[Discuss-gnuradio] Approaches for cross-spectra

[Marcus D. Leech]

I'm working on a continuum interferometer project (actually resurrecting 
one that has been dormant for several years).  I'm currently
  playing with approaches for spectral estimating with a two-antenna 
interferometer system.  I think, ideally, since you have two antennae,
  you'd like to take advantage of that for producing spectral estimates 
that have channel-electronics-induced-spurs reduced in magnitude,
  and the spectral components that are common to the two antennas 
enhanced.  I've tried various approaches:

    o Add the magnitudes of the FFTs for two antennae, low pass filter that

      This produces a *composite* spectral estimate, including the 
components due to independent spurs on the two channels.

    o Complex multiply the two FFT outputs, then compute the magnitude, 
low-pass filter that

       This seems to produce some suppression of distinct components 
that aren't common to the two sides
       The absolute magnitudes of non-suppressed components are 
obviously larger, due to the multiply.

The second approach seems very similar to what ALMA is doing for their 
FX correlators.

Any words of wisdom/advice?

--
Marcus Leech
Principal Investigator
Shirleys Bay Radio Astronomy Consortium
http://www.sbrac.org