Radar Recognition Research Articles

Pulse‐level work state recognition of multifunction radar based on MC‐RSG

AbstractAccurate work state recognition of multifunction radar (MFR) is crucial in electronic warfare, as it helps understand the enemy's intention and evaluate potential threats. A pulse‐level work state recognition method of MFR based on the residual block with spatial attention connected gated recurrent unit by features using metric coding and correlative embedding (MC‐RSG) is proposed. Metric coding is designed to generate the distance vector with time of arrival, and the correlative embedding is performed on the distance vector and raw data features to increase the feature information by extracting feature information associated with the previous and subsequent pulses in each feature sequence, respectively. Besides, we make use of the model called RSG containing the residual block with spatial attention connected gated recurrent unit to learn the features of pulse sequences and identify the radar work state label of each pulse. The experimental work shows that the method is robust and has achieved up to 97% recognition accuracy on the test dataset under ideal observation conditions and 5% higher than the comparison network in high noise observation conditions.

Using artificial intelligence methods to recognize radar stations based on the radio technical parameters of signals with filtering signals from radar stations not taken into account in the training set

Currently, the tasks of radio-electronic surveillance using spacecraft are the most relevant in ensuring the defense capability of the Russian Federation. In the course of solving the problem of radio-electronic surveillance, the issue of classifying the detected signals. Classical classification methods, based on comparing the obtained data on detected stations with reference a priori-known station parameters, do not allow obtaining results with a high degree of reliability, so to solve this issue we turned to machine learning technologies. This paper describes the solution to the problem of recognizing radar stations by the radio technical parameters of signals received during radio-electronic monitoring (SEM) of the Earth, using machine learning methods, taking into account the possibility of signals from unknown radars appearing in the test sample. In the process of work, an analysis of existing machine learning methods for processing SEM information was carried out using a developed mathematical model, implemented in program code, using simulated test data. In order to probabilistically evaluate methods for recognizing types of radar stations (radars) as part of the analysis of SEM information in a complex signal-interference environment, modeling of the initial data, consisting of a set of pairs “classification object - class label” was carried out. The radar type number was chosen as the class label, and the radar signal, which can be described using three characteristics: frequency, period, and signal pulse duration, was chosen as the classification object. The mathematical model for generating the initial data includes the ability to add a normally distributed measurement error depending on the specified standard deviation (RMSD) for each of the characteristics. In addition, the possibility of using various machine learning methods to search for outliers in the test sample was considered, and a probabilistic assessment of the accuracy of their work was also carried out. As a result of the research, based on the results obtained, the two most preferred methods of radar recognition were selected, as well as the most promising method of filtering signals from unknown radars. The use of the developed algorithms makes it possible to significantly speed up the recognition of radars based on the radio technical parameters of the signal with a high degree of recognition reliability.

Radar recognition system based on XG-Boost

Radar recognition system based on XG-Boost

Cluster-Based Input Weight Initialization for Echo State Networks.

Echo state networks (ESNs) are a special type of recurrent neural networks (RNNs), in which the input and recurrent connections are traditionally generated randomly, and only the output weights are trained. Despite the recent success of ESNs in various tasks of audio, image, and radar recognition, we postulate that a purely random initialization is not the ideal way of initializing ESNs. The aim of this work is to propose an unsupervised initialization of the input connections using the K-means algorithm on the training data. We show that for a large variety of datasets, this initialization performs equivalently or superior than a randomly initialized ESN while needing significantly less reservoir neurons. Furthermore, we discuss that this approach provides the opportunity to estimate a suitable size of the reservoir based on prior knowledge about the data.

