Cognitive radio technology is an important branch in the field of wireless communication, and automatic identification is a major part of cognitive radio technology. Convolutional neural network (CNN) is an advanced neural network, which is the forefront of application in the digital image recognition area. In this paper, we explore CNN in an automatic system to recognize the cognitive radio waveforms. Excitedly, it is a more effective model with high ratio of successful recognition (RSR) under high power background noise. The system can identify eight kinds of signals, including binary phase shift keying (Barker codes modulation) linear frequency modulation, Costas codes, Frank code, and polytime codes (T1, T2, T3, and T4). The recognition part includes a CNN classifier. First, we determine the appropriate architecture to make CNN effective for proposed system. Specifically, we focus on how many convolutional layers are needed, what appropriate number of hidden units is, and what the best pooling strategy is. Second, we research how to obtain the image features into CNN that based on Choi–Williams time-frequency distribution. Finally, by means of the simulations, the results of classification are demonstrated. Simulation results show the overall RSR is 93.7% when the signal-to-noise ratio is −2dB.

Convolutional Neural Networks for Automatic Cognitive Radio Waveform Recognition

In this paper, we apply the idea of deep learning to radar waveform recognition. Since the frequency variation with time is the most essential distinction among radar signals with different modulation types, we transform one-dimensional radar signals into time-frequency images (TFIs) using time-frequency analysis and design a convolutional neural network to recognize the frequency variation patterns exhibited in TFIs. Furthermore, we analyze the statistical characteristics of the noise in TFIs and introduce a naive approach to reduce its influence on the frequency variation patterns. Simulation results demonstrate the impressive recognition rate under very low SNR conditions and the strong generalization ability of our proposed recognition method.

Automatic radar waveform recognition based on time-frequency analysis and convolutional neural network

In this paper, an automatic radar waveform recognition system in a high noise environment is proposed. Signal waveform recognition techniques are widely applied in the field of cognitive radio, spectrum management and radar applications, etc. We devise a system to classify the modulating signals widely used in low probability of intercept (LPI) radar detection systems. The radar signals are divided into eight types of classifications, including linear frequency modulation (LFM), BPSK (Barker code modulation), Costas codes and polyphase codes (comprising Frank, P1, P2, P3 and P4). The classifier is Elman neural network (ENN), and it is a supervised classification based on features extracted from the system. Through the techniques of image filtering, image opening operation, skeleton extraction, principal component analysis (PCA), image binarization algorithm and Pseudo–Zernike moments, etc., the features are extracted from the Choi–Williams time-frequency distribution (CWD) image of the received data. In order to reduce the redundant features and simplify calculation, the features selection algorithm based on mutual information between classes and features vectors are applied. The superiority of the proposed classification system is demonstrated by the simulations and analysis. Simulation results show that the overall ratio of successful recognition (RSR) is 94.7% at signal-to-noise ratio (SNR) of −2 dB.

LPI Radar Waveform Recognition Based on Time-Frequency Distribution.

In this paper, to solve the problem of the low recognition rate of the existing approaches at low signal-to-noise ratio (SNR), an intra-pulse modulation recognition approach for radar signal is proposed. The approach identifies the modulation of radar signals using the techniques of time-frequency analysis, image processing, and convolutional neural network (CNN). Through Cohen class time-frequency distribution (CTFD), the time-frequency images (TFIs) of received signals are extracted. In order to obtain the high-quality TFIs of received signals, we introduce a new kernel function for the CTFD, which has stronger anti-noise ability than Choi–Williams time-frequency distribution. A series of image processing techniques, including 2-D Wiener filtering, bilinear interpolation, and Otsu method, are applied to remove the background noise of the TFI and obtain a fixed-size binary image that contains only morphological features of the TFI. We design a CNN classifier to identify the processed TFIs. The proposed approach can identify up to 12 kinds of modulation signals, including frequency modulation, phase modulation, and composite modulation. Simulation results show that, for 12 kinds of modulation signals, the proposed approach achieves an overall probability of successful recognition of 96.1% when SNR is −6 dB.

