RFI Mitigation Research And Application In CRAFTS Multi-beam Data

Radio frequency interference(RFI)is a well-recognized problem in radio astronomy research.As the most sensitive single-aperture telescope in the world,the Five hundred meter Aperture Spherical Radio telescope(FAST)is extremely sensitive to Radio Frequency Interference(RFI).The Commensal Radio Astronomy FAST Survey(CRAFTS)is a significant project conducted using the FAST’s 19-beam receiver,aiming to perform large-scale radio sky surveys.The project is expected to cover a substantial portion of the sky,approximately 57%,and generate unprecedented data with remarkable sensitivity and resolution.CRAFTS aims to search for distant galaxies,rapidly rotating neutron stars,and cosmic transient sources.Among them,pulsars and Fast Radio Bursts(FRBs)are higher-priority scientific targets in the CRAFTS project.However,due to the presence of RFI,the data quality of multi-beam observations is often affected to varying degrees.The primary concern in RFI mitigation is to remove RFI while preserving genuine astronomical signals.To improve the quality of RFI removal from multi-beam data,we propose a simple but powerful method,CCF-ST,which constructs a spatial filter based on Cross-Correlation Function(CCF)and Sum Threshold algorithms.The RFI marking results will be saved in a mask file,which is a binary format file used in the Pulsa R Exploration and Search TOolkit(PRESTO)to record the RFI location.The mask file can be used for RFI removal of raw data in the periodic search software or single-pulse search software in PRESTO.Due to the time-consuming calculation process of CCF,we parallelize this operation using Compute Unified Device Architecture(CUDA)based on graphics processing units(GPUs).Three real pulsars with different flow densities from CRAFTS observation data are used for experiments.Experimental results show that compared to rfifind(from PRESTO)and Ar PLS-ST,CCF-ST can effectively reduce false positives and increase the Signal-to-Noise Ratio(SNR)by ～26% and ～18%,respectively.Furthermore,compared to Ar PLS-SF,it has lower computational costs,reducing time consumption by ～40% and memory usage by ～90%.We plan to apply CCF-ST to the CRAFTS data processing production environment,to facilitate the discovery of more scientific goals,such as pulsars and FRBs.