Performance assessment of time–frequency RFI mitigation techniques in microwave radiometry

Abstract

Radio–frequency interference (RFI) signals are a well-known threat for microwave radiometry (MWR) applications. In order to alleviate this problem, different approaches for RFI detection and mitigation are currently under development. Since RFI signals are man made, they tend to have their power more concentrated in the time–frequency (TF) space as compared to naturally emitted noise. The aim of this paper is to perform an assessment of different TF RFI mitigation techniques in terms of probability of detection, resolution loss (RL), and mitigation performance. In this assessment, six different kinds of RFI signals have been considered: a glitch, a burst of pulses, a wide-band chirp, a narrow-band chirp, a continuous wave, and a wide-band modulation. The results show that the best performance occurs when the transform basis has a similar shape as compared to the RFI signal. For the best case performance, the maximum residual RFI temperature is 14.8 K, and the worst RL is 8.4%. Moreover, the multiresolution Fourier transform technique appears as a good tradeoff solution among all other techniques since it can mitigate all RFI signals under evaluation with a maximum residual RFI temperature of 21 K, and a worst RL of 26.3%. Although the obtained results are still far from an acceptable bias Misplaced < 1 K for MWR applications, there is still work to do in a combined test using the information gathered simultaneously by all mitigation techniques, which could improve the overall performance of RFI mitigation.