Selected aspects of measurement data processing in electronic warfare devices

Electronic warfare (Polish acronym WRE ‒ walka radioelektroniczna) is a set of militaryactions based on the use of the electromagnetic spectrum and it includes:— reception and identification of electromagnetic emissions,— reducing the effectiveness of the enemy’s electronic devices,— enabling effective use of the spectrum by own forces.The task of the WRE in peacetime is to obtain information about radio-electronic devices, especiallyradars. These tasks are performed by ELINT (Electronic Intelligence) and/or ESM (Electronic WarfareSupport Measures) devices. They operate passively in the 0.5÷18 GHz band, their detection is difficultand the information they provide allows to detect a threat to own forces and to take appropriatecountermeasures. Radar recognition is a complex process that involves the following stages. The firststage: gathering measurements and their grouping into the so-called packets and graphical display ofthe received pulse trains. The measurements are represented by the so-called measurement vectors(descriptors) containing signal parameters, including the TOA (Time of Arrival) pulse with nanosecondaccuracy. The second stage: associating pulse packets with logical sources in order to obtainrepresentative data strings. This stage is signal sorting and deinterleaving. The third stage: estimationof the WS signal vector based on the associated pulse packets. The final stage of data processing is thecomparison of the obtained signal descriptor parameters WS with Database (BD) radar patterns. Thereconnaissance result allows for the assessment of the threat resulting from the operation of this radar.Radar recognition requires precise determination of many parameters of its signal. For pulsed radars,important parameters are the Pulse Repetition Interval (PRI) and the type of PRI modulation. Radarsare characterised by a high complexity of PRI changes. This provides great utility and makes it possibleto distinguish even radar units and changes in their locations. This allows for the detection and fullidentification of objects that use radars (e.g. ships) with the given object completely out of sight; it alsoallows to predict intentions of the target. The basic inter-pulse PRI modulations include: stagger, dwelland switch, jitter or sliding. PRI changes in radar signals may be periodic with long cycles, difficultto identify in cluttered conditions. Unfortunately, the received pulse trains are usually distorted. Thismakes the WS estimation process difficult ‒ especially in automatic mode. PRI analysis has receivedmuch attention in the literature. The methods are computationally complex and have numerouslimitations. This article presents ASWC algorithm (Sequential Cycle Detection Algorithm) of PRIproprietary, with relatively low computational complexity. Examples of test results confirming its higheffectiveness in automatic PRI analysis performed under interference conditions were also presented.Keywords: radars, electronic warfare, radar recognition, ELINT, ESM, pulse descriptor, PRI

CubeLearn: End-to-End Learning for Human Motion Recognition From Raw mmWave Radar Signals

mmWave FMCW radar has attracted huge amount of research interest for human-centered applications in recent years, such as human gesture and activity recognition. Most existing pipelines are built upon conventional Discrete Fourier Transform (DFT) pre-processing and deep neural network classifier hybrid methods, with a majority of previous works focusing on designing the downstream classifier to improve overall accuracy. In this work, we take a step back and look at the pre-processing module. To avoid the drawbacks of conventional DFT pre-processing, we propose a complex-weighted learnable pre-processing module, named CubeLearn, to directly extract features from raw radar signal and build an end-to-end deep neural network for mmWave FMCW radar motion recognition applications. Extensive experiments show that our CubeLearn module consistently improves the classification accuracies of different pipelines, especially benefiting those simpler models, which are more likely to be used on edge devices due to their computational efficiency. We provide ablation studies on initialization methods and structure of the proposed module, as well as an evaluation of the running time on PC and edge devices. This work also serves as a comparison of different approaches towards data cube slicing. Through our task agnostic design, we propose a first step towards a generic end-to-end solution for radar recognition problems.

A radar anti‐jamming strategy optimisation based on Stackelberg game

AbstractRadar is an important sensor in electromagnetic spectrum warfare. Its confrontation with naval vessels has become increasingly competitive in recent years. However, current radar anti‐jamming methods are limited to some extent in complex electromagnetic environments, which poses a severe challenge to radar's detection and anti‐jamming capabilities. To improve the anti‐jamming capacity of radar, the authors propose a Stackelberg game‐based optimisation method to enhance the decision‐making of anti‐jamming strategies. First, we analyse the radar's winning conditions by considering the temporal constraints of non‐real‐time radar recognition and preparation actions and construct the radar's actual utility matrix. Second, we construct a Stackelberg game‐based model under the condition of a certain recognition probability, and update the recognition probability and recognition interval during the game. Finally, we discuss the conditions under which the radar can obtain a positive benefit in the game and compare the benefit to the reinforcement learning optimisation method. The simulation experimental results show that the proposed strategy can significantly improve the radar's winning probability in the confrontation.

