This study presents a numerical modeling approach that utilizes millimeter-wave (mm-Wave) Frequency-Modulated Continuous-Wave (FMCW) radar to reconstruct and classify five weapon types: grenades, knives, guns, iron rods, and wrenches. A dataset of 1000 images of these weapons was collected from various online sources and subsequently used to generate 3605 samples in the MATLAB (R2022b) environment for creating reflectivity-added images. Background reflectivity was considered to range from 0 to 0.3 (with 0 being a perfect absorber), while object reflectivity was set between 0.8 and 1 (with 1 representing a perfect electric conductor). These images were employed to reconstruct high-resolution weapon profiles using a monostatic two-dimensional (2D) Synthetic Aperture Radar (SAR) imaging technique. Subsequently, the reconstructed images were classified using a Convolutional Neural Network (CNN) algorithm in a Python (3.10.14) environment. The CNN architecture consists of 10 layers, including multiple convolutional, pooling, and fully connected layers, designed to effectively extract features and perform classification. The CNN model achieved high accuracy, with precision and recall values exceeding 98% across most categories, demonstrating the robustness and reliability of the model. This approach shows considerable promise for enhancing security screening technologies across a range of applications.

Achieving high imaging resolution in conventional monostatic radar imaging with mechanical scanning requires excessive acquisition time. Although real aperture radar systems might not suffer from such a limitation in acquisition time, they may still face challenges in achieving high imaging resolution, especially in near-field (NF) scenarios, due to diffraction-limited performance. Even with sophisticated electronic scanning techniques, increasing the aperture size to improve resolution can lead to complex hardware setups and may not always be feasible in certain practical scenarios. Multistatic systems can virtually increase the effective aperture but introduce challenges due to the required number of antennas and channels, making them expensive, bulky and power-intensive. An alternative solution that has been proposed in recent years is the compression of the physical layer using metasurface transducers. This paper presents a novel NF radar imaging approach leveraging dynamic metasurface antennas with multiple tuning states called masks, in a bistatic structure, using the Kirchhoff migration principle. The method involves expanding the compressed measured signal from the mask-frequency domain to the spatial-frequency domain to decode the scene's spatial content. The Kirchhoff integral is then developed based on the introduced special imaging structure to retrieve the three-dimensional spatial information of the target. Comprehensive numerical simulations analyze the masks' characteristics and their behavior under different conditions. The performance of the image reconstruction algorithm is evaluated for visual quality and computing time using both central processing units and graphics processing units. The results of computer simulations confirm the high reliability of the proposed approach in various cases.

In recent decades, microwave imaging technology has been used in a variety of applications including security, medicine, nondestructive testing and structural health monitoring. Traditional microwave imaging systems often suffer from drawbacks such as long acquisition times and complex array structures. To address these issues, this paper introduces a panel-to-panel microwave computational imaging (CI) technique for near-field operation using dynamic metasurface antennas for both transmission and reception, enhancing system diversity and enabling real-time applications. The paper also outlines mathematical models for three-dimensional image reconstruction algorithms tailored to this scenario. The results of numerical and electromagnetic simulations show the feasibility of this approach for CI-based imaging, with both Fourier and least squares-based image reconstruction techniques.

Machine learning (ML) techniques have been applied for radar applications in recent years. It is still changeling to classify images or objects accurately. This work has modeled 1600 reconstructed object shapes of four different objects like triangles, circles, squares, and rectangles using millimeter-wave (mmWave) FMCW radar principle based on the 2D SAR imaging technique, and the numerical analysis is performed in MATLAB. The Convolution Neural Network (CNN) technique is implemented to perform the objects’ classification in a Python environment. The results give a good prospect for the study of ML techniques to classify mmWave FMCW radar data.

Millimeter-Wave FMCW radar offers highresolution imaging and object detection capabilities, making it ideal for various applications including automotive radar, industrial sensing, security systems, and Imaging. This work presents the modeling of mmWave FMCW radar system for 2D SAR imaging for different objects like square, rectangle, triangle, and circle. The main objective is to reconstruct the object shapes with high resolution by developing numerical analysis. Utilizing MATLAB, the system is modeled and analyzed to achieve high-resolution imaging. The results show good prospects for the use of mmWave FMCW radar systems for imaging multiple objects with different shapes for SAR imaging.

