Abstract:
As one of the important methods to detect the cosmic reionization epoch, the
core goal of the low-frequency all-sky total power experiment is to detect the extremely
weak neutral hydrogen 21 cm signal during the cosmic reionization period in the all-sky
total power spectrum with a spectrum range of 50∼200 MHz. At present, the fast Fourier
transform (FFT) is usually used to calculate the power spectrum, and the spectral leakage
problem makes it impossible to completely deduct Radio Frequency Interference (RFI), which
seriously interferes with the extraction of weak cosmic reionization signals. Polyphase filter
banks (PFB) is a technique that evenly divides signals by frequency, which can replace
FFT to calculate the power spectrum. In this paper, the basic principle of PFB and its
implementation based on compute unified device architecture (CUDA) are introduced, and
the performance of PFB and FFT is compared through simulation, and the performance
differences of PFB of different orders are compared. Finally, the measured and experimental
data are processed to verify the superiority of PFB technology, which provides an effective
solution for the spectrum analysis of the all-sky total power experiment.