Working Mode Recognition of Non-Specific Radar Based on ResNet-SVM Learning Framework.

Mode recognition is a basic task to interpret the behavior of multi-functional radar. The existing methods need to train complex and huge neural networks to improve the recognition ability, and it is difficult to deal with the mismatch between the training set and the test set. In this paper, a learning framework based on residual neural network (ResNet) and support vector machine (SVM) is designed, to solve the problem of mode recognition for non-specific radar, called multi-source joint recognition framework (MSJR). The key idea of the framework is to embed the prior knowledge of radar mode into the machine learning model, and combine the manual intervention and automatic extraction of features. The model can purposefully learn the feature representation of the signal on the working mode, which weakens the impact brought by the mismatch between training and test data. In order to solve the problem of difficult recognition under signal defect conditions, a two-stage cascade training method is designed, to give full play to the data representation ability of ResNet and the high-dimensional feature classification ability of SVM. Experiments show that the average recognition rate of the proposed model, with embedded radar knowledge, is improved by 33.7% compared with the purely data-driven model. Compared with other similar state-of-the-art reported models, such as AlexNet, VGGNet, LeNet, ResNet, and ConvNet, the recognition rate is increased by 12%. Under the condition of 0-35% leaky pulses in the independent test set, MSJR still has a recognition rate of more than 90%, which also proves its effectiveness and robustness in the recognition of unknown signals with similar semantic characteristics.

Reconstruction of Radar Pulse Repetition Pattern via Semantic Coding of Intercepted Pulse Trains

The timing structure of multiple successive radar pulses constructs a high-dimensional pattern for intercepted pulse trains, which is called as pulse repetition interval (PRI) pattern. By treating the radar as a machine that uses differently permuted pulses to communicate with surroundings, this pattern acts as the grammatical structure of its language. Compared with the discrete PRI set that is conventionally used for pulse train description, PRI pattern contains richer and more condensed structural information about radar pulse train, which is helpful for pulse deinterleaving and radar recognition. This article introduces the semantic coding theory to reveal and reconstruct PRI pattern from the intercepted radar pulse train. We first define the coding complexity of a pulse train, which is divided into two parts: 1) the complexity of encoding a PRI pattern dictionary with each element consisting of several successive PRIs; and 2) encoding the intercepted pulse train based on this dictionary. The coding complexity is then minimized by optimizing the components in dictionary, and the PRI timing patterns are finally obtained from the dictionary when the minimization is reached. The effectiveness of semantic coding model and PRI pattern reconstruction method is verified in the simulation part.

Broadband one-dimensional range profiles characteristic of rough surface in terahertz band

<sec>The one-dimensional (1D) range profile is an important back scattering characteristic of objective, which reveals the longitudinal distribution of radar cross section (RCS) along the detection beam. Since the shape and posture can be reflected by the 1D range profile, it is of great significance in military to determine the target orientation, velocity and whether it is armed. In this paper, broadband terahertz 1D-range-profile measurement system is built based on the time-domain spectroscopy (TDS) system. It is in bistatic configuration (bistatic angle of 9°) and the signal-to-noise ratio (SNR) is 34.5 dB, with a gold mirror used as a reflector. Benefiting from the ultrashort terahertz pulse width (full pulse width of 0.52 ps), the bandwidth covers the frequency range from 0.1 THz to 2.5 THz (peaked at 0.9 THz), corresponding to the range resolution on a submillimeter scale.</sec><sec>Firstly, the 1D range profiles of several objects in different shapes are measured, including the step, cylinder, step cone and their combination, which indicates that the geometric profile of the target in the detection direction is adequate to identify the shape feature of the target and proves the reliability of the 1D range profile measuring system based on TDS. Secondly, aluminum plates with different surface roughness in a range of 0–25 μm are also characterized. The Kirchhoff approximation theory and small perturbation method (SPM) are introduced to illustrate the characteristics of broadband terahertz 1D range profile related to the surface roughness of target. It is found that the scattering characteristic of metal object in the terahertz range is sensitive to surface roughness. If the surface roughness of the object is larger, the peak intensity of the 1D range profile will be weaker and the echo signal pulse width becomes wider. The rule is also applicable for the cases with different incident angles. Furthermore, it is revealed that the time delay of the 1D range profile in the bistatic system is related to the rotation direction of the target, which is useful in estimating the posture of the target. In summary, the characteristics of 1D range profile for metal objects relating to shape, surface roughness and posture are studied. The conclusions have certain guiding significance for the target detection and recognition of terahertz radar.</sec>