The content of this chapter provides a thoughtful analysis of case studies that highlight the detection capabilities of FMCW radar systems operating in mmWave configurations. The case studies demonstrate how mmWave FMCW radar technology may be used to detect objects, motions, and changes in both line of sight and non-line of sight settings with accuracy, and efficiency. Each case study explores the unique difficulties presented by the application environment, which can include anything from identifying impediments in automotive safety systems to detecting minute movements for vital sign monitoring in healthcare. The steps for detecting different gesture recognition using IWR 1843 BOOST FMCW radar system and its processing are focussed upon. The document highlights the technology's excellent resolution, motion sensitivity, and adaptability to a variety of challenging environments in LoS and NLoS scenarios and with the technical details of operating mmWave radars, including signal processing methods, machine learning algorithms, and mitigating interference from surrounding objects.

Direction of arrival (DoA) estimation plays a crucial role in channel characterization and is a critical step to execute the necessary beam-shaping operations needed from antennas present within a wireless environment. Conventional DoA estimation frameworks rely on array-based topologies, making use of the phase difference information between the individual channels to retrieve the DoA data. Recently, the idea of computational imaging has been shown to offer a promising solution to conventional raster-scan-based techniques. This is mainly due to the physical layer compression facilitated by special types of compressive apertures (or surfaces) that leverage the idea of synthesizing quasi-orthogonal radiation patterns (modes) to probe and encode the scene information in an indirect manner and compress it into a single channel (or a reduced number of channels). The application of computational imaging to the channel characterization problem is intriguing. Yet, a system level of knowledge of the design parameters, which are key to understanding the development of compressive surfaces for computational DoA estimation, is far from comprehensive. In this paper, we demonstrate different techniques to synthesize spatio-temporally incoherent field patterns, a key requirement for computational DoA estimation, and provide a study of the different system level parameters needed to design such antennas. We show that by increasing the orthogonality of the radiated modes (and thus reducing the information redundancy), a single-channel compressive antenna can retrieve the DoA pattern of multiple far-field sources even under a signal-to-noise ratio (SNR) level of as low as 0 dB, without the necessity to use a multi-channel physical architecture.

Frequency domain image reconstruction algorithms offer significant advantages, especially for applications for which the reconstruction time is crucial. In the frequency domain, image reconstruction can be realized using fast Fourier transformations, reducing the complexity and thus the execution time of the reconstruction algorithm. In this paper, we adopt range migration techniques to reconstruct radar images from numerical and experimental data. The aim is to examine the robustness of the range migration algorithm (RMA) as a function of sampling sparsity under varying noise levels. Considering that sampling at Nyquist rates can be quite challenging for the conventional synthetic aperture radar (SAR) acquisition, we investigate the behavior of the reconstruction algorithm for larger sampling steps.

In this article, to accelerate data acquisition and image reconstruction procedures in a multistatic short-range microwave imaging scenario, an orthogonal coding approach with Fourier domain processing is presented. First, a special 2-D multiple-input multiple-output (MIMO) structure is introduced to fully electronically synthesize the 2-D aperture. Then, the model of the transmitted and received signals by a MIMO stepped-frequency-modulated radar is presented, with special considerations about orthogonal, balanced, and optimal sequences. On the receiver side, the backscatter frequency response extraction process is formulated with the aim of obtaining individual information of all channels. Finally, based on the introduced model, a fast Fourier-based algorithm with reduced dimensions, named MIMO coded generalized reduced dimension Fourier (CGRDF), is mathematically derived. It includes extracting phase and amplitude compensators with the aim of mapping 4-D to 2-D spatial data, transferring the backscatter transfer function from the spatial domain to the wavenumber domain, extracting the smoothing filter, compensating the curvature of the wavefront of all scatterers, and extracting the reflectivity function and an additional range compensator. The results of numerical simulations show the satisfactory and reliable performance of the proposed approach in terms of the information retrieval process and processing speed.

