Online Clustering Algorithms for Radar Emitter Classification
IEEE Transactions on Pattern Analysis and Machine Intelligence
Optimization of time and frequency resolution for radar transmitter identification
ICASSP '99 Proceedings of the Acoustics, Speech, and Signal Processing, 1999. on 1999 IEEE International Conference - Volume 03
Automatic target recognition using waveform diversity in radar sensor networks
Pattern Recognition Letters
Optimizing time-frequency kernels for classification
IEEE Transactions on Signal Processing
| Hi-index | 0.00 |
Radar emitter identification has attracted increasing interests in the last decade. The class-dependent method in [5,6] to optimize Time-Frequency kernel of ambiguity function (AF) needs to rank kernel points in the whole AF plane and is sensitive to sampling data length. In this paper, an ambiguity function zero-slice based feature optimization algorithm is proposed for radar emitter recognition. It efficiently extracts the zero-slice feature of AF as intermediate feature set and avoids "out of memory" problem as in large whole-plane optimization. Further, a Direct Discriminant Ratio (DDR) criterion is employed to rank the kernel points along the obtained slice. The resulting scheme not only preserves the most discriminant features of individual emitters, but also improves the recognition accuracy greatly. The experiments on both simulation radar data from U.S. Naval Research Laboratory and real radar emitter data demonstrate the feasibility and effectiveness of the proposed method