Research on a Vehicle Recognition Method Based on Radar and Camera Information Fusion

To improve the accuracy and real-time performance of vehicle recognition in an advanced driving-assistance system (ADAS), a vehicle recognition method based on radar and camera information fusion is proposed. Firstly, the millimeter-wave radar and camera are calibrated jointly, the radar recognition information is mapped on the camera image, and the region of interest is established. Then, based on operator edge detection, global threshold binarization is performed on the image of the region of interest (ROI) to obtain the contour information of the vehicle in front, and Hough transform is used to fit the vehicle contour edge straight line. Finally, a sliding window is established according to the symmetry characteristics of the fitting line, which can find the vehicle region with the highest symmetry and complete the identification of the vehicle. The experimental results show that compared to the original recognition region of the radar, the mean square error of this algorithm is reduced by 13.4 and the single frame detection time is reduced to 28 ms. It is proven that the algorithm has better accuracy and a faster detection rate, and it can solve the problem of an inaccurate recognition region caused by radar error.

Target Recognition of Millimeter-wave Radar based on YOLOX

The application of magnetic field detector in the actual road section mainly depends on the layout of these detectors in the actual traffic environment. For example, geomagnetism and coils are easily disturbed by vehicles near the road, resulting in missed or false detection; Video detection equipment depends on weather, visibility and other environmental conditions, and there will be a large degree of missed detection. Compared with the disadvantages of the above sensors, Millimeter-wave Radar has many unique advantages, including insensitivity to light or weather, wider application range compared with vision based technology, and higher accuracy,which makes millimeter-wave radar more outstanding in traffic monitoring applications. This paper presents a detection model of millimeter wave radar spectrum based on YOLOX, which is used to monitor traffic. Firstly, this paper enhances the millimeter-wave radar spectrum data used for training by gray-scale transformation, solves the problem that the Doppler target echo is not obvious, and obtains a clearer target image. Secondly, the latest YOLOX algorithm is used to train 2844 radar spectrum diagrams, and the performance evaluation is carried out to obtain the YOLOX average mAP@0.5 The value is 0.916 and FPS is 35.8. Finally, experiments show that YOLOX algorithm is better than YOLOv5 algorithm mAP@0.5 It is 2.9% higher, which proves the superiority of the algorithm.

초해상화 기반 CNN을 이용한 군사용 SAR 자동표적인식 모델 연구

Herein, we propose employing the super-resolution-based convolutional neural network (CNN) to design the automatic target recognition (ATR) of military synthetic aperture radar (SAR) images. Previous SAR ATR methods showed a good recognition performance with a low depression angle, but poor performance with a high depression angle. In a warfighting environment, good recognition performance is required even with a high depression angle. To address this issue, we combine the super-resolution method and the CNN. In comparison with the conventional VGGnet with a high depression angle, the proposed super-resolution-based CNN showed a 3%-4% improvement in accuracy. The MSTAR SAR dataset was utilized for validation.

Phase Correlation Single Channel Continuous Wave Doppler Radar Recognition of Multiple Sources.

