SAR Imaging with Noise Radar - IEEE Xplore

I. INTRODUCTION

SAR Imaging with Noise Radar

D. TARCHI Joint Research Centre of the European Commission Italy K. LUKIN, Fellow, IEEE National Academy of Sciences of Ukraine J. FORTUNY-GUASCH, Senior Member, IEEE Joint Research Centre of the European Commission Italy A. MOGYLA P. VYPLAVIN National Academy of Sciences of Ukraine A. SIEBER, Fellow, IEEE Joint Research Centre of the European Commission Italy

Two ground-based coherent imaging systems designed on the basis of noise radar technology (NRT) using continuous waveform with a linear synthetic aperture and pulse coherent noise waveform with a circular synthetic aperture are presented. A short description of noise waveform synthetic aperture radar (SAR) theory and data processing algorithms are given. Experimental results obtained show that the quality of the imagery is comparable to that of frequency-modulation continuous-wave (FMCW) or stepped frequency radar systems. Experimental validation of the technique at different locations provides confidence in the ability of noise waveform SAR to achieve high-resolution imaging at low cost. The exploitation of the interferometric phase combining pairs of images spaced in time is shown to allow the detection of displacements in the scene with a submillimeter accuracy.

Manuscript received July 26, 2007; revised June 17, December 24, 2008; released for publication March 9, 2009. IEEE Log No. T-AES/46/3/937968. Refereeing of this contribution was handled by M. Rangaswamy. Authors’ addresses: D. Tarchi, J. Fortuny-Guasch, and A. Sieber, Institute for the Protection and Security of the Citizen, Joint Research Centre of the European Commission, Ispra, Italy; K. Lukin, A. Mogila, and P. Vyplavin, Institute for Radiophysics and Electronics, National Academy of Sciences of Ukraine, LNDES, IRE NASU 12 Acad. Proskura St., Kharkov, 61085, Ukraine, E-mail: ([email protected]). c 2010 IEEE 0018-9251/10/$26.00 ° 1214

There are a number of applications where high-resolution microwave imaging of relatively small areas is required. Recently, a number of ground-based (GB) synthetic aperture radar (SAR) systems with submeter resolution both in range and cross-range were developed for that purpose. In [1]—[3], an X-band GB SAR using stepped frequency waveform is used for monitoring of ground instabilities and/or structural deformations of man-made structures using differential interferometric SAR (D-InSAR) techniques. In [4], a fully-polarimetric X-band SAR for assessment of the operational potentials and limitations of airborne and satellite SAR platforms is described. Yet these systems have a low electromagnetic compatibility due to the use of a stepped frequency waveform. Also the system described in [1]—[4] has associated long measurement times (typically a few minutes) basically needed for the stepped frequency data acquisition. Noise radar technology (NRT) [5—8] applied to GB SAR naturally can improve the electromagnetic compatibility. Besides, due to the fact that all the frequencies of noise signal are transmitted simultaneously, NRT enables us to shorten the measurement time in comparison with stepped frequency radar. Use of the noise signals for near-range X-band SAR imaging for the first time was suggested in [9]. In [10], another type of GB SAR, the ArcSAR was designed on the basis of an ultrawideband noise waveform. Noise waveform has also been used in inverse SAR (ISAR) systems in [11], [12] and in a system with digital beamforming [13]. A bistatic noise Ka-band ground-based SAR was developed for monitoring of the Chernobyl sarcophagus in [14]. This system is intended for detection of small shifts or changes in a scene using the D-InSAR techniques, and it provides a rather high spatial resolution and phase measurement precision. In this paper, we present some further investigations of implementations of noise radar technology in X-band GB SAR. The radars under consideration use a narrowband noise waveform (NW) as a probe signal and perform range focusing by correlating noise radar returns with a reference signal sampled before its transmission. Azimuth compression is similar to that in conventional SAR. Hereinafter we refer to this type of SAR as GB NW-SAR. Application of noise radar technology offers a possibility to use an extremely low transmitted power and still achieve fast data acquisition, high electromagnetic compatibility, small size, and light weight. More importantly, this technology is very affordable. In addition, an NW with a wide enough Gaussian power spectrum is able to minimize the ambiguity function range sidelobes [15]. The main area of possible applications of the designed

IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS VOL. 46, NO. 3

JULY 2010

TABLE I Main Parameters of the Linear NW-SAR

Fig. 1. Block diagram of linear NW SAR.

