Radio frequency interference identification using eigenvalue decomposition for multi-beam observations

TL;DR

The paper addresses RFI contamination in high-time-resolution radio astronomy with multi-beam receivers by introducing mRAID, a cross-correlation-matrix–based detector that uses eigenvalue decomposition to identify interference. The method combines ArPLS-based bandpass normalisation with CCM construction and an eigen-decomposition framework to flag RFI, followed by iterative masking using Gaussian fits to eigenvalue distributions and per-beam eigenvector thresholds; it is highly parallelisable across subintervals and frequency channels. On FAST 19-beam data, mRAID demonstrates improved identification of weak and short-duration RFI and yields cleaner data with fewer false candidates than rfifind, while preserving genuine astronomical signals. The approach is computationally scalable and adaptable to other multi-beam systems, offering a practical pathway to enhanced RFI mitigation for next-generation surveys and PAF-based instruments, with open-source software available for community use.

Abstract

With the installation of next-generation phased array feed (PAF) receivers on radio telescopes, there is an urgent need to develop effective and computationally efficient radio frequency interference (RFI) mitigation methods for large-scale surveys. Here we present a new RFI mitigation package, called mRAID (multi-beam RAdio frequency Interference Detector), which uses the eigenvalue decomposition algorithm to identify RFI in cross-correlation matrix (CCM) of data recorded by multiple beams. When applied to high time-resolution pulsar search data from the Five-hundred-meter Aperture Spherical Radio Telescope (FAST), mRAID demonstrates excellent performance in identifying RFI over short timescales, thereby enhancing the efficiency of pulsar and fast radio burst (FRB) searches. Since the computation of the CCM and the eigenvalue decomposition for each time sub-integration and frequency channel are independent, the process is fully parallelisable. As a result, mRAID offers a significant computational advantage over commonly used RFI detection methods.

Figures (7)

• Figure 1: Histogram of dominant eigenvalues derived for all subintervals and channels of the FAST 19-beam test observation. A Gaussian fit to the distribution is shown as the red line.
• Figure 2: The left panel shows dominant eigenvalues derived for the FAST 19-beam test observation. The right panel shows the result after masking out RFI affected subintervals and channels, where the eigenvalue distribution becomes noise-like, indicating the effectiveness of the iterative RFI identification process.
• Figure 3: Histogram of dominant eigenvectors of non-RFI channels (those not flagged as RFI in Figure \ref{['Eigval_fit']}) derived for the FAST 19-beam test observations. A Gaussian fit to the distribution is shown as the red line.
• Figure 4: Time averaged spectrum of each beam of the FAST observation. Raw spectra are shown as black points; results of rfifind are shown as green; results of mRAID are shown as blue. Compared with the raw spectra, while both methods effectively identify strongly RFI-affected channels, mRAID shows superior performance in identifying weak RFI.
• Figure 5: RFI masked data of beam 11 of the FAST observation. Left: results of rfifind; right: results of mRAID. Compared with rfifind, mRAID produces much cleaner data in time and frequency after masking, effectively preserving uncontaminated regions while removing both narrow-band and broadband RFI.