LG001 Lare aperture Experiment to Detect the Dark Age (LEDA)
Greenhill, L., 

The Large-Aperture Experiment to Detect the Dark Age (LEDA) project
targets a frontier challenge in observational cosmology,
characterization of physical conditions in the intergalactic medium
(IGM) during the transition from the Dark Age to the Epoch of
Reionization, when the first generations of stars lit up the
Universe. The 21 cm transition of atomic Hydrogen is a unique tracer
of the IGM in the early universe. Ongoing projects seeking to detect a
reionization signal in the 21 cm line focus on redshifts of
6–10. Results will be filtered through models accounting for the
complex astrophysics of star formation. LEDA focuses on higher
redshifts, z~20, at which pockets of star formation were small, and
the IGM was more homogeneous and readily modeled. LEDA aims to infer
the source population at the end of the Dark Age and the initial
conditions for reionization, via measurement of the spectrum of
sky-averaged (total) power in the 21 cm line at z~20.

The peak signal at z~20 is anticipated to be O(10x) stronger than that
for reionization and in principle less challenging to detect than
power spectra of fluctuations on the sky. Nonetheless, low-frequency
total-power measurements are difficult. They require accurate
calibration of instrument gain patterns and the horizon-to-horizon sky
brightness distribution, which is dominated by bright radio galaxies
and diffuse galactic synchrotron emission. This makes calibration
difficult for single dipoles, but when incorporated into large arrays,
gain patterns and sky brightness can be estimated separately with high
precision, foreground emission can be subtracted away, and redundancy
in measurement of total-power spectra tightens control over

The LEDA instrument comprises (i) a large-N correlator deployed to an
operational array (Long Wavelength Array station-1), (ii) hardware to
equip a subset of dipoles for calibrated total-power measurements, and
(iii) a data calibration pipeline with which to commission the system
and demonstrate measurement of the spectrum of sky-averaged power with
an array. Cross-correlation for a large-N array scales as ~O(N2) and
poses significant computing challenges. Hundreds of apertures and
frequency channels, and tens of MHz bandwidth require > 1014
ops/s. The LEDA FX correlator (512 inputs and 50 MHz) will combine
Field Programmable Gate Arrays serving the O(N) F stage and Graphics
Processing Units (GPUs) serving the O(N2) X stage. The system will be
cost effective, power efficient, rapidly deployable, and
reconfigurable, thus enabling various further processing to be
performed in situ (e.g., pulsar gating for differential calibration of
dipole gain patterns).