This paper presents an analysis of a 1-bit unit cell for reflectarray antennas in X-band applications. Reflectarray antennas offer advantages such as lower profiles and lighter weights compared to traditional antennas. The unit cell is made up of four arrow-shaped structures with a single head, enabling the manipulation of electromagnetic waves and beam steering capabilities. The unit cell operates in two modes, bit-0 and bit-l. Simulation results demonstrate that the unit cell achieves a phase difference of 180 degrees for the co-polarization reflection coefficient across a wide range of frequency. Additionally, the paper discusses the potential use of tunable elements and graphene for switching operations to reduce fabrication costs. This research contributes to the development of compact and efficient reflectarray antennas for X-band applications.

Millimetre-wave (mmW) reconstructed images are of complex-valued in nature, suggesting that they contain both magnitude and phase. It is known that from the phase aspect of the reconstructed images, meaningful feature information can be extracted about the imaged objects, which in turn, is beneficial to solve computer vision problems such as classification. To this end, a comparative study is shown in this paper wherein two Convolutional Neural Network (CNN) models are considered: one trained with magnitude aspect of mmW reconstructed images, and the other is trained with both the magnitude and the phase aspects of mmW reconstructed images. After training, when these two models are tested, a higher classification accuracy is obtained in the performance of the classification model trained with both the magnitude and phase information of mmW images, as compared to the other model.

This chapter presents the machine learning (ML) concept for standard RF component design in microwave frequency. It will explain the use of the deep machine learning concept for antenna and other RF components, such as RF filter, and all relevant analysis will be based on the CST simulations. The comparative study of the ML approach and the antenna design tool (such as CST) will be presented in the form of their performance. This chapter will explain the perspectives of ML in RF system design and analysis. This chapter will present the design of the antenna and filter as examples. The simulated results will be obtained by the CST MW Studio and the ML will be implemented in MATLAB.

Imaging systems leveraging millimetre-wave (mmW) frequencies have several advantages, however, such systems suffer from poor resolution images as compared to higher frequency reconstructions such as in optical regime. Also, practical radar systems are susceptible to noise such as clutter, thermal noise, motion blurs, etc. To recover the original mmW image from these poorly resolved noisy images, two individual image processing steps are required, that is, super-resolution and denoising. This paper focuses on using a complex-valued convolutional neural network (CV-CNN) to combine the two individual processing steps into one single algorithm. By designing the CV-CNN to accommodate complex-valued reconstruction data, the phase information content of the input images, along with the magnitude information, is considered in the process. A computational imaging (CI) numerical model, instead of an experimental imaging system, is used to train and test the neural network. By comparing the performance metrics of the final reconstruction images, it is observed that the developed CV-CNN can resolve and de-noise the poorly resolved noisy input mmW images to a high degree of fidelity.

A scheme for reducing the need of multi-scan measurements of near electric-field information to a single-scan process is presented to simplify the construction of the sensing matrix required for performing computational microwave imaging system at K-band frequencies. For this purpose, we propose to perform only a single scan, and the obtained information is used to construct the multiple radiated near electric-field information associated to all transmit and receive units. Further, the performance of the constructed near electric-field is verified by generating an image of a test target by means of computational imaging using the scene reflectivity information. The proposed scheme helps in reducing the overall complexity of the nearfield scanning process associated to the computational imaging system.

The emerging technology of dynamic metasurface antennas (DMAs) offers a promising solution to revolutionize future wireless communications by reducing hardware costs, physical size and power consumption. Especially in the field of radar imaging, DMAs can be used as an effective alternative platform for modern computational imaging; because they can simplify the physical hardware architecture and increase the data acquisition rate. Fourier transform (FT)-based scene image reconstruction techniques are known as cost-effective computing solutions for the imaging system processing unit. However, due to the physical layer compression in DMAs and the fact that they do not produce uniform radiation patterns, the information provided by them is not compatible with Fourier-based techniques and cannot be applied directly. In this article, we first introduce a 3D near-field bistatic imaging approach using two one-dimensional (1D) DMAs as a panel-to-panel model in a Mills Cross structure. Then, based on the introduced mathematical model, we derive a Fourier-based algorithm for the image reconstruction problem. The proposed algorithm consists of five main steps: (i) pre-processing (to transfer the data provided by the transmitter and receiver DMAs to a set of equivalent spatial measurements), (ii) applying FTs to the transferred signal, (iii) filtering in the Fourier domain, (iv) a simplified interpolation, and (v) applying a 3D inverse FT to retrieve scene information. The results of numerical simulations confirm the satisfactory performance of the proposed approach.

