High‐range‐resolution Two‐dimensional Imaging Using Frequency Diversity Multiple‐input–multiple‐output Sonar
Xionghou Liu,Chao Sun,Yixin Yang,Jie Zhuo
DOI: https://doi.org/10.1049/iet-rsn.2014.0559
2016-01-01
IET Radar, Sonar & Navigation
Abstract:IET Radar, Sonar & NavigationVolume 10, Issue 5 p. 983-991 Research ArticleFree Access High-range-resolution two-dimensional imaging using frequency diversity multiple-input–multiple-output sonar Correction(s) for this article Erratum: ‘High-range-resolution two-dimensional imaging using frequency diversity multiple-input–multiple-output sonar’ Volume 10Issue 4IET Radar, Sonar & Navigation pages: 816-816 First Published online: April 1, 2016 Xionghou Liu, Xionghou Liu School of Marine Science and Technology, Northwestern Polytechnical University, Xi'an, 710072 People's Republic of China State Key Laboratory of Acoustics, Institute of Acoustics, Chinese Academy of Sciences, Beijing, 100190 People's Republic of ChinaSearch for more papers by this authorChao Sun, Corresponding Author Chao Sun csun@nwpu.edu.cn State Key Laboratory of Acoustics, Institute of Acoustics, Chinese Academy of Sciences, Beijing, 100190 People's Republic of ChinaSearch for more papers by this authorYixin Yang, Yixin Yang School of Marine Science and Technology, Northwestern Polytechnical University, Xi'an, 710072 People's Republic of ChinaSearch for more papers by this authorJie Zhuo, Jie Zhuo School of Marine Science and Technology, Northwestern Polytechnical University, Xi'an, 710072 People's Republic of ChinaSearch for more papers by this author Xionghou Liu, Xionghou Liu School of Marine Science and Technology, Northwestern Polytechnical University, Xi'an, 710072 People's Republic of China State Key Laboratory of Acoustics, Institute of Acoustics, Chinese Academy of Sciences, Beijing, 100190 People's Republic of ChinaSearch for more papers by this authorChao Sun, Corresponding Author Chao Sun csun@nwpu.edu.cn State Key Laboratory of Acoustics, Institute of Acoustics, Chinese Academy of Sciences, Beijing, 100190 People's Republic of ChinaSearch for more papers by this authorYixin Yang, Yixin Yang School of Marine Science and Technology, Northwestern Polytechnical University, Xi'an, 710072 People's Republic of ChinaSearch for more papers by this authorJie Zhuo, Jie Zhuo School of Marine Science and Technology, Northwestern Polytechnical University, Xi'an, 710072 People's Republic of ChinaSearch for more papers by this author First published: 01 June 2016 https://doi.org/10.1049/iet-rsn.2014.0559Citations: 9AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinkedInRedditWechat Abstract A high-range-resolution imaging method for a multiple-input–multiple-output sonar is proposed. In this method, a set of frequency diversity linear frequency modulation (FD-LFM) pulses is transmitted, with a zero bandgap between two adjacent FD-LFM pulses to suppress cross-correlation functions. At the processing end, the synthesised matched filtering, which combines the matched filtering and the frequency band synthesising into one step, is proposed to reduce the computation burden. Moreover, the frequency-domain weighting is adopted to suppress to the sidelobes of the synthesised auto-correlation function to further improve the image quality. The effectiveness of the proposed method is validated via the theoretical analysis, numerical simulations and water tank experiments. 1 Introduction An imaging sonar is useful to display the intensity distribution of an underwater scene [1-3]. To improve the imaging result, increasing the range resolution is necessary, which is determined by the obtainable signal bandwidth [1, 4-11]. However, large-bandwidth signals require a large instant system bandwidth, making the imaging sonar suffer from a high hardware cost [10]. To reduce the system cost and obtain a fine range resolution, a stepped-frequency system using frequency diversity (FD) pulse-trains provides a feasible way [7-11]. Nevertheless, due to the long illumination time, the stepped-frequency system encounters the range migration problem when relative motion between the sonar and the target occurs. The range migration defocuses