GB NW-SAR is associated with the 24/7 monitoring of small areas and detection of changes in the scene of interest applying D-InSAR technique. We briefly describe both types of the GB NW-SAR and report on a series of experiments aimed to validate the technology for small area surveillance applications. In particular, we suggest the exploitation of the interferometric phase to detect displacements occurring in between the images acquisition with submillimeter accuracy. II. SAR SYSTEMS DESCRIPTION A. Linear GB NW-SAR An X-band linear continuous NW SAR has been designed in the Institute for the Protection and Security of the Citizen, in the European Commission Joint Research Centre. The block diagram of the linear GB NW-SAR is shown in Fig. 1. This system has an aperture length of 5 m, which is synthesized through a stop-and-go linear scanning of the radar’s Tx/Rx antennas. In the short range experiments, we used only a half of this aperture. A total of 251 measurement points along the linear aperture are used in order to avoid aliases in the azimuth direction. The continuous noise generator included in the noise waveform transmitter (NWT) unit has been developed on the basis of dynamical chaotization principles [7]. It generates chaotic signals with a selectable central frequency in the range from 9.05 to 9.6 GHz and bandwidth up to 200 MHz. Its output power is about 150 mW in CW regime. The two demodulators have a minimum intermediate frequency (IF) bandwidth of 300 MHz. The IF output is a signal with 150 MHz central frequency and 300 MHz bandwidth. The two IF outputs are digitized simultaneously using a multi-channel 1Gsample digital oscilloscope. The internal memory of the oscilloscope allows the storage of 200K points per channel. The A/D conversion is made with a resolution of 8-bit. Interestingly, the oscilloscope is also able to estimate in quasi-real time cross-correlation between the two acquired signals. Alternatively the two digitized signals can be stored in a hard disk of the computer for a subsequent off-line processing. TARCHI ET AL.: SAR IMAGING WITH NOISE RADAR

Linear NWT centre frequency Linear NWT frequency bandwidth Transmitted power Polarization Antenna height Synthetic Aperture Spatial step Range resolution Cross-range resolution at 50 range

9.1 GHz 150 MHz 30 mW VV 1.5 m 2.5 m 0.01 m 1 m 0.6 m

We implement D-InSAR technique without compensation of irregularities of the antenna trajectory. Therefore high repeatability of this trajectory is needed. Influence of distortions in antenna path on the precision of interferometric measurements is discussed in [16]. The linear rail of the GB-SAR system [1] provides precise (< 0:5 mm repeatability) linear movement of the transmitter/receiver over a range up to 5 m. The motion can be either continuous (velocity in the range of 10 cm/s) or in user-defined steps (minimum step: 1 mm). All the measurement procedure is controlled by a PC where the acquired data are finally stored. Further processing steps, performed on the PC, comprise the range data compression if not performed on the oscilloscope, data calibration, and azimuth compression. In each azimuth position, a series of measurements are repeated and coherently summed up in order to increase the integration time and to reduce the artifacts due to decorrelating targets in the scenario. The integration time in each position is in the order of few milliseconds, whereas the total measuring time for a single image is in the order of few minutes due to the actual very low data transfer rate between the oscilloscope and the computer. The acquired data have been calibrated using an additional measurement of a reference target (metallic disc) at a known distance and range compressed by correlating radar return and reference signal in each position. Fig. 2 shows a photograph of the linear GB NW-SAR used in the experimental validation. The main parameters of the system used in measurements are given in Table I. B. Circular SAR An X-band circular GB NW-SAR has been designed in the LNDES IRE NAS, Ukraine. It uses a pulsed NW as a probing signal and correlation processing for coherent reception of radar returns with capability of pulse-to-pulse coherent integration. To perform such coherent integration, we have to sample the transmitted noise pulse after its down-conversion and use it as a reference. We use the same channel for sampling the noise radar returns (also down 1215

Fig. 2. Photograph of linear NW SAR.

