet, though. It's based on the
>> derivative of the half cosine waveform of the  RRC filter rolloff. The
>> system behaves much better this way than just generating any random
>> band edge filter.
>>
>> In theory, this should work for any signal using an RRC pulse shaping.
>> For specific constellations, you could use a Costas loop with a wider
>> lock in bandwidth to handle the frequency offset.
>>
>> Oh, and I might be the only one who calls this the "FLL band edge
>> filter" specifically to point out that this is only one possible
>> implementation of an FLL for coarse frequency tracking. Other
>> algorithms are welcome :)
>
> Thanks for these pointers. For future reference (and the mailing list
> archives):
>
> * The paper "Let's Assume the System is Synchronized" is a good
>   introduction and provides a great and simple explanation.
> * Harris' book "Multirate Signal Processing: for communication systems"
>   is the more complete version, down to polyphase implementation
>   details.
> * I should have realized this is a way to implement an ML Frequency
>   correction for RRC spectra without a-priori knowledge. This is more
>   generally derived in "Synchronization techniques for digital
>   receivers" by Mengali/D'Andrea[1], and probably many other places.
>
> M
>
> [1] In general, Fred Harris books and papers are more fun to read,
>     though.

Yeah, Mengali's great, but dense and mathematical. harris provides a
more big-picture, system's oriented look at the problems.

Tom