the target response into several range bins and severely degrades the imaging resolution. Although several methods have been reported with the ability to align the migrated ranges and refocus the target [12-16], the long illumination time is always a harsh restrict, especially for the manoeuvring target imaging [17, 18]. Different from the stepped-frequency system, an FD multiple-input–multiple-output (FD-MIMO) sonar emits the individual pulses of the pulse-train from different transmitting elements simultaneously [19]. At the receiving end, echoes of different pulses are separated via matched filtering at each receiver, and then outputs of the matched filters at all receivers are processed jointly to form a high-resolution range profile [20-26]. By doing so, the illumination time is equal to the length of a single pulse, which is generally much shorter than that of the stepped-frequency system. A shorter illumination time means fewer range migration, which consequently causes less performance degradation in the FD-MIMO sonar image. In general, two different processing schemes are adopted at the receiving end of the FD-MIMO sonar in obtaining a fine range resolution. The first one is basically a parameter estimation approach [19-23]. Owing to the frequency offset across the transmitting array, phase differences occur at different target ranges (e.g. see (2) in [20] and (10) in [23]). High-resolution methods such as the subspace decomposition algorithms, which are commonly used in direction-of-arrival estimation, are adopted to estimate the target range precisely and then provide a high-resolution range profile [23]. Nevertheless, this method fails to work when the underwater target scene is continuous where the rank deficient problem occurs due to the limited number of transmitting elements. In the second scheme, the FD-MIMO sonar synthesises waveforms with relatively small bandwidths into a waveform with a large bandwidth to provide a similar range resolution as the stepped-frequency system [24-26]. In this paper, we focus on the second scheme in implementing the FD-MIMO sonar image, which will be addressed as the MIMO sonar for simplicity from now on. Conventionally, a stepped-frequency system transmits a large number of pulses in the pulse-train with overlapped bands. However, when implemented in the MIMO sonar, the large number of pulses requires a large number of transmitting elements, which increase the system cost greatly. Besides, although overlapped bands are effective for the stepped-frequency system to suppress the sidelobes of the synthesised auto-correlation function (ACF-SLs), they also bring high-cross-correlation functions (CCFs) for the MIMO sonar. Moreover, the overlapped setting is also a waste of the final obtainable bandwidth. In our method, a small number of FD linear frequency modulation (FD-LFM) pulses are utilised. On the basis of the fact that the CCF peak and the ACF-SLs cannot be simultaneously reduced, a zero bandgap is used to suppress the CCF and the frequency-domain weighting after the band synthesis is adopted to reduce the ACF-SLs. Furthermore, the synthesised matched filtering (SMF) is proposed to reduce the computation burden. The SMF combines the matched filtering and the large bandwidth synthesising into one step, showing a similar imaging ability as the traditional processing procedure while requiring a lower computation burden. Numerical simulations and water tank experiments are carried out to validate the effectiveness of the proposed method. The remainders of the text are organised as follows. In Section 2, the signal model of the proposed MIMO configuration is presented. The imaging procedure with the SMF is given in Section 3. In Section 4, the parameters which determine the imaging performance are investigated. To demonstrate the effectiveness of the imaging method, numerical simulations and water tank experiments are provided in Section 5. Finally, conclusions are given in Section 6. 