Radar Signal Intra-Pulse Modulation Recognition Based on Convolutional Neural Network

Signal modulation classification is an important research subject in both military and civilian field. This study proposed a novel blind modulation classification method based on the time-frequency distribution and convolutional neural network (CNN). This is the first attempt to treat the time-frequency map as a picture and use an outstanding (CNN-based) algorithm in computer vision area for signal recognition. The combination offers a novel feature extraction strategy, to some extent, which also conforms to intuition. Simulation results show that the method proposed in this study is efficient and robust and enables a high degree of automation for extracting features, training weights and making decisions. Additionally, a remarkable performance emerges with small samples and repeated training, which distinguishes this method from many other classification methods.

https://ietresearch.onlinelibrary.wiley.com/doi/pdf/10.1049/iet-rsn.2017.0265

Modulation classification method for frequency modulation signals based on the time–frequency distribution and CNN

It is an important work to classify the modulation type of the intercepted radar signal for an electronic intelligence (ELINT) receiver in a non-cooperative environment. The authors use the Rihaczek distribution (RD) and the Hough transform (HT) to concentrate the energy in time-frequency plane and derive two new characteristic features, namely the ratio of the minimum to the maximum of the HT and the peak number of the HT of the real part of the RD, to improve the probabilities of successful recognition (PSRs) to recognise the classical low probability of intercept (LPI) radar signals. The first feature is especially suitable for the linear frequency modulation (LFM), whereas the second one is specifically designed for frequency shift keying (FSK). The choice of thresholds and the effects of signal parameters are analysed. Simulations show that the PSRs can reach 90% when the signal-to-noise ratio (SNR) is above -4 dB. The proposed algorithm is better than the previous algorithms by just using ambiguity function.

Automatic modulation classification of radar signals using the Rihaczek distribution and hough transform

The automatic modulation classification (AMC) of a detected radar signal is a challenging task of an electronic intelligence (ELINT) receiver in a non-cooperative environment. With the aim to realise the AMC of five kinds of radar signals under negative signal-to-noise ratio (SNR), the authors have gained four characteristic features, namely, the ratio of sum of absolute slope, the coefficient of polynomial curve fitting, the number of ridge stairs and the normalised coefficient of difference of the extreme, from the generalised time-frequency representation of Zhao, Atlas and Marks (ZAM-GTFR). Simulation results show the probabilities of successful recognition (PSRs) can reach 90% when SNR is above -2%dB. The algorithm is suitable for the ELINT receiver when the detection range is critical.

Automatic modulation classification of radar signals using the generalised time-frequency representation of Zhao, Atlas and Marks

Based on the ambiguity function, an approach to intra-pulse modulation recognition is presented. This paper defines the frequency energy accumulation function (FEAF) and time-domain extremum function (TEF) to describe the Doppler frequency shift and time delay properties of signals, respectively. This approach can recognize six types of signals by four characteristic parameters. Simulation results show that the probability of successful recognition (PSR) can reach 90% when the signal to noise ratio (SNR) is above −1 dB.

An Approach to Intra-Pulse Modulation Recognition Based on the Ambiguity Function

SFM signal parameter estimation algorithm based on Carson rule: SFM signal parameter estimation algorithm based on Carson rule

Recognition of intra-pulse modulation signal types based on phase difference: Recognition of intra-pulse modulation signal types based on phase difference

Deguo Zeng

Papers

Automatic modulation classification of radar signals using the Rihaczek distribution and hough transform

Automatic modulation classification of radar signals using the generalised time-frequency representation of Zhao, Atlas and Marks

An Approach to Intra-Pulse Modulation Recognition Based on the Ambiguity Function

SFM signal parameter estimation algorithm based on Carson rule: SFM signal parameter estimation algorithm based on Carson rule

Recognition of intra-pulse modulation signal types based on phase difference: Recognition of intra-pulse modulation signal types based on phase difference