On Detection of Radio Frequency Interference in Spaceborne Microwave Radiometers using Negentropy Approximations and Complex Signal Kurtosis as Test Statistics

Microwave radiometers are passive, sensitive radio receivers used as spacecraft instruments for Earth remote sensing. Microwave radiometers work by measuring average power of naturally occurring thermal emission from Earth. Radiometer data are used by researchers to monitor Earth's hydrosphere and a number of important geophysical processes critical to our understanding of and survival on the planet.;Besides measuring natural thermal noise, radiometers also measure inadvertent signal emissions from human-made sources, such as surveillance radar, communications systems, reflected signals from broadcast satellites, and signals from a wide array of wireless technologies. Signals of this nature are called radio-frequency interference (RFI) and corrupt radiometric measurements, leading to erroneous geophysical retrievals. RFI threatens the utility of radiometer data as a result of this corruption.;This research investigates two new approaches for detecting RFI. The first approach investigates the suitability of several different negentropy approximations as test-statistics for detecting RFI in the radiometric signal. The second approach combines polarimetric and baseband quadrature signals into a complex signal model and employs a complex signal kurtosis test-statistic to improve RFI detection relative, to the current state of the art that uses only real-valued signal kurtosis. It is shown that various negentropy approximations do not outperform detection performance of the real-valued kurtosis, but does provide insight as to why kurtosis works well. It is also shown that RFI detection in the complex domain using the complex signal kurtosis offers improved detection performance and less data rate, leading to new digital signal processing systems for future spaceborne microwave radiometers.