2 Signal model Assume that the MIMO sonar is composed of an M-element transmitting uniform linear array (ULA) and an N-element receiving ULA. Both ULAs are located on the x-axis and centred at the coordinate origin. The inter-element spacing of the receiving ULA, dr, is set to dr = λ/2, where λ denotes the wavelength corresponding to the operating frequency which is the centre frequency of the whole frequency band herein. Moreover, that of the transmitting ULA, dt, is set to dt = 2dr (Note that the relationship in [27], that is, dt = Ndr, is not required here.). The array layout of the MIMO sonar is depicted in Fig. 1. Fig. 1Open in figure viewerPowerPoint Array layouts of the MIMO sonar composed of two ULAs located on the same straight line The MIMO sonar emits FD-LFM pulses from the M transmitting elements simultaneously. Suppose all pulses share the same bandwidth B0, the same pulse length T0, and the same bandgap ΔB. The mth LFM pulse applied to the mth transmitting element can be expressed as (1) where (2) and f0 denotes the centre frequency of the mth pulse and the whole frequency band. To simplify the investigation, the transmission loss and the medium absorption loss are ignored. The MIMO sonar is supposed to be stationary to the target during the illumination. The far-field seabed target is modelled as P ideal scatterers. Accordingly, the echo at the nth (n = 1, 2, …, N) receiving element, denoted by xn(t), can be expressed as the sum of M LFM pulses with different time delays and attenuations [27-30] (3) where σp denotes the reflectivity of the pth scatterer, denotes the time delay of the mth (m = 1, 2, …, M) transmitted pulse from the mth transmitting element to the pth scatterer, denotes the time delay from the pth scatterer to the nth receiving element, and zn(t) denotes the additive white Gaussian noise at the nth receiving element. At the receiving end, using copies of the M LFM pulses to matched filter echoes at each of the N-element receiving ULA, resulting in MN outputs in total. By doing so, the pulse-compressed echoes corresponding to different frequency bands are separated [26]. Specifically, the output of the mth matched filter attached to the nth output (denoted by ym, n(t)) can be expressed as (4) where (5) denotes the impulse response function of the matched filter associated with the mth LFM pulse, * denotes the convolution, and []c denotes the conjugate. As described in [27-30], a matched filter can be considered as a correlator under a low Doppler shift condition, and hence the output of a matched filter can be expressed as the sum of auto- and cross-correlation terms. Since transmitting pulses are independent from the noise and the CCFs are low enough when compared with the ACF peak, (4) can be simplified as the summation of auto-correlation terms [27-30] (6) where (7) denotes the ACF of the mth LFM pulse. From (7) we know that after the matched filtering, echo components corresponding to different LFM pulses are picked out. Thus, the MN outputs of matched filters can be divided into M groups and each group contains N components corresponding to a certain frequency band. To illustrate clearly, the echo processing structure of the MIMO sonar is depicted in Fig. 2, where MFm indicates the mth matched filter. Fig. 2Open in figure viewerPowerPoint Echo processing structure of the MIMO sonar, where MFm indicates the mth matched filter 3 High-range-resolution imaging processing 3.1 Traditional imaging procedure As shown in Fig. 2, echoes at the N-element receiving ULA are compressed into M groups of echoes corresponding to different frequency bands after matched filtering. Afterwards, the M groups of echoes are beamformed individually and processed jointly so as to synthesise an image with high-range resolutions. This procedure is quite different from the methods in [18, 27-30] where a high-cross-range resolution is achieved while the range resolution remains unchanged. Taking the narrowband case as an example, the phase-shift beamformer (PSBF) is applied to the N-element receiving ULA. For the mth group, the output of the qth (q = 1, 2, …, Q) PSBF, , is (8) where (9) denotes the complex weight at the nth receiving element with An as the amplitude weight and θq denotes the steering direction of the qth beam, which is the relative angle to the broadside of the MIMO sonar. After compensating for