Radio Frequency Mitigation Using Independent Component Analysis

Radio astronomy has become an important way to study the universe. However, with the development of human activities, radio frequency interference (RFI) has more and more serious impact on radio astronomical observation. The quality of observation is related to the quality of scientific achievements and even the authenticity of conclusions. At present, RFI detection based on threshold is widely used, and part of the observed data is directly discarded for the interference. Such methods have difficulties in determining threshold values, and reduce observation bandwidth and time. Observed the fact that interferences and radio signals are statistically independent and non-Gaussian, we propose a novel approach for RFI mitigation using independent component analysis to decompose the mixed signal, then identifying the pulsar signal according to the different distribution characteristics between the pulsar signal and RFI signals. The pulsar observations received from 40-meter radio telescope in Yunnan Observatories are processed by the new approach. The results show:RFI signals in pulsar observations are cleanly mitigated while pulsar signal is barely affected and good signal-to-noise ratio is achieved.