In this paper, a preliminary study related to the detection of breast cancer based on a computational microwave imaging system is presented at K-band frequencies.In comparison to normal tissues, the different dielectric properties (permittivity and conductivity) of the malignant tissues can be exploited in order to detect the presence of tumour through a microwave imaging system.This work demonstrates the detection of breast tumour as an application of a computational imaging technique by leveraging the concept of the dynamic metasurface antenna (DMA) aperture as a transmitter.In this framework, the computational imaging aperture can provide a compact system design with a low-cost deployment for detecting tumours in early-stage, and this can speed up the screening process of the population at risk.Particularly, early detection of Stage 1 can help with the determination of the best treatment method and enhance a patient's prognosis.

The unique characteristics of the millimeter-wave (mmW) frequency band have led to its widespread use in various fields such as communications, imaging, and wireless sensing. This paper addresses two different mmW imaging structures, monostatic and multistatic, in the face of a sparse spatial sampling scenario. By using compressive sensing theory, a solution for image reconstruction, consistent with fast Fourier-based techniques, is presented with compressed data obtained from monostatic imaging. This solution is then generalized to a multiple-input multiple-output (MIMO) imaging case using a multistatic-to-monostatic conversion. Reconstructed images from numerical and experimental data show the satisfactory performance of the presented approach.

A two-dimensional (2D) dynamically reconfigurable metasurface aperture is presented to perform frequency selective through wall imaging (TWI) with an unknown structure of the wall.Generally, in TWI, the medium properties and thickness of the wall need to be known in advance, which is not always possible.Moreover, compensating for these effects can significantly increase the computational complexity.We propose a two-stage method that leverages the concept of a dynamically reconfigurable metasurface antenna (DMA) in a narrow frequency band in which the effects of the wall are minimum to perform TWI.First, two simple probe antennas are used to evaluate the reflection response of the wall by means of a simple backscatter measurement.Based on these characteristics, a narrow band frequency selective window is identified.Second, a DMA consisting of an array of tunable metamaterial elements is used for TWI in the identified frequency selective window.The DMA aperture enables the scene information to be sampled through a set of spatio-temporally varying quasi-random modes using a single-channel transmit and receive architecture.This physical-layer compression scheme can significantly simplify the data acquisition while the quasi-random sampling of the scene information eliminates the need for conventional raster-scan based modalities.

Electromagnetic (EM) waves at millimeter-wave (mmW) frequencies have found applications in a variety of imaging systems, from security screening to defense and automotive radars, with the research and development of mmW imaging systems gaining interest in recent years. Despite their significant advantages, mmW imaging systems suffer from poor resolution compared to higher frequency reconstructions, such as optical images. To improve the resolution of mmW images, various super-resolution (SR) techniques have been introduced. One such technique is the use of machine learning algorithms in the signal processing layer of the imaging system without altering any of the system’s parameters. This article focuses on the use of a convolutional neural network (CNN) architecture to achieve SR when applied to 3-D mmW input images. To exploit the phase information content of the input images along with the magnitude, a complex-valued CNN is designed, which can accommodate complex-valued data. To simplify the learning process, the resolution difference between the input and output images is divided into smaller parts by using subnetworks in the CNN architecture. The trained model is tested on simulated and experimental targets. The average mean square error score and the structural similarity index obtained on a test dataset of 460 samples are 0.0127 and 0.9225, respectively. It can be inferred that the model has the capability to improve the resolution of input mmW images to a high degree of fidelity, hence paving the way for an end-to-end SR imaging system.

Microwave frequencies based on frequency-modulated continuous-wave (FMCW) technique as applied to synthetic aperture radar (SAR) imaging have many potential applications related to security, automotive, mapping, medical, and surveillance. In this paper, a three-dimensional (3D) SAR imaging system at K band frequencies is presented and its effective performance is verified with the reconstructed images from a set of uniformly sampled data from the scene. The collected backscattered signals are measured over a 2D plane, in both horizontal and vertical directions, using a Nearfield Systems Inc. (NSI) measurement platform. This paper presents the complete details about the measurement steps along with the signal processing SAR technique for 3D image reconstruction. The presented work demonstrates the reconstruction of image for two different targets. The performance of the work is also verified with the 3D reconstructed image of one target in a concealed scenario.