Fig. 3. Block diagram of X-band GB NW-SAR system. Tx is transmitter, Rx is receiver, AFS is antenna feeding system, AFC is automated frequency control, ADC is analog-to-digital converter, AS is antenna switch, CU is control unit, ADS is angle data sensor, PC is personal computer.

converted). The GB NW-SAR consists of three units: transmit/receive (Tx/Rx) unit; antenna with antenna pillar; data acquisition & processing unit) and control unit (synchronizer). Block diagram of the circular GB NW-SAR is shown in Fig. 3. The Tx/Rx unit consists of a transmitter (Tx), a receiver (Rx), frequency locked loop (FLL), an antenna switch (AS), and an antenna feeding system (AFS). The transmitter generates NW pulses with 9.2 GHz central frequency having a duration of 210 ns and spectrum bandwidth of 250 MHz with pulse repetition frequency (PRF) up to 100 kHz. Central frequency of these signals is stabilized with the help of FLL. In order to provide a reference for correlation processing a part of the sounding signal is taken using a directional coupler, down converted amplified, and sampled using 8 bit CompuScope82G analog-to-digital converter (ADC) from GaGe Company having a maximal sampling frequency of 2Gsamples. After the transmission of a noise pulse the antenna switch connects antenna output to the radar 1216

receiver input. The radar return is amplified, down converted, and sampled using the same frequency converter, IF amplifier, and ADC. With the available 2 MByte onboard memory, up to 250 pulses can be transmitted and received at each antenna position. The antenna unit is responsible for the synthesis of the circular aperture. The horn antenna is mounted at one end of a special antenna boom. The second end of the antenna boom is mounted onto the axis of the main drive of the antenna pillar along with an angular sensor. An additional mechanism keeps the horn antenna pointing in a fixed direction when the boom is rotating. During the measurement, the angular velocity is kept constant, different from the step-and-go mode in the linear GB NW-SAR. The angular extent of the circular aperture is 99 deg, and a total of 111 measurement points are sampled. The precision of the angular sensor is » 0:06 deg which for the 1.75 m boom length enables controlling the horn antenna position along the arc with accuracy of better than 0.18 mm. When the angle reaches a certain value, a special optoelectronic sensor sends a pulse to the synchronization system which forms instructions for the Tx/Rx unit and ADC to start transmission and data acquisition accordingly. The acquired data are stored in the PC hard disc. The same PC is used for control the GB NW-SAR and all the measurement procedure. Fig. 4 shows a photograph of the circular GB NW-SAR. One can see the main box containing the antenna pillar, the Tx/Rx unit, and part of the control unit. All other parts such as the boom with the transmit/receive horn antenna at its end and the power and signal cables connected to the control panel of the GB NW-SAR are clearly seen in Fig. 4 as well. The measurement parameters of the circular GB NW-SAR are summarized in Table II.


JULY 2010

Fig. 4. Photograph of circular GB NW-SAR. TABLE II Main Parameters of the Circular NW SAR Circular NWT central frequency Circular NWT frequency bandwidth Peak transmitted power Average transmitted power Polarization Incidence angle (with respect to the normal) Antenna height Arc radius Scanning angle Scanning step Range resolution Cross-range resolution at 50 range

9.2 GHz 250 MHz 30 mW 2 mW VV ¡30± 20 m 1.75 m 99± 0:9± 0.6 m 0.3 m

III. SAR IMAGING PROCEDURES The antennas of the linear NW SAR move in the stop-and-go mode and they do not move during the data acquisition. It enables separating SAR processing in two independent stages: 1) range compression and 2) azimuth compression. The first one enables estimation of ranges to the targets whereas the second enables one to resolve targets in azimuth. Let us consider SAR image generation in more detail. The algorithm for range compression is the same for both linear and circular SAR, while azimuth compression is to be adopted for the used type of radar aperture. Let us first consider the range compression technique for the NW SAR following the papers [5—7]. Supposing that the random waveform is stationary and ergodic, we may evaluate the cross-correlation between the reference signal and radar returns as follows: Z 1 T=2 X(t ¡ ¿ 0 )X(t ¡ ¿ )dt (1) R(¿ ) = lim T!1 T ¡T=2 TARCHI ET AL.: SAR IMAGING WITH NOISE RADAR

where X(t ¡ ¿ ) is the reference signal, i.e., the part of the transmitted signal delayed in a delay line with variable delay ¿ ; X(t ¡ ¿ 0 ) is the radar return or received signal, which is a delayed copy, of the transmitted signal; is the delay of the transmitted signal due to its propagation towards a target and back. Substitutions t0 = t ¡ ¿ ; ¿0 = ¿ 0 ¡ ¿ transform (1) to the following form: Z 1 T=2 R(¿0 ) = lim X(t0 ¡ ¿0 )X(t0 )dt0 : (2) T!1 T ¡T=2 For a random stationary signal, the Wiener-Khintchin theorem is valid, which allows us to write (2) in the spectral domain as follows: Z 1 1 jF(!)j2 e¡j!¿0 d! (3) R(¿0 ) = 2¼ ¡1 where jF(!)j2 is the power spectral density of the NW. The NW generator used in the radar produces signals with the shape of a power spectral density close to Gaussian [5—7]; i.e., jF(!)j2 = e¡(!¡!0 )

