اشترك بالحزمة الذهبية واحصل على وصول غير محدود شمرا أكاديميا
تسجيل مستخدم جديدDeep Learning Approach for Target Locating in Through-the-Wall Radar under Electromagnetic Complex Wall
57
0
0.0
(
0
)
اسأل ChatGPT حول البحث
ﻻ يوجد ملخص باللغة العربية
In this paper, we used the deep learning approach to perform two-dimensional, multi-target locating in Throughthe-Wall Radar under conditions where the wall is modeled as a complex electromagnetic media. We have assumed 5 models for the wall and 3 modes for the number of targets. The target modes are single, double and triple. The wall scenarios are homogeneous wall, wall with airgap, inhomogeneous wall, anisotropic wall and inhomogeneous-anisotropic wall. For this purpose, we have used the deep neural network algorithm. Using the Python FDTD library, we generated a dataset, and then modeled it with deep learning. Assuming the wall as a complex electromagnetic media, we achieved 97:7% accuracy for single-target 2D locating, and for two-targets, three-targets we achieved an accuracy of 94:1% and 62:2%, respectively.
قيم البحث
اقرأ أيضاً
Radar images of humans and other concealed objects are considerably distorted by attenuation, refraction and multipath clutter in indoor through-wall environments. While several methods have been proposed for removing target independent static and dy
namic clutter, there still remain considerable challenges in mitigating target dependent clutter especially when the knowledge of the exact propagation characteristics or analytical framework is unavailable. In this work we focus on mitigating wall effects using a machine learning based solution -- denoising autoencoders -- that does not require prior information of the wall parameters or room geometry. Instead, the method relies on the availability of a large volume of training radar images gathered in through-wall conditions and the corresponding clean images captured in line-of-sight conditions. During the training phase, the autoencoder learns how to denoise the corrupted through-wall images in order to resemble the free space images. We have validated the performance of the proposed solution for both static and dynamic human subjects. The frontal radar images of static targets are obtained by processing wideband planar array measurement data with two-dimensional array and range processing. The frontal radar images of dynamic targets are simulated using narrowband planar array data processed with two-dimensional array and Doppler processing. In both simulation and measurement processes, we incorporate considerable diversity in the target and propagation conditions. Our experimental results, from both simulation and measurement data, show that the denoised images are considerably more similar to the free-space images when compared to the original through-wall images.
Cognitive radars are systems that rely on learning through interactions of the radar with the surrounding environment. To realize this, radar transmit parameters can be adapted such that they facilitate some downstream task. This paper proposes the u
se of deep reinforcement learning (RL) to learn policies for gain control under the object detection task. The YOLOv3 single-shot object detector is used for the downstream task and will be concurrently used alongside the RL agent. Furthermore, a synthetic dataset is introduced which models the radar environment with use of the Grand Theft Auto V game engine. This approach allows for simulation of vast amounts of data with flexible assignment of the radar parameters to aid in the active learning process.
Millimeter-wave (mmWave) radars are being increasingly integrated in commercial vehicles to support new Adaptive Driver Assisted Systems (ADAS) features that require accurate location and Doppler velocity estimates of objects, independent of environm
ental conditions. To explore radar-based ADAS applications, we have updated our test-bed with Texas Instruments 4-chip cascaded FMCW radar (TIDEP-01012) that forms a non-uniform 2D MIMO virtual array. In this paper, we develop the necessary received signal models for applying different direction of arrival (DoA) estimation algorithms and experimentally validating their performance on formed virtual array under controlled scenarios. To test the robustness of mmWave radars under adverse weather conditions, we collected raw radar dataset (I-Q samples post demodulated) for various objects by a driven vehicle-mounted platform, specifically for snowy and foggy situations where cameras are largely ineffective. Initial results from radar imaging algorithms to this dataset are presented.
The existence of multipath brings extra looks of targets. This paper considers the extended target detection problem with a narrow band Multiple-Input Multiple-Output(MIMO) radar in the presence of multipath from the view of waveform-filter design. T
he goal is to maximize the worst-case Signal-to-Interference-pulse-Noise Ratio(SINR) at the receiver against the uncertainties of the target and multipath reflection coefficients. Moreover, a Constant Modulus Constraint(CMC) is imposed on the transmit waveform to meet the actual demands of radar. Two types of uncertainty sets are taken into consideration. One is the spherical uncertainty set. In this case, the max-min waveform-filter design problem belongs to the non-convex concave minimax problems, and the inner minimization problem is converted to a maximization problem based on Lagrange duality with the strong duality property. Then the optimal waveform is optimized with Semi-Definite Relaxation(SDR) and randomization schemes. Therefore, we call the optimization algorithm Duality Maximization Semi-Definite Relaxation(DMSDR). Additionally, we further study the case of annular uncertainty set which belongs to non-convex non-concave minimax problems. In order to address it, the SDR is utilized to approximate the inner minimization problem with a convex problem, then the inner minimization problem is reformulated as a maximization problem based on Lagrange duality. We resort to a sequential optimization procedure alternating between two SDR problems to optimize the covariance matrix of transmit waveform and receive filter, so we call the algorithm Duality Maximization Double Semi-Definite Relaxation(DMDSDR). The convergences of DMDSDR are proved theoretically. Finally, numerical results highlight the effectiveness and competitiveness of the proposed algorithms as well as the optimized waveform-filter pair.
We revisit the electromagnetic form factors of the proton and neutron in the holographic soft wall model. At low momentum transfer, we show that by matching the nucleon and rho Regge slopes and fixing the nucleon anomalous dimension by the nucleon ma
ss, a perfect match to the world average charge radii from e-p scattering (including the recent small charge radius of the proton measured by the PRad collaboration at JLab) follows. At high momentum transfer, the nucleon anomalous dimension runs up to match the hard scaling rule.
سجل دخول لتتمكن من نشر تعليقات
التعليقات
جاري جلب التعليقات
سجل دخول لتتمكن من متابعة معايير البحث التي قمت باختيارها
