Deep Learning-Based Approach for Low Probability of Intercept Radar Signal Detection and Classification
- PDF / 1,546,460 Bytes
- 13 Pages / 612 x 792 pts (letter) Page_size
- 17 Downloads / 219 Views
HEORY AND METHODS OF SIGNAL PROCESSING
Deep Learning-Based Approach for Low Probability of Intercept Radar Signal Detection and Classification G. Ghadimia, Y. Norouzib, R. Bayderkhania, *, M. M. Nayebic, and S. M. Karbasic aEngineering
Department, Islamic Azad University Central Tehran Branch, Tehran, Iran of Electrical Engineering, Amirkabir University of Technology, Tehran, Iran c Electronic Engineering Department, University of Sharif, Tehran, Iran *e-mail: [email protected]
bDepartment
Received April 1, 2020; revised April 1, 2020; accepted May 7, 2020
Abstract—Detection and classification of Low Probability of Interception (LPI) radar signals is one of the most important challenges in electronic warfare (EW), since there are limited methods for identifying these type of signals. In this paper, a radar waveform automatic identification system for detecting and classifying LPI radar is studied, and accordingly we propose a method based on deep learning networks to detect and classify LPI radar waveforms. To this end, the GoogLeNet architecture as one of the well-known convolutional neural networks (CNN) is utilized. We employ the Short Time Fourier Transform (STFT) for timefrequency analysis in order to construct the entry image for proposed method 1,2 (improved the GoogLeNet and AlexNet networks) to recognize offline training and online recognition. After the training procedure with the supervised data sets the proposed method 1,2 can detect and classify nine modulation types of LPI radar, including LFM, poly-phase (P1, P2, P3, P4) and poly-time (T1, T2, T3, T4) waveforms. The numerical results for proposed method 1, show considerable accuracies up to 98.7% at the SNR level of –15 dB, which outperforms the existing methods. Keywords: low probability of intercept radar, short time fourier transform, googlenet, convolutional neural network, deep learning DOI: 10.1134/S1064226920100034
1. INTRODUCTION In the past, many designed radars used short-duration and high peak power pulses to reduce their electromagnetic radiation losses. When it comes military purposes, it was important to keep these signals hidden from enemy forces. However, nowadays, by using low probability of intercept (LPI) techniques, it is possible to design a LPI radar which is efficient against intercept electronic warfare (EW) receivers [1]. The LPI radar systems utilize techniques such as low transmit power, wide bandwidth and frequency variations to make the interception process more difficult for intercept EW receivers [1]. LPI waveforms include linear and nonlinear Frequency Modulation (LFM, NLFM), polyphase-codes such as Frank; P1, P2, P3 and P4; frequency hopping; combined frequency hopping-phase coding; and ploy time-coded such as T1, T2, T3 and T4. This requires intercept receivers to be equipped with an automatic LPI radar waveform recognition function, with two main steps, to recognize the presence of LPI radar signals [2], i.e., the feature extraction step followed by the detection and classification step.
The main goal
Data Loading...
