Neural Networks to Increase Range Resolution of FMCW Radar

Schutz, M.; Decroze, Cyril; Lalande, Michèle; Lenoir, B.

doi:10.1109/lsens.2020.3010015

Cited by 8 publications

(6 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…where M represents the batch size, i y represents the label value of the th i group in each batch, and ' i y represents the predicted output value [25]. The lower the MSE, the more accurate the network prediction [26].…”

Section: Network Trainingmentioning

confidence: 99%

A New BiRNN-SVA Method for Side Lobe Suppression

Liu,

Jia,

Liu

et al. 2024

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

The spatially variant apodization (SVA) algorithm, a classic super-resolution method for synthetic aperture radar (SAR) images, can suppress side lobes while maintain the resolution of the main lobe. To address the problem of residual side lobes or loss of main lobe energy in improved SVA algorithms, the paper proposes a new side lobe suppression method combining the bidirectional recurrent neural network (BiRNN) and the SVA algorithm, employing BiRNN to extract the main lobe and side lobe features of radar data to achieve side lobe suppression at any Nyquist sampling rate. The land flight experiment data of the fully polarized microwave scatterometer is used to quantitatively evaluate the side lobe suppression performance and the main lobe energy in order to verify the effectiveness of the BiRNN-SVA method. The experimental results demonstrate that the BiRNN-SVA method can be applied to data at any Nyquist sampling rate and has superior PSLR and ISLR compared to the GSVA algorithm and MSVA algorithm. The image processed with the proposed method retains more fine details and edge features. In comparison to the GSVA algorithm and MSVA algorithm, the image contrast (IE) and focus have increased by 31.6% and 3.6%, respectively, and by 4.4% and 1.1%.

show abstract

Section: Network Trainingmentioning

confidence: 99%

A New BiRNN-SVA Method for Side Lobe Suppression

Liu,

Jia,

Liu

et al. 2024

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

show abstract

“…In addition, a neural network does not have the resolution limitation as in conventional DFT signal processing. In fact, recent research has shown that super-resolution of Range, Doppler and AoA can be achieved through deep neural networks for raw radar signals [8,25]. Furthermore, the pre-processing module could be trained to extract more informative features, rather than noise and environmental reflections, potentially carrying more useful information to the downstream classifier.…”

Section: Cubelearn Module Architecturementioning

confidence: 99%

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

Zhao¹,

Lu²,

Wang³

et al. 2021

Preprint

View full text Add to dashboard Cite

mmWave FMCW radar has attracted huge amount of research interest for human-centered applications in recent years, such as human gesture/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 learnable preprocessing 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 previously weaker models. 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. CCS CONCEPTS• Computing methodologies → Machine learning.

show abstract

“…Recently, algorithms for improving range resolution through artificial intelligence have been developed [ 22 , 23 ]. Through deep learning, a time domain signal containing a beat frequency is learned and so the point of the signal is expanded to improve the range resolution.…”

Section: Introductionmentioning

confidence: 99%

“…Breath monitoring (BM) extracts the feature of the signal after STFT (Short-Time Fourier Transform) of the signal and detects the situation through a Support Vector Machine (SVM) [ 21 ]. (Neural Networks (NNs) extend time signals through artificial intelligence [ 22 ], and finally, deep learning-2D (DL-2D) reduces the estimation error of objects measured through 2D DNNs [ 23 ]. Through this, the performance of existing studies can be confirmed.…”

Section: Introductionmentioning

confidence: 99%

Method for Improving Range Resolution of Indoor FMCW Radar Systems Using DNN

Park

Kim

Jung

et al. 2022

Sensors

View full text Add to dashboard Cite

Various studies on object detection are being conducted, and in this regard, research on frequency-modulated continuous wave (FMCW) RADAR is being actively conducted. FMCW RADAR requires high-distance resolution to accurately detect objects. However, if the distance resolution is high, a high-modulation bandwidth is required, which has a prohibitively high cost. To address this issue, we propose a two-step algorithm to detect the location of an object through DNN using many low-cost FMCW RADARs. The algorithm first infers the sector by measuring the distance to the object for each FMCW RADAR and then measures the position through the grid according to the inferred sector. This improves the distance resolution beyond the modulation bandwidth. Additionally, to detect multiple targets, we propose a Gaussian filter. Multiple targets are detected through an ordered-statistic constant false-alarm rate (OS-CFAR), and there is an 11% probability that multiple targets cannot be detected. In the lattice structure proposed in this paper, the performance of the proposed algorithm compared to those in existing works was confirmed with respect to the cost function. The difference in performance versus complexity was also confirmed when the proposed algorithm had the same complexity and the same performance, and it was confirmed that there was a performance improvement of up to five-fold compared to those in previous papers. In addition, multi-target detection was shown in this paper. Through MATLAB simulation and actual measurement on a single target, RMSEs were 0.3542 and 0.41002 m, respectively, and through MATLAB simulation and actual measurement on multiple targets, RMSEs were confirmed to be 0.548265 and 0.762542 m, respectively. Through this, it was confirmed that this algorithm works in real RADAR.

show abstract

Neural Networks to Increase Range Resolution of FMCW Radar

Cited by 8 publications

References 12 publications

A New BiRNN-SVA Method for Side Lobe Suppression

A New BiRNN-SVA Method for Side Lobe Suppression

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

Method for Improving Range Resolution of Indoor FMCW Radar Systems Using DNN

Contact Info

Product

Resources

About