2

=2B 2

: (4) p Such a power spectrum has 1= e—level circular bandwidth B and is centered at the circular frequency !0 . Substitution of (4) into (3) and taking the integral yields B 2 2 (5) R(¿0 ) = p e¡(B ¿0 =2)¡j!¿0 2 ¼ or, substituting ¿0 = ¿ ¡ ¿ 0 back gives us B 2 0 2 0 R(¿ ¡ ¿ 0 ) = p e¡(B (¿ ¡¿ ) =2)¡j!(¿ ¡¿ ) : 2 ¼

(6)

1217

It is seen from (5)—(6) that theoretically, a radar based upon noise signal with a Gaussian-like shape of power spectrum will provide simultaneously 1) high rate compression of the received signal at the receiver output, 2) optimal reception of radar returns, and 3) minimal sidelobes in the radar ambiguity function. However, the given relations are written for the case of infinite interval in both receiver bandwidth and signal duration. Finite width of the spectrum in realistic noise radars causes the appearance of range sidelobes, the level of which depends on the ratio between the width of the Gaussian power spectrum of the random signal and the receiver bandwidth [15]. It has been shown in [15] that ¡40 dB sidelobe level can be achieved in the case of the receiver bandwidth being twice the ¡3 dB width of the power spectral density of the Gaussian NW. Finiteness of the signal duration leads to appearance of residual noise, which exists even if the NW is cross-correlated with itself (without adding any external noise to the channels). Randomness of residual noise enables it to be reduced dramatically via long time integration of the CW random signals or, alternately, via additional coherent summation of cross-correlations obtained from different pulses with random filling. Thus, in the case of a wide enough receiver bandwidth compared to the spectrum width of the signal and long integration time, NW may provide a performance close to the theoretical one. That makes NW useful in many radar applications and, in particular, in SAR imaging. Now let us describe azimuth compression techniques. In the linear GB NW-SAR we split the sounding signal into samples for each of which we find spectra using fast Fourier transform (FFT) and perform a correlation procedure in the frequency domain using the discrete analog of (2) with finite limits. Results are integrated over all samples. Finally, we obtain array Sa,l of range compressed data related to the set of discrete antenna positions a (equivalent to the slow time) and discrete time l. Azimuth compression is done by summing up responses from the target obtained from all antenna positions with corresponding phase correction [17] using the following relation for reflectivity ³(x, y, z) of the point of interest (x, y, z): ³(x, y, z) =

A X a=0

Wa Sa

μ

¶ Ra (x, y, z) expfikRa (x, y, z)g c (7)

where a exp(¡i!t) time dependence is assumed, A is the total number of antenna positions; Wa is a weighting factor to lower the sidelobes; Ra (x, y, z) is the distance which the signal propagates from antenna to the focused point (x, y, z) and back; Sa (Ra (x, y, z)=c) is a function built on the basis of Sa,l by Lagrange 1218

Fig. 5. Linear and circular SAR geometry.

interpolation over four closest neighbors of the point of interest; km = 2¼Fcar =c is the wave number; Fcar is the central frequency of the signal. As both systems can be considered monostatic, one may write Ra (x, y, z) =

q (x ¡ xa )2 + (y ¡ ya )2 + (z ¡ za )2

(8)

where (xa , ya , za ) are coordinates of the antenna at position a. For a linear SAR oriented in Cartesian coordinates as depicted in Fig. 5(a), ¶ μ A ¢x xa = a ¡ 2 ya = 0

