Phase enhancement model based on supervised convolutional neural network for coherent DOA estimation

When the elevation of targets is smaller than beamwidth, the coherent multi-path signals will significantly degrade the direction of arrival (DOA) estimation accuracy of existing methods for a very-high-frequency (VHF) radar system. Through detailed theoretical analysis, we demonstrate that the phas...

Full description

Saved in:
Bibliographic Details
Published in:Applied intelligence (Dordrecht, Netherlands) Netherlands), 2020-08, Vol.50 (8), p.2411-2422
Main Authors: Xiang, Houhong, Chen, Baixiao, Yang, Ting, Liu, Dong
Format: Article
Language:English
Subjects:
Accuracy
Artificial Intelligence
Artificial neural networks
Coherence
Computer Science
Computer simulation
Direction of arrival
Machines
Manufacturing
Mechanical Engineering
Neural networks
Phase distortion
Processes
Radar equipment
Very high frequencies
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:When the elevation of targets is smaller than beamwidth, the coherent multi-path signals will significantly degrade the direction of arrival (DOA) estimation accuracy of existing methods for a very-high-frequency (VHF) radar system. Through detailed theoretical analysis, we demonstrate that the phase distortion is the key factor of degrading the accuracy of DOA estimation. Hence, a novel phase enhancement model based on supervised convolutional neural network (CNN) for coherent DOA estimation is proposed to mitigate the phase distortion and improve estimation accuracy. The results of simulation experiments and real data have demonstrated the superiority of proposed method in DOA estimation accuracy and resolution compared to classic physics-driven methods. Moreover, the proposed scheme is suitable for the coherent DOA estimation compared with existing data-driven methods.
ISSN:0924-669X
1573-7497
DOI:10.1007/s10489-020-01678-4