Three-Dimensional Microwave Imaging for Indoor Environments
Simon Scott

Citation
Simon Scott. "Three-Dimensional Microwave Imaging for Indoor Environments". PhD thesis, University of California Berkeley, August, 2017.

Abstract
Microwave imaging involves the use of antenna arrays, operating at microwave and millimeter-wave frequencies, for capturing images of real-world objects. Typically, one or more antennas in the array illuminate the scene with a radio-frequency (RF) signal. Part of this signal reflects back to the other antennas, which record both the amplitude and phase of the reflected signal. These reflected RF signals are then processed to form an image of the scene. This work focuses on using planar antenna arrays, operating between 17 and 26 GHz, to capture three-dimensional images of people and other objects inside a room. Such an imaging system enables applications such as indoor positioning and tracking, health monitoring and hand gesture recognition. Microwave imaging techniques based on beamforming cannot be used for indoor imaging, as most objects lie within the array near-field. Therefore, the range-migration algorithm (RMA) is used instead, as it compensates for the curvature of the reflected wavefronts, hence enabling near-field imaging. It is also based on fast-Fourier transforms and is therefore computationally efficient. A number of novel RMA variants were developed to support a wider variety of antenna array configurations, as well as to generate 3-D velocity maps of objects moving around a room. The choice of antenna array configuration, microwave transceiver components and transmit power has a significant effect on both the energy consumed by the imaging system and the quality of the resulting images. A generic microwave imaging testbed was therefore built to characterize the effect of these antenna array parameters on image quality in the 20 GHz band. All variants of the RMA were compared and found to produce good quality three-dimensional images with transmit power levels as low as 1 uW. With an array size of 80x80 antennas, most of the imaging algorithms were able to image objects at 0.5 m range with 12.5 mm resolution, although some were only able to achieve 20 mm resolution. Increasing the size of the antenna array further results in a proportional improvement in image resolution and image SNR, until the resolution reaches the half-wavelength limit. While microwave imaging is not a new technology, it has seen little commercial success due to the cost and power consumption of the large number of antennas and radio transceivers required to build such a system. The cost and power consumption can be reduced by using low-power and low-cost components in both the transmit and receive RF chains, even if these components have poor noise figures. Alternatively, the cost and power consumption can be reduced by decreasing the number of antennas in the array, while keeping the aperture constant. This reduction in antenna count is achieved by randomly depopulating the array, resulting in a sparse antenna array. A novel compressive sensing algorithm, coupled with the wavelet transform, is used to process the samples collected by the sparse array and form a 3-D image of the scene. This algorithm works well for antenna arrays that are up to 96% sparse, equating to a 25 times reduction in the number of required antennas. For microwave imaging to be useful, it needs to capture images of the scene in real time. The architecture of a system capable of capturing real-time 3-D microwave images is therefore designed. The system consists of a modular antenna array, constructed by plugging RF daughtercards into a carrier board. Each daughtercard is a self-contained radio system, containing an antenna, RF transceiver baseband signal chain, and analog-to-digital converters. A small number of daughtercards have been built, and proven to be suitable for real-time microwave imaging. By arranging these daughtercards in different ways, any antenna array pattern can be built. This architecture allows real-time microwave imaging systems to be rapidly prototyped, while still being able to generate images at video frame rates.

Electronic downloads

Internal. This publication has been marked by the author for TerraSwarm-only distribution, so electronic downloads are not available without logging in.

Citation formats

• HTML

  Simon Scott. <a
  href="http://www.terraswarm.org/pubs/994.html"
  ><i>Three-Dimensional Microwave Imaging for Indoor
  Environments</i></a>, PhD thesis,  University of
  California Berkeley, August, 2017.

• Plain text

  Simon Scott. "Three-Dimensional Microwave Imaging for
  Indoor Environments". PhD thesis,  University of
  California Berkeley, August, 2017.