(9)

za = h where ¢x is the spatial step of the synthetic aperture. For azimuth weighting in linear SAR we used the following function in order to minimize the azimuth sidelobes: ¶ μ a+1 2 : (10) Wa = sin ¼ A+2 In the circular GB NW-SAR, the acquisition time at each antenna position is about 130 ¹s. This corresponds to an antenna shift of » 7 ¹m which is much shorter than the signal wavelength. These very short shifts allow us to consider the antenna as static when data acquisition is taking place and therefore to split SAR processing into range and azimuth compression steps. The range compression in circular GB NW-SAR is similar to the case of CW NW-SAR with the only difference that the sounding signal is a pulse train. The correlation was calculated for each pair of pulses, and its return and integration was done for all pulses transmitted at the given azimuth. The azimuth compression can be done in the same way using (7) and (8) but taking into account the circular motion of the horn antenna. Using geometry depicted in Fig. 5(b) one can find the following


JULY 2010

Fig. 6. Photograph of area for outdoor measurements with linear GB NW-SAR.

expressions for the antennas positions (xa , ya , za ): ¶ ¶ μμ A ¢μ a¡ xa = L ¢ cos 2 ¶ ¶ μμ A (11) ¢μ ya = L ¢ sin a¡ 2 za = h where ¢μ is the angular step of scanning, L is the length of the boom. Relations (11) are to be used instead of (9). The circular motion of the antenna causes irregular amplitude distribution of the received signal over equivalent linear aperture, which gives sidelobes in azimuth. In order to overcome this effect we used the following windowing function: ¶ μ sin(μa ) ¡ sin(μmin ) (12) Wa = cos(μa ) sin2 sin(μmax ) ¡ sin(μmin ) where μa = (a ¡ A=2)¢μ ¡ μ0 is the angle between the antenna direction μ0 and the boom direction; μmin and μmax are minimal and maximal values of the μa . The first factor of the window function stands for linearization of the amplitude distribution and the second one forms equivalent of the sin2 (: : :) window. This window function is applicable in the cases of ¡¼=2 < ®a < ¼=2 for all antenna positions a. IV. EXPERIMENTAL RESULTS A. Linear SAR A photograph and a sketch of the scene where the outdoor measurements have been performed are shown in Figs. 6 and 7. The scenario is characterized by a flat area covered by asphalt with the following TARCHI ET AL.: SAR IMAGING WITH NOISE RADAR

Fig. 7. Map of area for outdoor measurements with linear GB NW-SAR.

structures: a building, about 5 m in height made up of bricks and with a series of regularly spaced iron beams (labeled A); a structure, about 7 m in height made up of metallic containers and pipelines (labeled B); a building, 15 m in height made up of bricks (labeled C); a building, 4 m in height made up of metal (labeled D); two areas covered by high fir trees (labeled E). A metallic truck trailer, about 7 m long, was present on the right side. As indicated in these figures, three reference targets (corner reflectors) have been placed at known positions in the scenario. The acquisitions have been repeated with exactly the same measurement parameters in order to evaluate the stability of the system. In Fig. 8, the power normalized to the maximum of the obtained SAR image is displayed. The image is projected on a horizontal plane exactly corresponding to the area 1219

compared with the values derived from the radar image. A very good agreement has been found in all cases. To evaluate the stability of the system, two images acquired at different times have been compared. Fig. 9 shows the histogram of the phase of the interferogram formed with the two images, considering pixels with coherence higher than 0.5, where the coherence is defined as hI1 I2¤ i p (13) °=p hI1 I1¤ i hI2 I2¤ i

Fig. 8. NW SAR image of area depicted in Fig. 7.

in the map of Fig. 7. By comparing it with Fig. 7, most of the relevant features in the scenario turn out to be clearly recognizable in the radar image. In particular the layout of the buildings and of the metallic structure on the left and on the upper part is correctly imaged as well as the trailer on the right and the three reference targets. Additional features can be identified with a more detailed analysis of the image: 1) a series of bright spots corresponding to the facade of the building on the left side due to the iron beams; 2) five spots close to the upper right corner of the building on the left side corresponding to cars; 3) a series of less intense spots close to the third corner reflector corresponding to the trunks of fir trees. The positions and the relative distances between most of the recognizable features in the scene have been measured with traditional tools and then

where I1 and I2 are, respectively, the master and slave images, and the operator h i indicates a spatial averaging typically applied over 3 £ 3 resolution cells. The average value turns out to be very close to zero whereas the standard deviation as derived from the histogram is in the order of 15 deg, which leads to a submillimeter displacement accuracy. Skew in the histogram can be explained by a minor antenna path distortion due to imperfection of the mechanical part or by atmospheric effects. B. Circular GB NW-SAR The measurement setup was the following: the circular NW SAR was placed in a room on the 5th floor of a building at a height of » 20 m above the ground. The boom with horn antenna mounted was placed outside a window. The horn antenna pattern width was 50 deg, and it was directed along the line with an incidence angle of ¡30 deg and directed ¡25 deg with respect to the center position of the boom. Fig. 10 shows the photograph of the area of experiment, and Fig. 11 shows its sketch. The ground surface was covered with grass and also contains an asphalt road. A building nearly 15 m high (labeled A) with a metal border of the roof and 12 metal vent tubes on the top was at a horizontal distance of 80 m from NW SAR. Another building (labeled B) with a metal roof was at a distance of 230 m from the radar. Both buildings were made up of bricks. Two groups

Fig. 9. Phase histogram formed with two images obtained with help of linear NW SAR. 1220


JULY 2010

Fig. 10. Photograph of area for outdoor measurements with circular NW SAR.

Fig. 11. Sketch of measurements area for circular SAR experiment.

of trees were in front of the building B (labeled C) and behind it (labeled D). A horn antenna (labeled E) with a short circuit plunger in a waveguide was placed on the ground at the distance of 30 m from the radar and used as a target with variable phase center. The phase of the reflected signals could be adjusted with high precision using the plunger. Sets of scans with different plunger positions were performed. The obtained images show good resolution and sensitivity of the radar (see Fig. 12). TARCHI ET AL.: SAR IMAGING WITH NOISE RADAR

Fig. 12. Image obtained with circular SAR (sketch of scene is shown in Fig. 6).

The following objects of the scene (Figs. 10 and 11) are clearly seen in the SAR image (Fig. 12) as the metal parts of the buildings A and B, some of the vent tubes on the top of building A, groups of trees C and D. Fig. 13 shows an interferogram between two radar images obtained at different times. Amplitude threshold with a level of ¡30 dB is applied to this interferogram. It can be seen that phase shifts are close to zero on stable targets like ground and buildings. Phase shifts on less stable targets like trees have a random structure. Measurements with 250 pulses integration at each antenna position showed 1221

V.

Fig. 13. Interferogram obtained with circular SAR. Pixels with signal level below threshold of ¡30 dB are masked by black color.

that the standard deviation of the measured phase from the mean value at the horn antenna was about 10 deg. This error was caused by low reflectivity of the antenna (the return of the antenna was only 12 dB above the noise level). Such precision of the phase measurement enables us to estimate shifts of the objects with 0.8 mm accuracy. Fig. 14 shows a histogram of the phase difference of two images, taking into account pixels with an amplitude higher than ¡30 dB with respect to the maximum. We applied a fixed value of signal-to-noise (SNR) as a threshold rather than coherency (13) because of high temporal stability of the mapped scene during the measurements. A ¡7 deg offset is caused by the radar platform shift. Standard deviation as derived from the histogram is about 12 deg. This leads to submillimeter accuracy in measuring relative displacement in the scene.

CONCLUSION

In this paper, some theoretical and experimental aspects of NW SAR imaging are discussed. In particular, the basics of the NW SAR imaging are presented. The experimental implementation of X-band NW SAR with linear and circular apertures enabled us to obtain high quality radar images of small areas. Evaluation of the phase measurements precision with the help of ground-based NW SAR was carried out on the basis of the obtained interferograms. It has been shown that NW SAR provides quite high precision of displacement measurements between two acquisitions spaced in time. Furthermore, it was demonstrated that NRT may provide rather high performance of radar designed for short-range applications with interferometric capabilities including ground-based SAR and D-InSAR systems. Airborne and spaceborne SAR can also benefit from using the NRT because of their properties like interference immunity, possibility of generating a signal with Gaussian spectra with a low sidelobe level, and randomness of the pulses, which enables using higher PRF than for the deterministic signals. ACKNOWLEDGMENT We thank V. Palamarchuk, O. Zemlyany, and V. Skresanov for the support provided in the design of the antenna, data acquisition and control unit, and in the performance of the field measurements. The support provided by D. Leva during the performance of the field measurements in Ispra is also acknowledged. REFERENCES [1]

Tarchi, D., Rudolf, H., and Luzi, G. SAR interferometry for structural changes detection: A demonstration test on a DAM. In Proceedings of the International Geoscience and Remote Sensing Symposium (IGARSS), Hamburg, Germany, 1999, 1522—1524.

