Dual Channel S-Band Frequency Modulated Continuous Wave Through-Wall Radar Imaging

Ying-chun Li, Daegun Oh, Sunwoo Kim, Jong-Wha Chong2018 · DOI

TL;DR

This paper introduces a low-complexity, dual-channel S-band FMCW radar system for through-wall imaging (TWRI). By employing a novel 3D shift-invariant subspace algorithm, it achieves high-resolution imaging of targets behind walls, outperforming traditional SAR-based methods in terms of system complexity and mobility.

What Problem Does This Paper Address?

Through-wall radar imaging (TWRI) is crucial for applications like search and rescue, law enforcement, and military operations, enabling visibility into environments obscured by walls. However, existing TWRI systems often face challenges:

  • Wall Clutter: Signals reflected by the wall itself are significantly stronger than those from targets behind it, overwhelming receivers and limiting dynamic range. This interference can even prevent the use of advanced techniques like compressive sensing.
  • System Complexity and Mobility: Many high-resolution TWRI systems rely on Synthetic Aperture Radar (SAR) principles, which typically require multiple antennas and radar movement. This makes them bulky, complex, and less suitable for rapid deployment in emergency situations.
  • Cost: UWB impulse radars offer high resolution but are complex and expensive due to high sampling rate requirements.

This paper aims to address these issues by developing a simpler, more mobile, and cost-effective TWRI system that can still provide high-quality imaging of targets behind walls.

Methods

The core of this work lies in the development of a dual-channel S-band Frequency Modulated Continuous Wave (FMCW) system and a novel 3D shift-invariant subspace algorithm for target imaging.

  • System Design: The implemented system uses a single transmitting antenna (Tx) and two receiving antennas (Rx). This dual-channel configuration significantly reduces the number of antennas compared to traditional SAR systems. The system operates in the S-band, utilizing FMCW waveforms, which are favored for their lower cost and simpler hardware requirements compared to UWB impulse radars.
  • 3D Shift-Invariant Subspace Algorithm: This algorithm is the key innovation for achieving high-resolution imaging. It leverages the concept of a "3D shift invariant structure" derived from the received radar signals.
  • Hankel Matrices: The received beat signals are organized into Hankel matrices, which capture the shift-invariant properties of the data across different domains (sample, pulse, and antenna).
  • Singular Value Decomposition (SVD): SVD is applied to these matrices to separate the signal subspace (containing target information) from the noise subspace (containing clutter and interference).
  • Pseudo-Spectrum Estimation: A 3D pseudo-spectrum is then computed to estimate the range, angle (azimuth), and Doppler (velocity) of targets. This method allows for joint estimation of these parameters.
  • Peak Detection: By identifying peaks in the pseudo-spectrum, the system estimates the number of targets and their respective parameters.

This approach aims to exploit the underlying structure of the radar data to achieve high resolution without the need for complex SAR processing or extensive antenna arrays.

Datasets

Not discussed in the paper. The experiments were conducted using real-world scenarios with controlled targets (water blocks, person) and simulated wall clutter.

Key Insights and Ideas

  • Dual-Channel Simplicity: The realization that a dual-channel system (one Tx, two Rx) can significantly reduce hardware complexity while still enabling advanced imaging techniques is a key takeaway. This contrasts with SAR systems that often require many antennas.
  • 3D Shift Invariance for Subspace Methods: The paper demonstrates the power of exploiting the "3D shift invariant structure" of FMCW radar data. By organizing the data into smoothed Hankel matrices, the underlying signal properties can be effectively extracted using subspace methods.
  • Joint Range-Angle-Doppler Estimation: The proposed algorithm's ability to jointly estimate range, azimuth angle, and Doppler frequency from the radar data is a significant advantage for comprehensive target characterization.
  • Practicality for Emergency Scenarios: The emphasis on low complexity and mobility makes this approach highly relevant for real-world applications where rapid deployment is critical, such as in rescue operations.

Key Results

The paper presents experimental results from both indoor and outdoor scenarios, demonstrating the effectiveness of the proposed dual-channel FMCW TWRI system and its associated algorithm:

  • High-Quality Imaging: The system successfully imaged targets behind a 20 cm thick concrete wall with embedded iron bars.
  • Target Detection:
    • Indoor Experiment: Detected four identical square water blocks, estimating M=5 targets (likely due to some ambiguity or additional clutter). The results showed significant peaks at the wall range, which the authors note can be mitigated.
  • Outdoor Experiment (Water Blocks): Detected multiple water blocks, with one frame showing M=5 targets. The results indicated a crossrange of 1m and downrange of 1.8m, corresponding to an azimuth angle of approximately 29°. The range-Doppler map showed a straight-line range of about 2.1m and a Doppler frequency of ~25 Hz.
  • Outdoor Experiment (Person): Detected a moving person, estimating M=8 targets (likely representing different body parts). The azimuth angle was about 27°, with about 1m crossrange and 2m downrange. The range-Doppler map showed a range of ~2.2m and a Doppler frequency of ~-25 Hz.
  • Wall Clutter Mitigation: While not explicitly quantified, the results suggest that the system and algorithm can effectively handle wall clutter, allowing for the detection of targets behind it. The use of a bandpass filter also contributes to this.

Limitations and Future Work

  • Algorithm Complexity: The authors acknowledge that the proposed algorithm, involving operations like SVD, is currently implemented in MATLAB and not yet on FPGA/DSP due to large operating costs.
  • Real-time Implementation: A significant future direction is to develop a low-complexity version of the algorithm that can be implemented in real-time on hardware like FPGAs and DSPs, enabling a truly real-time TWRI system with two receiving channels.
  • Elevation Angle: The current system, with receiving antennas in a horizontal array, does not consider the elevation angle of human components, which limits the detailed characterization of human targets. Future work could explore systems that incorporate elevation angle estimation.
  • Wall Parameter Estimation: While the paper focuses on imaging, other research areas like wall parameter estimation are mentioned as separate topics, suggesting potential integration for improved performance.

Summary updated

BibTeX
@misc{li2018dual,
  author = {Ying-chun Li and Daegun Oh and Sunwoo Kim and Jong-Wha Chong},
  title = {{Dual Channel S-Band Frequency Modulated Continuous Wave Through-Wall Radar Imaging}},
  year = {2018},
  doi = {10.3390/s18010311},
}
Figure