• BibTeX

  @phdthesis{Scott17_ThreeDimensionalMicrowaveImagingForIndoorEnvironments,
      author = {Simon Scott},
      title = {Three-Dimensional Microwave Imaging for Indoor
                Environments},
      school = {University of California Berkeley},
      month = {August},
      year = {2017},
      abstract = {Microwave imaging involves the use of antenna
                arrays, operating at microwave and millimeter-wave
                frequencies, for capturing images of real-world
                objects. Typically, one or more antennas in the
                array illuminate the scene with a radio-frequency
                (RF) signal. Part of this signal reflects back to
                the other antennas, which record both the
                amplitude and phase of the reflected signal. These
                reflected RF signals are then processed to form an
                image of the scene. This work focuses on using
                planar antenna arrays, operating between 17 and 26
                GHz, to capture three-dimensional images of people
                and other objects inside a room. Such an imaging
                system enables applications such as indoor
                positioning and tracking, health monitoring and
                hand gesture recognition. Microwave imaging
                techniques based on beamforming cannot be used for
                indoor imaging, as most objects lie within the
                array near-field. Therefore, the range-migration
                algorithm (RMA) is used instead, as it compensates
                for the curvature of the reflected wavefronts,
                hence enabling near-field imaging. It is also
                based on fast-Fourier transforms and is therefore
                computationally efficient. A number of novel RMA
                variants were developed to support a wider variety
                of antenna array configurations, as well as to
                generate 3-D velocity maps of objects moving
                around a room. The choice of antenna array
                configuration, microwave transceiver components
                and transmit power has a significant effect on
                both the energy consumed by the imaging system and
                the quality of the resulting images. A generic
                microwave imaging testbed was therefore built to
                characterize the effect of these antenna array
                parameters on image quality in the 20 GHz band.
                All variants of the RMA were compared and found to
                produce good quality three-dimensional images with
                transmit power levels as low as 1 uW. With an
                array size of 80x80 antennas, most of the imaging
                algorithms were able to image objects at 0.5 m
                range with 12.5 mm resolution, although some were
                only able to achieve 20 mm resolution. Increasing
                the size of the antenna array further results in a
                proportional improvement in image resolution and
                image SNR, until the resolution reaches the
                half-wavelength limit. While microwave imaging is
                not a new technology, it has seen little
                commercial success due to the cost and power
                consumption of the large number of antennas and
                radio transceivers required to build such a
                system. The cost and power consumption can be
                reduced by using low-power and low-cost components
                in both the transmit and receive RF chains, even
                if these components have poor noise figures.
                Alternatively, the cost and power consumption can
                be reduced by decreasing the number of antennas in
                the array, while keeping the aperture constant.
                This reduction in antenna count is achieved by
                randomly depopulating the array, resulting in a
                sparse antenna array. A novel compressive sensing
                algorithm, coupled with the wavelet transform, is
                used to process the samples collected by the
                sparse array and form a 3-D image of the scene.
                This algorithm works well for antenna arrays that
                are up to 96% sparse, equating to a 25 times
                reduction in the number of required antennas. For
                microwave imaging to be useful, it needs to
                capture images of the scene in real time. The
                architecture of a system capable of capturing
                real-time 3-D microwave images is therefore
                designed. The system consists of a modular antenna
                array, constructed by plugging RF daughtercards
                into a carrier board. Each daughtercard is a
                self-contained radio system, containing an
                antenna, RF transceiver baseband signal chain, and
                analog-to-digital converters. A small number of
                daughtercards have been built, and proven to be
                suitable for real-time microwave imaging. By
                arranging these daughtercards in different ways,
                any antenna array pattern can be built. This
                architecture allows real-time microwave imaging
                systems to be rapidly prototyped, while still
                being able to generate images at video frame rates.},
      URL = {http://terraswarm.org/pubs/994.html}
  }
  

Posted by Simon Scott on 4 Sep 2017.