phase shifts between M transmitting elements, beam outputs covering different frequency bands are synthesised into a single beam output with a large bandwidth (10) where (11) denotes the complex weight at the mth transmitting element. Taking the absolute value of all beam outputs as the output intensities and map the results onto the angle-range plane, the two-dimensional (2D) image is obtained. The traditional imaging procedure is depicted in Fig. 3, where the matched filtering is ignored for brevity. Fig. 3Open in figure viewerPowerPoint Traditional imaging procedure to form a 2D image with a high-range resolution, where the matched filtering is ignored for brevity and BF is the abbreviation of beamforming 3.2 Synthesised matched filtering However, from Fig. 3 we know that MN matched filters are required in the traditional imaging procedure to gain a high-range resolution, which results in a heavy computation burden. To reduce the computation load, we propose a new imaging procedure which combines the matched filtering and the large bandwidth synthesising together in order to reduce the number of matched filters. Substituting (8) into (10) yields (12) Further, substituting (4) into (12), we get (13) As seen from (13), the matched filtering is conducted after the receiving beamforming, and the impulse response function is , which is the output after the large bandwidth synthesising. Accordingly, the matched filtering and the large bandwidth synthesising are realised in one step, which is referred to as the SMF. Thus, only Q matched filters are required. Usually, the product of the element number, that is, MN, is larger than the number of beamformers, that is, Q; thus, the computation burden is effectively reduced by the SMF. The new imaging procedure using the SMF is given in Fig. 4. Fig. 4Open in figure viewerPowerPoint New imaging procedure using the SMF 4 Resolution and parameter constraint 4.1 Range resolution Recalling (10) and (13), the beam outputs corresponding to M frequency bands are summed into the beam output with a single large bandwidth. Thus, the synthesised ACF can be expressed as (14) Substituting (2) into (14), the synthesised ACF turns into (15) From (15), the synthesised ACF is the product of two terms: the first term is a sinc function, which essentially is the ACF of a single LFM pulse; the second term is similar to a beampattern of an ULA (we refer to it as the pseudo beampattern). To illustrate intuitively, a set of four FD-LFM pulses (the parameters are given in Table 1) is adopted. Moreover, the sinc function (the first term), the pseudo beampattern (the second term), and the synthesised ACF are given in Fig. 5. Table 1. Parameters of a set of four LFM pulses M Frequency bands, kHz B0, kHz ΔB, kHz T0, ms 4 390–395 5 0 8 395–400 400–405 405–410 Fig. 5Open in figure viewerPowerPoint Sinc function (the first term), the pseudo beampattern (the second term), and the synthesised ACF a Sinc function (the first term) and the pseudo beampattern (the second term) b Single ACF (corresponding to a single LFM pulse) and the synthesised ACF (corresponding to all the LFM pulses) at B0 = 5 kHz, ΔB = 0 kHz, and T0 = 8 ms. Only values within [−1, 1] ms are zoomed As seen from Fig. 5a, the pseudo beampattern (the second term) is similar to an ULA beampattern with grating lobes. In Fig. 5b, the mainlobe of the synthesised ACF is narrower than that of the single ACF, showing that a better range resolution is obtained after the band synthesis. Accordingly, the final bandwidth, which determines the range resolution of the MIMO sonar, is (16) 4.2 Angle resolution Owing to different frequency bands at different transmitting elements, outputs of matched filters cannot be coherently processed to form a large aperture. Consequently, the transmitting aperture is lost and the cross-range resolution is only determined by the receiving array. Therefore, the effective aperture of the MIMO sonar is the same as that of the single-input–multiple-output (SIMO) (17) Note that the effective aperture here is quite different from those in [18, 27-30] where the full aperture produced by the spatial convolution of transmitting and receiving arrays is available. 