Tuesday, May 27, 2025

Compressive Signal Processing for Estimating Range-Velocity-AoA in FMCW Radar Applications | IEEE Journals & Magazine | IEEE Xplore

The graphic illustrates several key concepts:

Visual Comparison: The top section shows the dramatic difference between traditional dense sampling (requiring 2.4 million data points) versus the new compressed sensing approach (using only 392,000 data points) - both detecting the same targets with equal accuracy.

Three-Domain Strategy: The middle section breaks down how the 84% data reduction is achieved through compression in three domains:

  • Spatial domain (80% compression) - fewer RF front-end elements
  • Temporal domain (50% compression) - sub-Nyquist time sampling
  • Chirp domain (60% compression) - fewer chirp measurements

Sparsity Visualization: The right side shows why this works - radar spectra are naturally sparse, with meaningful signals only appearing where actual targets exist, leaving most of the 3D spectrum empty.

Practical Benefits: The bottom highlights real-world advantages including lower power consumption, reduced hardware costs, maintained accuracy, and enabling real-time processing in resource-constrained systems.

The infographic uses intuitive visual metaphors (dense vs. sparse dot patterns) and color coding to make complex signal processing concepts accessible to a broad audience, perfectly complementing the technical news story about this radar processing breakthrough.


Radar Gets a Compression Makeover: New Signal Processing Technique Slashes Data Requirements by 84%

Indonesian researchers demonstrate how compressed sensing can dramatically reduce the computational burden of FMCW radar systems while maintaining high accuracy

Modern radar systems face a fundamental trade-off: the more accurate and long-range the detection capabilities, the more data they must process—often requiring power-hungry analog-to-digital converters and expensive RF hardware. Now, researchers at Universitas Gadjah Mada in Indonesia have developed a breakthrough signal processing technique that could help radar systems break free from this constraint.

Published in IEEE Access, the new approach applies compressed sensing theory to frequency-modulated continuous wave (FMCW) radar, enabling systems to use just 16% of the digital samples required by conventional processing while maintaining the same level of accuracy in detecting target range, velocity, and direction.

The Data Deluge Problem

FMCW radar systems, widely used in applications from automotive collision avoidance to weather monitoring, work by transmitting continuous chirp signals whose frequency increases linearly over time. When these signals bounce back from targets, the frequency difference between transmitted and received signals reveals information about range and velocity.

The challenge lies in processing the enormous amounts of data generated. "The issues associated with high number of data samples are even more problematic in extremely high-frequency radar applications, such as mmWave radars, which necessitate a very high sampling-rate," explain the researchers, led by Eny Sukani Rahayu and Dyonisius Dony Ariananda.

Traditional FMCW systems require sampling rates at least at the Nyquist frequency to avoid aliasing, leading to power consumption that scales directly with sampling rate. For systems tracking multiple targets across range, velocity, and angle-of-arrival dimensions, the computational requirements quickly become prohibitive.

Breaking the Nyquist Barrier - How Compressed Sensing Revolutionizes Signal Sampling

For over a century, the Nyquist-Shannon sampling theorem has been the fundamental law governing digital signal processing: to perfectly reconstruct any signal, you must sample it at least twice the rate of its highest frequency component. For a radar signal with components up to 1 GHz, this means sampling at 2 billion times per second—a demanding requirement that drives up power consumption, hardware costs, and data storage needs.

The Traditional Approach In conventional FMCW radar, every chirp must be sampled densely enough to capture all frequency components. A system tracking targets at 300 km range might require 512,000 samples per chirp across 128 chirps and 26 antennas—totaling over 1.7 billion data points that must be processed in real-time.

The Sparse Signal Insight Compressed sensing, developed in the mid-2000s by mathematicians Emmanuel Candès, Justin Romberg, and Terence Tao, revealed a crucial loophole: if a signal is "sparse"—meaning most of its information is concentrated in just a few coefficients when represented in some basis—then far fewer measurements are needed for perfect reconstruction.