Fig. 14. Phase histogram formed with two images obtained with help of circular NW SAR. 1222


JULY 2010

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

Pieraccini, M., Tarchi, D., Rudolf, H., Leva, D., Luzi, G., and Atzeni, C. Interferometric radar for remote monitoring of building deformations. Electronics Letters, 36, 6 (Mar. 16, 2000), 569—570. Leva, D., Nico, G., Tarchi, D., et al. Temporal analysis of a landslide by means of a ground-based SAR interferometer. IEEE Transactions on Geoscience and Remote Sensing, 41, 4 (Apr. 2003), 745—752. Morrison, K., Bennett, J. C., Cookmartin, G., McDonald, A. J., and Quegan, S. Three-dimensional X-band SAR imaging of a small conifer tree. International Journal of Remote Sensing, 22, 4 (Mar. 10, 2001), 705—710. Lukin, K. A. Noise radar technology. Radiophysics and Electronics, Kharkiv, Institute for Radiophysics and Electronics NASU, 4, 3 (1999), 105—111. Lukin, K. A. The principles of noise radar technology. In Proceedings of the NRTW, Yalta, Crimea, Ukraine, Sept. 18—20, 2002, 13—22. Lukin, K. Radar design using noise/random waveforms. In Proceedings of the International Radar Symposium 2006, Krakow, Poland, May 22—26, 2006, 355—358. Lukin, K. A. Noise radar technology: The principles and short overview. Applied Radio Electronics, 4, 1 (2005), 4—13. Tarchi, D., Leva, D., Lukin, K., et al. Short range imaging applications using noise radar technology. In Proceedings of the EUSAR 2000 3rd European Conference on SAR, Munich, Germany, May 23—25, 2000, 361—364.

TARCHI ET AL.: SAR IMAGING WITH NOISE RADAR

[10]

[11]

[12]

[13]

[14]

[15]

[16]

[17]

Garmatyuk, D. S. and Narayanan, R. M. Ultra-wideband continuous random noise arc-SAR. IEEE Transactions on Geoscience and Remote Sensing, 40, 12 (Dec. 2002), 745—752. Walton, E. K., Fillimon, V., and Gunawan, S. ISAR imaging using UWB noise radar. In Proceedings of the 18th AMTA Symposium, Seattle, WA, Sept. 30—Oct. 3, 1996, 167—171. Bell, D. C. and Narayanan, R. M. ISAR imaging using a coherent ultrawideband random noise radar system. Optical Engineering, 40, 11 (Nov. 2001), 2612—2622. Axelsson, S. R. J. Noise radar for range/Doppler processing and digital beam forming using binary ADC. In Proceedings of the IGARSS’2001, Sydney, Australia, July 9—13, 2001. Lukin, K. Ground based noise-waveform-SAR for monitoring of Chernobyl sarcophagus. In Proceedings of International Radar Symposium (IRS-2005), Berlin, Germany, Sept. 6—8, 2005, 655—659. Mogyla, A. A., Lukin, K., and Shiyan, Y. Ambiguity function of noise pulses train and it’s distortions in noise radar receiver. In Proceedings of the International Radar Symposium, Warszawa, Poland, May 19—21, 2004, 37—42. Vyplavin, P. L., Lukin, K. A., and Mogyla, A. A. Phase errors in noise waveform D-InSAR due to trajectory distortions in synthetic aperture sliding antenna. In Proceedings of SPIE, vol. 6159, Feb. 2006, 407—412. Skolnik, M. I. Introduction to Radar Systems. Columbus, OH: McGraw-Hill, 1980, 581.

