LMOR-Net: Lightweight Local-Manifold Open-Set Detection and Candidate-Region Retrieval for Compound Radar Interference

Compound radar interference can mix known components with interference categories absent from the training data. Analysis of these recordings requires both a decision on unknown presence and candidate time–frequency regions in which to examine the unknown contribution. We present LMOR-Net to provide these complementary outputs. A recording detector compares the features of the complete signal with known training examples. A source-recovery network estimates component waveforms; their features are compared with known-source references to select candidate regions. This separates the decision about a complete mixture from the search for its unknown contribution. Network training and numerical threshold calibration use only known-category data. On a test set independently generated by the same simulator after the method and thresholds were fixed, the area under the receiver operating characteristic curve (AUROC) is 0.895±0.019 across three training seeds. Candidate-box and pixel evaluations show overlap between the retrieved regions and unknown reference support. Restricting regional retrieval to recordings flagged by the detector reduces false proposals and average processing time, with a reduction in regional coverage. LMOR-Net thus extends unknown-interference detection to candidate-region discovery through a compact two-network system with a measured trade-off between coverage, false proposals and processing cost.

Authors

Institutions

Publication Details

Journal
Sensors
Published
2026-11-16
DOI
https://doi.org/10.3390/s26227214
Primary Topic
Radar Systems and Signal Processing
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

LMOR-Net: Lightweight Local-Manifold Open-Set Detection and Candidate-Region Retrieval for Compound Radar Interference

宋祖勋, Haotian Zheng
Sensors
Radar Systems and Signal Processing
article

LMOR-Net: Lightweight Local-Manifold Open-Set Detection and Candidate-Region Retrieval for Compound Radar Interference

宋祖勋, Haotian Zheng
article en

Abstract

Compound radar interference can mix known components with interference categories absent from the training data. Analysis of these recordings requires both a decision on unknown presence and candidate time–frequency regions in which to examine the unknown contribution. We present LMOR-Net to provide these complementary outputs. A recording detector compares the features of the complete signal with known training examples. A source-recovery network estimates component waveforms; their features are compared with known-source references to select candidate regions. This separates the decision about a complete mixture from the search for its unknown contribution. Network training and numerical threshold calibration use only known-category data. On a test set independently generated by the same simulator after the method and thresholds were fixed, the area under the receiver operating characteristic curve (AUROC) is 0.895±0.019 across three training seeds. Candidate-box and pixel evaluations show overlap between the retrieved regions and unknown reference support. Restricting regional retrieval to recordings flagged by the detector reduces false proposals and average processing time, with a reduction in regional coverage. LMOR-Net thus extends unknown-interference detection to candidate-region discovery through a compact two-network system with a measured trade-off between coverage, false proposals and processing cost.

SensorsVol. 26(22)
Northwestern Polytechnical University (CN)
Radar Systems and Signal Processing
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon