Microwave Sensors

Within the framework of military and civilian oriented research the group develops and constructs active and passive microwave sensors for microwave remote sensing. Collaterally a detailed system simulation is mandatory, supporting the work in addition to the required instrumentation. Furthermore the relevant computations of microwave signatures are performed in order to investigate, predict, and compare those, and in order to be able to modify the behaviour and performance of the sensor systems. The development and construction work is mostly performed on a system engineering level, i.e. commercially available subsystems or single components are externally purchased, if a reasonable value for money is given. Non-available components are developed and constructed in house. Beside microwave techniques, required activities in the analogue and digital low-frequency range are exercised. Signal conditioning and data acquisition plays an important role in remote sensing. Here the subsystems are as well externally purchased and combined via adequate control software following the specific requirements of an application, e.g. real-time capability. The global objective for the sensor development is the realization of highest performance at low costs. Consequently novel imaging principles are of major interest.


Active sensors

Ground-based imaging radar

Modern radar applications for reconnaissance and security make use of highly resolved descriptions of the radar cross section (RCS) distribution necessary for advanced target recognition. Polarimetric features can act as an additional information source, allowing primarily the identification of scattering mechanisms. The use of different wavebands, e.g. from P to Ku-band, enable different scattering and penetration phenomena to be revealed for complicated objects. To investigate the potential of these capabilities experimentally, and for collecting true radar signatures, a powerful radar system called UNIRAD was developed. UNIRAD was designed as a stepped-frequency radar to be used for Inverse Synthetic Aperture Radar (ISAR) imaging in a tower-turntable environment, and for the application as a ground-based SAR mounted on a platform moving along a road or a rail track. UNIRAD covers L/S, C, X, and Ku-band. It has two transmit (TX) and two receive (RX) channels, which can be used for fully-polarimetric imaging, when both the two TX and the two RX channels are connected to a horizontally (H) and a vertically (V) linearly polarized antenna. In addition, the two TX and RX channels can be used as independent channels for the same polarization in order to investigate multi-channel radar architectures or bistatic modes.

Fully-polarimetric imaging

Figure 1 shows the RCS distribution of a car for 5 cm spatial resolution as a Red-Green-Blue (RGB) superimposition of all polarimetric signatures. Many details represented by tiny scattering centers can be observed. The VV channel in green appears to be dominant all over the car, while the front part roughly shows the same signal strength for the VV and HH polarization combination, shown in yellow. The left and right front parts also show some cross-polarising behaviour especially around the headlight region, visible as small blue spots.

Figure 1: RGB superimposition of the X-band RCS of a car for all polarimetric channels. Only the front part of the car is shown, radar illumination is looking towards the car.

Impact of spatial resolution

Figure 2 shows a series of SAR images of the same car with open doors and open hatchback. The SAR images are superimposed on an optical image taken from a bird’s eye perspective, in order to relate individual scattering centers to physical objects on the car. The SAR images were processed to four different spatial resolutions, in order to illustrate the loss of information as the resolution decreases. While the 5 cm image allows the detection of many fine scattering centers around the car, the coarse resolution of 80 cm can only indicate the existence of one or two scatterers. Note that, today, the spatial resolution for civilian spaceborne SAR systems is still worse than 80 cm.

Figure 2: Series of X-band RCS images for HH polarization and different levels of spatial resolution (indicated top left). For better comparison, the SAR images have been made transparent below a certain RCS level, in order to superimpose them correctly in size in relation to their optical equivalent. Radar illumination is from bottom.

Three-dimensional (3-D) imaging

Inverse SAR (ISAR) allows the collection of very precise high-resolution radar signatures from objects. For a spaceborne radar system, the imaging geometry is similar, differing only by geometrical transformation and rotation. Using ISAR techniques, a high spatial resolution in the decimeter range or higher is accomplished by rotating an object on a turntable with respect to a spatially fixed broadband radar system, and by recording a sequence of corresponding range profiles within a specific azimuth angular range. Despite the high resolution, visual object recognition is usually difficult, in particular for the typical 2-D images of the RCS distribution. Therefore, specific tower-turntable ISAR measurements were carried out for generating 3-D datasets by spanning a synthetic aperture not only in azimuth (by rotating the turntable), but also in elevation (by subsequent tilting of the turntable). High resolution 3-D images enable the analyst to arbitrarily change the perspective of the imaged object considerably, facilitating visual image interpretation. Now, the single scattering centers can be individually assigned to specific parts of the object under test. In order to enhance the visual perception, all single 3-D images for a complete turntable rotation can be incoherently superimposed. An example of such experiments is shown in Figure 3. This projection of the 3-D image especially reveals the wheel rims, the front mudguards, and the rear corners of the cargo area of the vehicle. Close to the vehicle sides, four additional small radar reference trihedrals on the turntable are visible in the radar image.

Figure 3: Photograph of a small truck (top) and its 3-D ISAR image for the perspective of about 45° elevation and 45° azimuth (bottom).


Passive sensors

Microwave radiometry

Radiometry addresses the domain of the passive measurement of the natural, thermal electromagnetic radiation of matter at a physical temperature above 0 K. In the case of microwave or millimeter-wave Earth observation, significant contrasts can be observed between reflective and absorbing materials, due to the impact of reflected sky radiation of cosmic origin.

Figure 4: Photographs and measured results for concealed object detection (object location indicated by orange symbol). Top: handgun in a notebook bag. Center: metal can in a backpack. Bottom: In contrast to the other experiments this one shows an indoor measurement of a handgun tucked in the trouser waistband.
The incident radiation power measured by a radiometer system is usually expressed in an apparent temperature, the brightness temperature. For Earth observation, an approximate range from 3 K to more than 300 K can be observed. The spatial two-dimensional brightness temperature distribution can be used as a daytime and almost weather independent indicator for many different physical phenomena. Based on a large experience in using radiometric imaging, our focus now is more on various security applications.

Security applications

The continuous threats by international terrorism increase the danger to the public and create a new and more complex threat dimension. This evolution can only be combated by the application of new counter-measure methods like advanced imaging technologies for surveillance and the detection of concealed dangerous objects. For the observation of a variety of security critical premises, borders, and maritime coastal areas, there is a strong demand on wide field-of-view imaging for intruder detection under all adverse ambient conditions. The imaging of persons with respect to weapons and explosives detection is of increasing interest, particularly for airlines, transportation services, or public events with large crowds.

The penetration capability of microwaves allows the detection of objects through atmospheric obstacles, like bad weather, fog, dust, vapour and smoke, as well as through thin non-metallic materials and clothing. For the latter, the detection of hidden objects like weapons, explosives, and contraband is possible by monitoring dielectric anomalies. Furthermore, the acquisition of polarimetric object characteristics can increase the detection capability by gathering complementary object information. Based on the physical principles of microwave radiometry, images have a quasi-optical appearance, simplifying the image interpretation for the operator. In addition, the sensor operation is inherently passive and covert.

Concealed object detection

Various methods to record high-resolution and high-sensitivity images in close to real-time exist at least theoretically. Apart from performance, the hardware complexity and cost is a major driver for the system design. Nowadays, the full electronic scanning of a scene is still very expensive, so that fully mechanical imagers are still attractive, in particular for experimental equipment for signature measurements and phenomenological studies. Consequently, various innovative mechanical imaging techniques have been developed. The LPAS systems were primarily designed to be used for low-cost, near-field experiments on applications like people screening at a distance of a few meters. The imaging principles are based on one receiver only. Nevertheless, image acquisition times in the order of up to 10 s for an image of about 1 m width and 1.5 m height have been achieved. The proof of concept and the capabilities of the LPAS systems have been demonstrated under laboratory conditions, as shown for selected results in Figure 4.