Continuous-wave Doppler radar (CWDR) can be used to remotely detect physiological parameters, such as respiration and heart signals. However, detecting and separating multiple targets remains a challenging task for CWDR. While complex transceiver architectures and advanced signal processing algorithms have been demonstrated as effective for multiple target separations in some scenarios, the separation of equidistant sources within a single antenna beam remains a challenge. This paper presents an alternative phase tuning approach that exploits the diversity among target distances and physiological parameters for multi-target detection. The design utilizes a voltage-controlled analog phase shifter to manipulate the phase correlation of the CWDR and thus create different signal mixtures from the multiple targets, then separates them in the frequency domain by suppressing individual signals sequentially. We implemented the phase correlation system based on a 2.4 GHz single-channel CWDR and evaluated it against multiple mechanical and human targets. The experimental results demonstrated successful separation of nearly equidistant targets within an antenna beam, equivalent to separating physiological signals of two people seated shoulder to shoulder.

РЕШЕНИЕ ЗАДАЧИ ОПТИМАЛЬНОГО РАДИОЛОКАЦИОННОГО РАСПОЗНАВАНИЯ ПОДВИЖНЫХ ВОЗДУШНЫХ ОБЪЕКТОВ НА ОСНОВЕ ТЕОРИИ ПРОВЕРКИ СТАТИСТИЧЕСКИХ ГИПОТЕЗ

The materials of this article are devoted to the development of one of the main aspects of artificial intelligence systems - pattern recognition. The relevance of the materials is due to the rapid development of radar systems for various purposes and the transition in some directions from radar to radio vision. Currently, much attention is paid to the development of radar systems with synthesizing the antenna aperture, for remote sensing of the earth and recognition of stationary ground objects, however, according to the author, radar recognition of mobile aerial objects is an important issue. The purpose of this article is to propose a solution to the problem of recognizing moving aerial objects by their radar portraits, based on the theory of statistical hypothesis testing. At the moment there are many methods of pattern recognition, this article discusses an algorithm that implements the function of matching the current image and the reference from a pre-formed catalog. As the current image, an azimuth-range radar portrait is considered, which is formed by super-resolution in azimuth, by synthesizing the aperture of the antenna and in range, using ultra-wideband signal. The author suggests, with a statistical approach to solving the problem of radar recognition, not to be tied to finding the probability of an object belonging to each of the pre-formed classes using a selected feature with a known probability distribution density of values, but to consider this process from the position of the signal at the output of the optimal recognition system to a specific image. A new approach to the description of probabilistic events when making a recognition decision is proposed. As a statistical classifier, it is proposed to use the Neumann-Pearson theory of statistical solutions.

SAR-PeGA: A Generation Method of Adversarial Examples for SAR Image Target Recognition Network

Deep learning (DL) is widely used in automatic target recognition (ATR) of synthetic aperture radar (SAR) images. Related researches show that DL models for SAR ATR are vulnerable to adversarial examples attack in the digital domain. However, how to generate adversarial examples in practical scenarios is critical and challenging. In this paper, we propose a systematic SAR perturbation generation algorithm (SAR-PeGA) for target recognition network. Firstly, assuming that some reflection phase tuning samples are located in the fixed area of SAR target, we adjust the phase characteristics of reflected signal with variable phase sequences. Secondly, we take the imperceptible perturbations from universal adversarial perturbations (UAP) as reference. Then, we construct the unconstrained minimum optimization model to find the specific phase sequences of tuning samples, and optimize the model with the adaptive moment estimation (Adam) optimizer. Finally, SAR adversarial examples can be flexibly generated through the proposed deceptive jamming model. Experimental results demonstrate that the proposed method can generate imperceptible jamming and effectively attack three classical recognition models.