4.3 Range SL As the bandgap ΔB increases, the CCF peak becomes low while the ACF-SL grows high. With the use of four FD-LFM pulses with B0 = 5 kHz and T0 = 8 ms, such relationship is depicted in Fig. 6a, where the CCF peak and the synthesised ACF-SLs are normalised to the synthesised ACF peak. According to Fig. 6a, the bandgap of ΔB = 0 kHz is chosen as a tradeoff to suppress CCFs and the synthesised ACF-SLs. To further suppress the ACF-SLs, the frequency weighting after the band synthesis is used, that is, weighting synthesised outputs of beamformers in the frequency domain [6]. As an example, the synthesised ACF in Fig. 5b and a weighted one by using a Chebyshev window (which achieves the lowest SLs while resulting in the narrowest mainlobe) with an SL level of −25 dB are depicted in Fig. 6b, which shows the frequency weighting can suppress the synthesised ACF-SLs effectively. Fig. 6Open in figure viewerPowerPoint Range SL a Relationship between the bandgap ΔB and the CCF peak and the synthesised ACF-SL peak b Synthesised ACF and weighted synthesised ACF, where a −25 dB Chebyshev window is used 5 Computer simulations and water tank experiments 5.1 Imaging a single-scatterer target First, the range resolution of the SIMO and the MIMO sonars are compared by imaging a single-scatterer target located at (15°, 20 m). The SIMO sonar is composed of a single transmitting element and a 48-element receiving ULA with dr = λ/2. The MIMO sonar is composed of a 4-element transmitting ULA with dt = λ and a 48-element receiving ULA with dr = λ/2. A single LFM pulse (with a frequency band of 397.5–402.5 kHz and a pulse length of 8 ms) is applied to the single transmitting element of the SIMO sonar, and a set of four FD-LFM pulses with the parameters given in Table 1 is applied to the 4-element transmitting ULA of the MIMO sonar. At the receiving end, the sampling frequency is set to fs = 400 kHz. The signal-to-noise ratio at each receiving element, defined as the ratio of the signal power to the noise power, is set to 10 dB with an additive white Gaussian noise (the noise variance is calculated over the frequency band from 0 Hz to fs/2). The beam steering direction is from −45° to 45° with a step of 1°. The traditional imaging procedure in Fig. 3 is adopted to process simulated echoes to produce the 2D image. To suppress angle SLs, a Chebyshev window with a −25 dB SL level is applied to the N-element receiving ULA. Moreover, a rectangular window is chosen for matched filtering. The imaging results of the SIMO and the MIMO sonars are given in Fig. 7. Comparing the 2D images in Fig. 7a and Fig. 7b, it can be seen that the MIMO sonar has a much better range resolution than the SIMO sonar. For quantitative analysis, the range slices at the angle of 15° are depicted in Fig. 7c. By calculating, the range sidelobe peak (SLP) of the SIMO sonar (−13.64 dB) is similar to that of the MIMO sonar (−13.47 dB). Moreover, the range resolution (−3 dB beamwidth) of the SIMO sonar is about 0.133 m while that of the MIMO sonar is about 0.036 m, showing that the range resolution of the MIMO sonar is approximately four times that of the SIMO sonar. The angle slices at the range of 20 m are depicted in Fig. 7d, which shows that the SIMO and MIMO sonars have the same angle resolution due to the sameness of the receiving apertures (Owing to the adoption of the Chebyshev weight (−25 dB), the SLs of angle slices are equal to or lower than −25 dB in Fig. 7d.). Fig. 7Open in figure viewerPowerPoint Imaging results of the SIMO and the MIMO sonars a Imaging results of the SIMO sonar b MIMO sonar using the traditional imaging procedure c Range cuts at the angle of 15° d Angle cuts at the range of 20 m Furthermore, to test the performances of the traditional and the new imaging procedures (depicted in Fig. 3 and Fig. 4, respectively) of the MIMO sonar, the corresponding imaging results are compared and analysed. The 2D image produced by the new imaging procedure is shown in Fig. 8a, which is similar to that shown in Fig. 7b produced by the traditional imaging procedure. The range and angle slices of 2D images produced by the two imaging procedures are