Fully-polarimetric imaging

The acquisition of polarimetric characteristics can increase the detection capability by gathering complementary object information. The different reflection/transmission behavior of radiation interacting with a dielectric body for two orthogonal polarizations is well known. However, the third and fourth component of the Stokes vector can also be of interest, as they are sensitive to anisotropic and periodic structures, sharp material transitions, and even small changes in the material composition and shape. Thus, the potential of fully-polarimetric information for security applications in general is evident. As a measurement example a car located in a typical environment was imaged using the ABOSCA imager, a mechanical scanner for wide field-of-view imaging. A photograph of the scene is shown together with the results of all four Stokes components in Figure 5. The car is located on a fairly smooth concrete surface, which is partly intersected by expansion gaps. One can see the typical difference between horizontal and vertical polarization, e.g. the differently pronounced mirror image of the car in the concrete area. The U and complementary V components show different features more pronounced, like material transitions and areas of structural anisotropy, e.g. the curved windows and the expansion gap, which are visible in all images. Note the higher noise level of the U and V component, due to the dynamic range of just a few Kelvin for those images being only little larger than the range of the sensitivity.

Figure 5: Photograph of a scene (top) and measured W-band images for all Stokes vector components of the brightness temperature. TH and TV indicate horizontal and vertical polarization, U and V are the real and imaginary part of their cross-correlation.

Fully-electronic imaging

Very often the main requirements for an imaging radiometer system are high resolution in parallel with low image acquisition time. Full electronic scanning would be preferred, but for budget constraints novel imaging principles using highly thinned arrays, like aperture synthesis, are also suitable. Consequently, research on application-oriented, high-performance low-cost solutions is performed. Two concepts are presently under investigation, the imaging low-frequency spectrometer with aperture synthesis, ANSAS, and the fully electronic scanner with aperture synthesis, VESAS. The ANSAS proof-of-concept demonstrator is a one-dimensional aperture synthesis array working in the frequency range of 1.4–6.5 GHz. The second image dimension is performed by array rotation, leading to a hybrid system of mechanical scanning and electronic beam steering. This configuration was chosen as a compromise effort between short imaging time, hardware expense and cost. By VESAS, a concept for novel, low-cost fully electronic scanning in the Ka-band is investigated. Two-dimensional scanning is realized by frequency scanning in one direction, i.e. beam steering by changing the received frequency band, and one-dimensional aperture synthesis in the other direction. The design goal, in this case, is to achieve a frame rate of around one second.

Figure 6: Principle of aperture synthesis for two-dimensional imaging using a sparse array. This technique can be applied as well for only one dimension using a linear sparse array.

Aperture synthesis is an imaging technique originally developed and used for radio astronomy. Here a large antenna is synthesized by a sparse array of small antennas as shown in Figure 6 for an Earth observation scenario. The signals from each antenna pair are complex cross-correlated, i.e. amplitude and phase or real and imaginary part are measured, and this value represents one sample of the spatial frequency spectrum, called as well the visibility function. The spatial frequency is the distance vector or baseline between two array elements considered for each sample. By sampling all baselines within the array according to the sampling theorem, the brightness temperature distribution can be reconstructed for an ideal system by a Fourier inversion of the sampled spatial frequency spectrum. The achievable spatial resolution is still depending on the size of the sparse array, as this is valid for any antenna, but aperture synthesis imaging requires no mechanical motion of the antenna and hence large antenna structures can be built for a high resolution. In addition, high frame rates are possible allowing real-time imaging, making this technique attractive for many security applications.


Dr.-Ing. Markus Peichl
Deutsches Zentrum für Luft- und Raumfahrt (DLR)

Institut für Hochfrequenztechnik und Radarsysteme
, Aufklärung und Sicherheit
Tel: +49 8153 28-2390

Fax: +49 8153 28-1135

E-Mail: markus.peichl@dlr.de
URL dieses Artikels