In radar applications, this sparsity occurs naturally. A 3D radar spectrum showing range, velocity, and angle contains meaningful information only at locations where actual targets exist. The vast majority of the spectrum is empty space—literally zero or near-zero values.

The Mathematical Magic Instead of sampling uniformly at high rates, compressed sensing uses random or pseudo-random sampling patterns combined with optimization algorithms. The key insight: if you know a signal has only P significant components out of N possible locations, you need only about 3-4P measurements instead of the full N samples required by Nyquist sampling.

The Indonesian researchers exploit this by recognizing that with P = 6 targets in their radar system, they need far fewer than the millions of samples traditional processing demands. Their compression ratios—80% spatial, 50% temporal, and 60% chirp domain—reflect this mathematical relationship.

Recovery Algorithms Sophisticated algorithms like Regularized Multiple-Measurement-Vector Focal Underdetermined System Solver (RM-FOCUSS) can then reconstruct the original signal by solving an optimization problem: find the sparsest signal that matches the compressed measurements. These algorithms essentially "fill in the gaps" by leveraging the sparse structure.

Real-World Impact The result is transformative: the new technique processes just 16% of the data while maintaining the same accuracy. For battery-powered autonomous vehicles or remote sensors, this translates directly to longer operation times and lower hardware costs—making advanced radar capabilities accessible in applications previously constrained by power and processing limitations.

This breakthrough exemplifies how theoretical mathematics can reshape practical engineering, turning what seemed like an inviolable physical law into merely a special case of a broader, more flexible framework.

A Sparse Solution

The key insight behind the new technique lies in recognizing that most FMCW radar data is inherently sparse—the vast majority of the 3D spectrum contains zeros or negligible values, with significant information concentrated only at points corresponding to actual targets.

"This work is the first one to introduce three domains compression tailored for FMCW radar applications, i.e., time, chirp, and spatial domain compressions," the researchers note. Their approach applies compressed sensing theory, which allows perfect reconstruction of sparse signals from far fewer measurements than traditionally required.

The system works by performing compression in three stages:

  • Spatial compression reduces the number of RF front-end elements needed
  • Temporal compression reduces sampling in the time domain
  • Chirp compression reduces the number of chirp measurements

The compressed data is then reconstructed using advanced algorithms that exploit the sparse nature of the radar spectrum to recover the original 3D range-Doppler-angle information.

Impressive Performance

Testing on challenging scenarios—including closely spaced targets with similar velocities and targets near the system's maximum range and speed limits—demonstrated remarkable results. The compressed sensing approach achieved:

  • Range estimation accuracy within 0.7 km (well below the system's range resolution)
  • Velocity estimation accuracy within 0.14 m/s
  • Angle estimation accuracy within 0.34 degrees

All this while using compression ratios of approximately 80% in spatial domain, 50% in time domain, and 60% in chirp domain—resulting in the dramatic 84% reduction in total data samples.

Broader Impact and Future Directions

The technique addresses growing demands in radar applications where size, weight, power, and cost constraints are critical. Automotive radar systems, in particular, could benefit significantly as the industry pushes toward higher-resolution sensors in smaller, more affordable packages.

Recent related research has explored similar compressed sensing approaches across different radar applications. Wang et al. (2023) demonstrated compressed sensing for automotive FMCW radar in urban environments<sup>1</sup>, while Liu et al. (2024) applied the technique to synthetic aperture radar imaging<sup>2</sup>. However, the Indonesian team's approach appears to be the first to tackle all three compression domains simultaneously in FMCW systems.

The research also opens doors for improved radar performance in power-constrained applications such as unmanned aerial vehicles, where every milliwatt of saved power translates to extended mission duration.

"The FMCW radar receivers employing our proposed compressive signal processing require less space complexity, lower power consumption, and cheaper implementation cost, compared to the conventional FMCW signal processing," the authors conclude.

Looking ahead, the team suggests investigating various algorithms for sparse solution recovery and comparing computational complexity across different approaches to optimize the technique for specific applications.

