Premium
Noise Integration Kernel Design for the Wave Distribution Function Method: Robust Direction Finding With Different Sensor Noise Levels
Author(s) -
Tanaka Yuji,
Ota Mamoru,
Kasahara Yoshiya
Publication year - 2021
Publication title -
radio science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.371
H-Index - 84
eISSN - 1944-799X
pISSN - 0048-6604
DOI - 10.1029/2021rs007291
Subject(s) - noise (video) , kernel (algebra) , noise measurement , earth's magnetic field , acoustics , gradient noise , robustness (evolution) , computer science , noise floor , physics , noise reduction , mathematics , magnetic field , artificial intelligence , combinatorics , quantum mechanics , image (mathematics) , gene , biochemistry , chemistry
Plasma wave direction determined by propagation characteristics and remote sensing technology provides important information for understanding not only the local plasma environment but also the global features of space plasma. The wave distribution function method, a direction finding method, derives the directional distribution of wave energy density using a priori information such as the propagation mode, plasma density, and geomagnetic field intensity. The Markov random field model includes a noise integration kernel that corresponds to the noise used for the estimation, which improves the robustness of the model. However, the effectiveness of this noise integration kernel was only verified for the case where the electromagnetic field sensor noise levels were equal. The noise levels of the electromagnetic field sensors on board scientific satellites often change due to the degradation of the sensors during long‐term instrument operation. Therefore, we proposed two design methods for a noise integration kernel to improve the estimation accuracy when the noise levels among electromagnetic field sensors are different. In proposed model 1, the noise integration kernel was designed for each sensor, and in proposed model 2, the noise integration kernel was designed with the information that the noise level ratio between the sensors was known. The simulation results showed that proposed model 2 improved the misalignment of the peak in the direction of arrival when the noise levels were different regardless of whether the sensor was parallel or perpendicular to the external magnetic field.