1223

Dario Tarchi was born in Florence, Italy, on April 6, 1963. He received the “laurea” in physics from the University of Florence in 1990. Since 1993 he has been with the Joint Research Centre of the European Commission, Ispra, Italy, where he joined the scientific team of the European Microwave Signature Laboratory working on the design and experimental validation of SAR imaging algorithms. He was involved in the design and implementation of a ground-based interferometric SAR system (LISA) as well as in the experimental validation of its use for real-time monitoring of natural hazards, such as landslides and snow avalanches. He was also leading a project on the use of satellite SAR images for the detection and monitoring of oil spill signature on the sea surface. His main research interests concern the application of SAR interferometric techniques for changes in detection in natural and man-made objects and the development and testing of novel SAR concepts and systems, such as parasitic SAR system and noise radar technology, for various applications. Konstantin Lukin (A’01–SM’04–F’09) received his diploma in radiophysics and electronics from Kharkov State University, Ukraine, in 1973. He completed his Candidate of Sciences thesis in IRE NASU and defended it in Moscow State University (MGU) in 1980. He completed his Doctor of Sciences dissertation in physical electronics in IRE NASU and defended it in Kharkov State University in 1989. Since 1973 he has been with the Institute for Radiophysics and Electronics, National Academy of Sciences of Ukraine, IRE NASU, Kharkov. He has been a visiting scientist at the International Center for Theoretical Physics (ICTP, Trieste, Italy) in 1996—1997 and a visiting professor at the Joint Research Center (JRC, Ispra, Italy) in 1998. His current research interests are nonlinear dynamics and chaos in electronic systems; generation and processing of chaotic/noise waveforms and their applications. He is the Chairman of RTO/NATO Task Group on NRT. Dr. Lukin is an author and coauthor of more than 70 journal publications and monograph. Joaquim Fortuny-Guasch (S’93–A’96–SM’08) was born in Tarragona, Spain, in 1964. He received the “Ingeniero” degree in telecommunications engineering from the Technical University of Catalonia (UPC), Barcelona, Spain, in 1988, and the Dr.-Ing. degree in electrical engineering from the University Karlsruhe (TH), Karlsruhe, Germany, in 2001. From 1988—1989, he was a research assistant in the Signal Theory Department at UPC. From 1990—1992, he was a graduate trainee in the RF Division at the European Space Technology Centre of ESA, The Netherlands. Since 1993, he has been with the Directorate General Joint Research Centre of the European Commission, Ispra, Italy, where he is a senior scientific officer. In the last 15 years, he has participated in more than 20 European research projects, which have been funded by EC’s Directorates General Information Society, Enterprise, Environment, and Research and Technological Development in the frame of the 5th, 6th, and 7th Framework Programs. Application areas addressed in these projects include subsurface sensing for humanitarian demining, automatic oil spill detection and classification, through the wall radar imaging, landslide and snow avalanches monitoring with ground-based synthetic aperture radar systems, and definition of certification procedures for ultra-wide band systems. His current research interests are in the fields of radar imaging techniques, multiple input multiple output (MIMO) radar, and UWB techniques for radar and wireless communications. 1224


JULY 2010

Mogyla Anatoly Andreevich was born on May 19, 1956. in 1978 he graduated from Kharkov Institute of Radioelectronics with a specialty in radio engineering. He received the Ph.D. in physics and mathematics in 1998 from NASU. Since 1998 he has been with the Laboratory for Nonlinear Dynamics of Electronic Systems of A. Usikov Institute for Radiophysics and Electronics NASU, where he is now a senior research follow. Area of scientific interests: statistical radiophysics, radiolocation, radio engineering systems, models and methods for signal processing.

Pavlo Leonidovych Vyplavin was born in 1982 in Kharkiv, Ukraine. In 2004 he graduated from Kharkiv National University with the M.S. degree in radiophysics and electronics. Since 2004 he has been with Laboratory of Nonlinear Dynamics of Electronic Systems of Institute of Radiophysics and Electronics of NAS of Ukraine. In 2005 he took part in inservice training at Joint Research Centre of the European Commission, Ispra, Italy where he studied SAR imaging using noise radars. His research interests are in the field of noise signal processing, noise radars, and SAR imaging.

Alois J. Sieber (F’99) received the M.S. degree in physics from the Technical University of Karslruhe, Germany in 1971, the Ph.D. in 1973 from the University of Tu¨ bingen, Germany, and the Habilitation degree in remote sensing in 1985 from the University of Stuttgart, Germany. Since 1991, he has been with the Joint Research Centre of the European Commission, Ispra, Italy, as the Head of the unit “Security Technologies Assessment” at the Institute for the Protection and Security of the Citizen. Since 1986, he has been a senior lecturer at the University of Stuttgart, and since 1977, a member of the German Institute for Radio Frequency Technology (DFVLR). His research interests are in radar and communications techniques for security applications. TARCHI ET AL.: SAR IMAGING WITH NOISE RADAR

1225