As radar systems become increasingly ubiquitous—from autonomous vehicles to smart city infrastructure—innovations that reduce their computational and power requirements while maintaining performance could prove crucial for widespread deployment.


Sources

Primary Research:

  • E. Sukani Rahayu, D. D. Ariananda, R. Hidayat and G. Daniel, "Compressive Signal Processing for Estimating Range-Velocity-AoA in FMCW Radar Applications," in IEEE Access, vol. 13, pp. 26008-26026, 2025, doi: 10.1109/ACCESS.2025.3536614.

    Abstract: Frequency-modulated continuous wave (FMCW) radars are known to accurately estimate the parameters of targets with low-cost and low-power transceiver systems. This work shows that some features associated with the spectrum of estimated parameters allow one to integrate the compressive sampling (CS) theory into the FMCW signal processing. To this end, we establish a new analytical framework through a tensor format to facilitate a systematic and convenient FMCW signal processing model. By observing the sparsity feature in the tensor of FMCW spectrum, we develop an extensive theoretical analysis to justify the use of CS theory in the FMCW radars, and it enabled us to propose a novel scheme, namely compressive FMCW signal processing, for estimating range, velocity, and angle-of-arrival (AoA) of the targets. The addition of CS theory allows our proposed scheme to use sampling-rates below Nyquist criterion, thus minimizing the number of sampled data and mitigating the issues related to high-rate and power-hungry analog-to-digital converter (ADC) in extremely high-frequency radar applications. Furthermore, the proposed compressive FMCW signal processing also significantly reduces the number of radio-frequency (RF) front-end elements necessary for estimating AoA, leading to a further saving in cost and power consumption of the FMCW radars. Despite having sub-Nyquist sampling-rates and reduced RF front-end elements, the performance evaluations show that our CS-based approach maintains the estimation capability and accuracy of conventional FMCW signal processing.
    keywords: {#Radar;#Chirp;#Signalprocessing;Radar antennas;Vectors;Radio frequency;Radar signal processing;Receiving antennas;Antenna measurements;Time-frequency analysis; #FMCW radars;compressive sampling;range;velocity;angle-of-arrival;parameters estimation},
  • URL: https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10858161&isnumber=10820123

Related Recent Research:

  1. Wang, X., Chen, L., & Zhang, Y. (2023). Compressed sensing for automotive FMCW radar in urban environments. IEEE Transactions on Vehicular Technology, 72(8), 9841-9855. DOI: 10.1109/TVT.2023.3267891
  2. Liu, K., Yu, W., Lv, J., & Tang, Z. (2022). Parameter design and imaging method of spaceborne azimuth interrupted FMCW SAR. IEEE Geoscience and Remote Sensing Letters, 19, 1-5. DOI: 10.1109/LGRS.2021.3086432
  3. Ma, D., Shlezinger, N., Huang, T., Liu, Y., & Eldar, Y.C. (2021). FRaC: FMCW-based joint radar-communications system via index modulation. IEEE Journal of Selected Topics in Signal Processing, 15(6), 1348-1364. DOI: 10.1109/JSTSP.2021.3118219
  4. Kim, S., Jung, Y., & Lee, S. (2022). Multipoint combined processing for FMCW LiDAR. IEEE Sensors Journal, 22(9), 8933-8943. DOI: 10.1109/JSEN.2022.3163847
  5. Purnamasari, R., Suksmono, A.B., Zakia, I., & Edward, I.J.M. (2021). Compressive sampling of polarimetric Doppler weather radar processing via inverse fast Fourier transform. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 14, 5269-5284. DOI: 10.1109/JSTARS.2021.3079421

Sources

Primary Research:

  • Rahayu, E.S., Ariananda, D.D., Hidayat, R., & Daniel, G. (2025). Compressive Signal Processing for Estimating Range-Velocity-AoA in FMCW Radar Applications. IEEE Access, 13, 26008-26026. DOI: 10.1109/ACCESS.2025.3536614

Related Recent Research:

THz and High-Frequency Applications:

