Re: [Discuss-gnuradio] Radio Astronomy GSoC 2017 group possible?

[Martin Braun]

Glen,

as the GSoC guy I have a couple of comments. But first of all, I'm glad
there's interest from more radio astronomers in GNU Radio -- as Marcus
points out, there's already a couple of RAs using GNU Radio, and you'll
find a bunch of nice experiments out there.

On 02/07/2017 01:20 PM, Glen I Langston wrote:
> Hello GNU radio folks,
> 
> A few radio astronomer friends have had a very active interest in GNU
> Radio, but
> I’m aware of relatively few Radio Astronomy oriented contributions to
> GNU radio.
> 
> This email is a request to start a discussion on
> some requirements of Radio Astronomy and the software support they would
> need.
> 
> The main GNU Radio enhancement items on my short list are:
> 1) Averaging of spectra for long periods (minutes to hours), while
> capturing every spectrum.
> 2) Writing average and transient spectra based on internal and external
> events.
> a) Maybe this already exists, but a spectrum message is needed so that
> averaging can be separated from writing.
> b) Transient event detection with spectrum (or time sequence) passed to
> a writing thread.
> c) When sudden increases of signal are noted, time sequences would be
> written. (When auto-detected).

I would guess that most of these are already possible in GR; they might
be put into a more accessible form.

> 3) Keeping tracking of information associated with the observing setup.
>  There are large numbers of ancillary data
> values needed to calibrate and map spectral observations (geographic
> location, precise time, horn/antenna azimuth, elevation
> gains, device types used for the observations, flags to indicate
> calibration spectra etc).

We've just started work on a file format we call "SigMF", which we plan
to integrate into GR, and would solve these issues.

> I’ve greatly appreciated the GNU Radio software and excellent quality of
> the GRC and all the code I’ve seen.
> I’ve extensively modified the ‘FFT sink' to optimize for averaging and
> added a write component inside that code.
> Writing inside averaging is probably a mistake, as writing suspends data
> collection for a short time.  I need to learn
> how to bring my code up to the GNU Radio quality standards etc and put
> the existing code in GNU Radio distribution.
> 
> Further, can we add a spectra message type in GNUradio so that spectral
> can be passed to different blocks?

Spectra are real- or complex-valued vectors, and we already have that type.

> To show that good progress has already been made, but still needs
> quality integration into GNU Radio,
> three figures are attached. Using an AIRSPY (10 MHz bandwidth) and GNU
> radio, I’ve mapped the Milky Way Galaxy in Neutral 
> hydrogen (1420.406 MHz). It would be great if we can get this
> functionality, with a few significant 
> enhancements, into the standard GNU Radio release.

It sounds like the best way forward would be to consolidate these things
into a radio astronomy OOT; I believe Marcus already has at least one of
those.



So, for GSoC, we'd need a more concise and specific project idea. In
particular, we'd need a better survey of what already exists (and what
not), and phrase this as a good project that can sensibly be done over a
period of 3 months.

Cheers,
Martin