depicted in Fig. 8b and Fig. 8c, respectively. By calculating the −3 dB beamwidth, the range resolutions corresponding to the two imaging procedures are both 0.036 m and the range SL levels are both −13.54 dB. In addition, the angle slices are superposed with each other. As a result, it can be concluded that the two imaging procedures have the same range and the angle resolutions. Fig. 8Open in figure viewerPowerPoint 2D image produced by the new imaging procedure a Imaging result of the MIMO sonar using the new imaging procedure b Range slices at the angle of 15° c Angle slices at the range of 20 m d Computation times of the two imaging procedures To compare the computation loads of the two imaging procedures quantitatively, the computation times are calculated. Toward this end, the programmes are run on a personal computer with Intel Core(TM) i3-3220 central processing unit at 3.30 GHz and a 4 GB memory, and MATLAB functions CLOCK and ETIME are used to calculate the computation time. Totally hundred runs are implemented for each procedure and the corresponding computation times are compared, as shown in Fig. 8d. By averaging the results of these hundred runs, the computation time of the new imaging procedure is about 70 ms, whereas that of the traditional one is about 170 ms. Since the new imaging procedure with SMF only needs 91 matched filters while the traditional one requires 4 × 48 = 192 in this situation, the computation burden of the new imaging procedure is much smaller than that of the traditional one. 5.2 Imaging a multiple-scatterer target In this simulation part, a more complicated target composed of multiple scatterers is considered and the SLP-ACF suppression ability of the frequency weighting method is tested. The new imaging procedure in Fig. 4 is applied to the MIMO sonar. To suppress ACF-SLs, a Chebyshev window with a −25 dB SL level is applied to the frequency-domain outputs of all beamformers. Other simulation parameters are the same as previous. The original angle and range distributions of scatterers in the target are depicted in Fig. 9a. The imaging results of the MIMO sonar are given in Fig. 9b, which suffers from high-range SLs. The 2D image after the frequency weighting is depicted in Fig. 9c, where the range SLs are suppressed to a lower level. Moreover, the projection results of the imaging results along the angle axis are given in Fig. 9d, where the range SL peak is reduced from −10.18 to −18.12 dB, showing that the frequency weighting method can suppress the ACF-SLs effectively. Fig. 9Open in figure viewerPowerPoint Original angle and range distributions of scatterers in the target a Original angle and range distributions of scatterers in the target b 2D image produced by the MIMO sonar using the new imaging procedure c Result after the ACF-SL suppression using the frequency weighting method d Projection images along the angle axis 5.3 Tank experiments To demonstrate the high-range resolution of the proposed method, an imaging experiment in an anechoic water tank was carried out. The schematic diagram of the tank experiment is shown in Fig. 10. The tank is of a dimension of 20 m × 8 m × 6 m. Three projectors and 26 hydrophones (forming a 26-element receiving ULA) were used and placed 3.6 m deep in the water. The inter-spacing of projectors was 0.02 m and that of the hydrophones was 0.01 m. The geometry centre of the ULA was (0, 0, −3.6) m, and the coordinate of the nth (n = 1, 2, …, 26) hydrophone was (n−12.5) × 0.01 m and those of three projectors were (−0.02, 0, −3.4), (0, 0, −3.4), and (0.02, 0, −3.4) m, respectively. An aluminium half-cylindrical shell filled with air was used as the target, positioned at (−0.9, 7, −3.5) m. Fig. 10Open in figure viewerPowerPoint Schematic diagram of the tank experiment, where Tm (m = 1, 2, 3) denotes the mth projector Parameters of three FD-LFM pulses, applied to the three projectors, are given in Table 2. An SIMO and two MIMO sonar configurations are used and the corresponding compositions are given in Table 3. The sampling frequency of each receiving channel is 250 kHz. The new imaging procedure with SMF is adopted by the two MIMO sonars. Moreover, the