RFI Localization Using Jointly Non-Convex Low-Rank Approximation and Expanded Virtual Array in Microwave Interferometric Radiometry
Yanyu Xu, Dong Zhu, Fei Hu, Peng Fu
Abstract
The scientific goal of the Soil Moisture and Ocean Salinity (SMOS) mission is to retrieve the geophysical parameter from brightness temperature (TB) maps. However, radio frequency interference (RFI) significantly influences the interpretation of TB maps, leading to a deteriorated retrieval performance. RFI localization is essential for switching off these illegal emitters and mitigating their impacts on TB maps. This article proposes a novel high-resolution RFI localization method via jointly non-convex low-rank approximation and expanded virtual array (EVA). Concretely, the RFI localization problem is first formulated from the perspective of non-convex low-rank recovery, which better approximates the rank of the covariance matrix collecting visibility samples. Then, we propose the EVA concept by relaxing the size constraint on the physical antenna array. Moreover, we use a new algorithm based on the joint Schatten- <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$p$ </tex-math></inline-formula> and <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$Lp$ </tex-math></inline-formula> (JSL) norms to solve the above non-convex low-rank recovery problem. This JSL algorithm can improve the spatial resolution for RFI localization. Combining the JSL algorithm and the EVA can further improve the detection performance and enhance the spatial resolution for RFI localization. The experimental results using synthetic data and real SMOS data prove that the proposed method shows enhanced spatial resolution, better detection performance, and competitive or better localization accuracy compared with the currently existing methods.