Enabling Joint Communication and Radar Sensing in Mobile Networks—A Survey

Mobile network is evolving from a communication-only network towards one with joint communication and radar/radio sensing (JCAS) capabilities, that we call perceptive mobile network (PMN). Radio sensing here refers to information retrieval from received mobile signals for objects of interest in the environment surrounding the radio transceivers, and it may go beyond the functions of localization, tracking, and object recognition of traditional radar. In PMNs, JCAS integrates sensing into communications, sharing a majority of system modules and the same transmitted signals. The PMN is expected to provide a ubiquitous radio sensing platform and enable a vast number of novel smart applications, whilst providing non-compromised communications. In this paper, we present a broad picture of the motivation, methodologies, challenges, and research opportunities of realizing PMN, by providing a comprehensive survey for systems and technologies developed mainly in the last ten years. Beginning by reviewing the work on coexisting communication and radar systems, we highlight their limits on addressing the interference problem, and then introduce the JCAS technology. We then set up JCAS in the mobile network context and envisage its potential applications. We continue to provide a brief review of three types of JCAS systems, with particular attention to their differences in design philosophy. We then introduce a framework of PMN, including the system platform and infrastructure, three types of sensing operations, and signals usable for sensing. Subsequently, we discuss required system modifications to enable sensing on current communication-only infrastructure. Within the context of PMN, we review stimulating research problems and potential solutions, organized under nine topics: performance bounds, waveform optimization, antenna array design, clutter suppression, sensing parameter estimation, resolution of sensing ambiguity, pattern analysis, networked sensing under cellular topology, and sensing-assisted communications. We conclude the paper by listing key open research problems for the aforementioned topics and sharing some lessons that we have learned.

Research on the target recognition of marine surveillance radar based on ensemble learning

The existence of corner reflector decoy makes the marine surveillance radar to be caught in severe challenges of target recognition. In order to enhance the accuracy of recognition, an ensemble classifier for marine targets is created based on the self-built dataset of high resolution range profile (HRRP). In addition, a confidence evaluation algorithm based on non-parametric estimation of probability density is proposed to reject unknown decoy target outside the database. With taking different interference conditions into consideration respectively, the comparison experiment between ensemble classifier and single classifier is carried out. The results show that the ensemble classifier is significantly better, which has strong robustness to noise as well as rejection ability to unknown target, and the classification accuracy can reach 92.63% under ideal conditions. This paper proves the feasibility of ensemble learning for maritime target recognition, and provides a reliable classification algorithm when the sample information is sufficient.

Distribution Characteristics of Ground Echo Amplitude and Recognition of Signal Grazing Angle.

With the continuous advancement of electronic technology, terahertz technology has gradually been applied on radar. Since short wavelength causes severe ground clutter, this paper studies the amplitude distribution statistical characteristics of the terahertz radar clutter based on the measured data, and provides technical support for the radar clutter suppression. Clutter distribution is the function of the radar glancing angle. In order to achieve targeted suppression, in this paper, selected axial integral bispectrum (selected AIB) feature is selected as deep belief network (DBN)input to complete the radar glancing angle recognition and the network structure, network training method, robustness are analyzed also. The ground clutter amplitude distribution can follow normal distribution at 0~45° grazing angles. The Weibull distribution and G0 distribution can describe the amplitude probability density function of ground clutter at grazing angles 85° and 65°. The recognition rate of different signal grazing angles can reach 91% on three different terrains. At the same time, the wide applicability of the selected AIB feature is verified. The analysis results of ground clutter amplitude characteristics play an important role in the suppression of radar ground clutter.

Specific Radar Recognition Based on Characteristics of Emitted Radio Waveforms Using Convolutional Neural Networks.

With the increasing complexity of the electromagnetic environment and continuous development of radar technology we can expect a large number of modern radars using agile waveforms to appear on the battlefield in the near future. Effectively identifying these radar signals in electronic warfare systems only by relying on traditional recognition models poses a serious challenge. In response to the above problem, this paper proposes a recognition method of emitted radar signals with agile waveforms based on the convolutional neural network (CNN). These signals are measured in the electronic recognition receivers and processed into digital data, after which they undergo recognition. The implementation of this system is presented in a simulation environment with the help of a signal generator that has the ability to make changes in signal signatures earlier recognized and written in the emitter database. This article contains a description of the software’s components, learning subsystem and signal generator. The problem of teaching neural networks with the use of the graphics processing units and the way of choosing the learning coefficients are also outlined. The correctness of the CNN operation was tested using a simulation environment that verified the operation’s effectiveness in a noisy environment and in conditions where many radar signals that interfere with each other are present. The effectiveness results of the applied solutions and the possibilities of developing the method of learning and processing algorithms are presented by means of tables and appropriate figures. The experimental results demonstrate that the proposed method can effectively solve the problem of recognizing raw radar signals with agile time waveforms, and achieve correct probability of recognition at the level of 92–99%.