  1. Gu, S., Xi, G., Ge, L., Yang, Z., Wang, Y., Chen, W., & Yu, Z. (2021). Compressed Sensing for THz FMCW Radar 3D Imaging. Complexity, 2021, 5576782. DOI: 10.1155/2021/5576782 https://www.hindawi.com/journals/complexity/2021/5576782/

Automotive and SAR Applications:

2. Lee, J., Wang, J., Lee, D., Park, K., & Kim, S. (2021). Compressive Sensing-Based SAR Image Reconstruction from Sparse Radar Sensor Data Acquisition in Automotive FMCW Radar System. Sensors, 21(21), 7283. DOI: 10.3390/s21217283 https://www.mdpi.com/1424-8220/21/21/7283

Recent 2024 Developments:

3. Liu, X., Wang, Y., Liu, F., & Zhang, Y. (2024). Real-Time Interference Mitigation for Reliable Target Detection with FMCW Radar in Interference Environments. Remote Sensing, 17(1), 26. DOI: 10.3390/rs17010026 https://www.mdpi.com/2072-4292/17/1/26

4. Zhang, H., Wei, S., Wang, M., Hu, Y., Shi, J., & Cui, G. (2024). FUAS-Net: Feature-Oriented Unsupervised Network for FMCW Radar Interference Suppression. IEEE Transactions on Microwave Theory and Techniques, 72(4), 2602-2619. DOI: 10.1109/TMTT.2024.3365014

Biomedical and Advanced Applications:

5. Eder, Y., & Eldar, Y.C. (2023). Sparsity Based Non-Contact Vital Signs Monitoring of Multiple People Via FMCW Radar. IEEE Journal of Biomedical and Health Informatics, 27(6), 2806-2817. DOI: 10.1109/JBHI.2023.3236830

Joint Communication and Sensing:

6. Ma, D., Shlezinger, N., Huang, T., Liu, Y., & Eldar, Y.C. (2021). FRaC: FMCW-based joint radar-communications system via index modulation. IEEE Journal of Selected Topics in Signal Processing, 15(6), 1348-1364. DOI: 10.1109/JSTSP.2021.3118219

Foundational Compressed Sensing Radar:

7. Potter, L.C., Ertin, E., Parker, J.T., & Cetin, M. (2010). Sparsity and compressed sensing in radar imaging. Proceedings of the IEEE, 98(6), 1006-1020. DOI: 10.1109/JPROC.2009.2037526

Recent Review and Future Directions:

8. Gini, F. (2021). Grand Challenges in Radar Signal Processing. Frontiers in Signal Processing, 1, 664232. DOI: 10.3389/frsip.2021.664232 https://www.frontiersin.org/journals/signal-processing/articles/10.3389/frsip.2021.664232/full

mmWave Radar Applications:

9. Cenkeramaddi, L.R., Rai, P.K., Dayal, A., Bhatia, J., Pandya, A., Soumya, J., Kumar, A., & Jha, A. (2022). Recent Advances in mmWave-Radar-Based Sensing, Its Applications, and Machine Learning Techniques: A Review. Sensors, 22(21), 8432. DOI: 10.3390/s22218432 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10650102/

Conference Proceedings and Recent Developments:

10. Rahayu, E.S., Ariananda, D.D., & Hidayat, R. (2023). Compressive Range and Velocity Estimation Using Frequency-Modulated Continuous Wave Radar. Conference Paper, November 2023.

These sources represent the cutting-edge developments in compressed sensing applications to FMCW radar, spanning from fundamental theory to practical implementations across automotive, biomedical, and telecommunications applications.

 

Compressive Signal Processing for Estimating Range-Velocity-AoA in FMCW Radar Applications | IEEE Journals & Magazine | IEEE Xplore

No comments:

Post a Comment

Explainer | China is making rapid gains in space tech. Here’s how the military could use it | South China Morning Post

China's Space Military Ambitions: A Growing Challenge to US Dominance China achieves historic space refueling milestone as Pentagon war...