1. Introduction
In modern warfare, signal sorting is an important part in the electronic support system (ESM); the sorting level has become an important symbol to measure the technology level of the reconnaissance equipment [
1,
2]. With the rapid development of radar technology and the increasing degree of radar electronic countermeasures, the electromagnetic environment is becoming more and more complex. The long-term used method based on the traditional five parameters, radio frequency (RF), pulse repeat interval (PRI), pulse width (PW), angle of arrival (AOA) and pulse amplitude (PA), cannot achieve the satisfactory sorting effect in today’s radar-signal environment; the phenomena of “combining-batch” and “increasing-batch” are very serious [
3,
4,
5]. Therefore, domestic and foreign scholars have done a lot of in-depth and systematic studies.
At present, the research results are divided into two categories. One is the intelligent sorting algorithms such as the clustering algorithm or classifier, which can process the feature-parameter uncertainty or overlapping of the measured emitter signals and classify them correctly. At present, many intelligent sorting algorithms and classifiers have been successfully applied, such as fuzzy clustering [
6], K-means clustering [
7], grid clustering [
8], density clustering [
9], support vector machine classification [
10], neural networks [
11,
12], and so on. However, some of them can only solve the problem of the uncertainty or overlapping of the emitter feature parameters to a certain extent, and some of them are very sensitive to signal-to-noise ratios (SNRs). The other is trying to find new characteristic parameters to make up for the shortcomings of the typical five parameters. Currently, many studies are on the intrapulse feature, such as entropy [
13], resemblance coefficient [
14], wavelet gray matrix feature [
15,
16], fractal feature [
17] and so on. To some extent, these methods have achieved some satisfactory results. However, when two radars that have the same signal parameters and signal modulation type are deployed in close proximity, a signal-sorting method based on the intrapulse feature is almost invalid. In this paper, the primary focus is on this.
In recent years, with the wide application of digital-image-processing technology in radar signal time-frequency processing [
18], it provides a new perspective for radar signal sorting. Bai et al. [
19] proposed a novel radar-signal-recognition method based on the image feature of Choi–Williams time–frequency distribution. Yong et al. [
20] used image-processing technology to extract recognition features from jet-engine modulation signals. Intrinsic time-scale decomposition (IDT) and texture analysis were used in [
21] for emitter recognition. On the basis of the research mentioned above, in view that coherence is an important characteristic of today’s modern radar [
22], a new signal-sorting method based on coherent characteristics is proposed. In this paper, the feasibility of utilizing the coherent characteristic as the feature parameter for signal sorting has been proved. The discrete Fourier transform (DFT) was taken to get the spectrum image from the radar signals. Then, the spectrum image pretreatment was made, and the central-moment feature of the spectrum image was extracted as the coherent feature. To check the sorting performance of the coherent feature, a number of simulations were carried out in terms of the coherent feature and the typical five parameters on complex sorting scenes.
It is noted that the term “coherent characteristics” mentioned in this paper represents that the initial phase of transmitting signals from the coherent radar is the same, while that from the noncoherent radar is random.
The rest of this paper is organized as follows. In
Section 2, a reconstruction signal model of signal sorting is given. Through the theoretical analysis, the feasibility of the coherent characteristics as the signal-sorting feature is proved, and the frequency-domain model of coherent characteristics is constructed. In
Section 3, image-processing technology is utilized to extract the coherent feature. In
Section 4, three experiments are performed to demonstrate the sorting capability of the proposed coherent feature. Conclusions are made in
Section 5.