rectangular window is utilised in beamforming and matched filtering. Table 2. Parameters of LFM pulses in experiment Projector Frequency band, kHz Pulse length, ms 1 55–65 8 2 65–75 3 75–85 Table 3. Compositions of SIMO and MIMO sonars in experiment Sonar type Composition SIMO projector 2 and ULA MIMO/2T projectors 1, 2, and ULA MIMO/3T projectors 1, 2, 3, and ULA The 2D imaging results are depicted in Fig. 11, from which it can be seen qualitatively that the MIMO/3T sonar has the best range resolution while the SIMO sonar has the worst. For quantitative comparison, the range and angle slices of imaging results are shown in Fig. 12. By calculating, the range resolutions (−3 dB beamwidth) of the SIMO, MIMO/2T, and MIMO/3T are 0.069, 0.032, and 0.021 m, respectively. According to Fig. 11 and Fig. 12a, the MIMO/2T sonar shows a better range resolution than the SIMO sonar due to a larger bandwidth (i.e. 20 kHz) synthesised at the processing end; and the MIMO/3T sonar has the best range resolution since the largest bandwidth (i.e. 30 kHz) is synthesised. Consequently, the range slices in Fig. 12a are in coincidence with the theoretical analysis that more transmitting elements ensure a larger synthesised bandwidth, and therefore a better range resolution. As seen from Fig. 12b, the angle resolutions are approximately the same, showing that the angle resolution of the MIMO sonar is only determined by the aperture of the receiving array (i.e. a 26-element ULA in the experiment) in this situation. Fig. 11Open in figure viewerPowerPoint 2D images produced by a SIMO sonar b MIMO/2T sonar c MIMO/3T sonar Fig. 12Open in figure viewerPowerPoint Range and angle slices of imaging results a Range b Angle slices of experimental results 6 Conclusions In this paper, we showed that an MIMO sonar could utilise the spatial-domain stepped-frequency technique to synthesise a large-bandwidth waveform during a single ping. Moreover, a 2D image with a high-range resolution was produced by the proposed MIMO configuration. For practical applications, we used a small number of FD-LFM pulses and chosen the zero bandgap to suppress CCFs. To suppress the synthesised ACF-SLs, the frequency weighting after the band synthesis was adopted. Furthermore, a new imaging procedure using the SMF was proposed to reduce the number of matched filters and the computation burden in turn. Numerical simulations and water tank experiments were conducted and the results were in good agreements with the theoretical analysis. 7 Acknowledgments Great thanks to the staff of Institute of Acoustic Engineering for the assistance on the experiment. This work was supported in part by the National Natural Science Foundation of China under Grant (51509204), the Opening Project of State Key Laboratory of Acoustics (SKLA201501), and the Fundamental Research Funds for the Central Universities (3102015ZY011). 8 References 1Sutton, J.L.: ‘Underwater acoustic imaging’, Proc. IEEE, 1979, 67, (4), pp. 554– 566 (doi: https://doi.org/10.1109/PROC.1979.11283) 2McHugh, R., Shaw, S., Taylor, N.: ‘Efficient digital signal processing algorithm for sonar imaging’, IEE Proc. Radar Sonar Navig., 1996, 143, (3), pp. 149– 156 (doi: https://doi.org/10.1049/ip-rsn:19960583) 3Bjørnø, L.: ‘ Developments in sonar and array technologies’. Proc. 2011 IEEE Symp. on Underwater Technology (UT) and 2011 Workshop on Scientific Use of Submarine Cables and Related Technologies (SSC), Tokyo, Japan, April 2011, pp. 1– 11 4Knight, M.C., Pridham, R.G., Kay, S.M.: ‘Digital signal processing for sonar’, Proc. IEEE, 1981, 69, (11), pp. 1451– 1506 (doi: https://doi.org/10.1109/PROC.1981.12186) 5Murino, V., Trucco, A.: ‘Three-dimensional image generation and processing in underwater acoustic vision’, Proc. IEEE, 2000, 88, (12), pp. 1903– 1948 (doi: https://doi.org/10.1109/5.899059) 6Van Trees, H.L.: ‘ Optimum array processing: part 4 of detection, estimation, and modulation theory’ ( John Wiley & Sons Inc., Hoboken, NJ, 2002) 7Levanon, N.: ‘Stepped-frequency pulse-train radar signal’, IEE Proc. Radar Sonar Navig., 2002, 194, (6), pp. 297– 309 (doi: https://doi.org/10.1049/ip-rsn:20020432) 8Levanon, N., Mozeson, E.: ‘Nullifying ACF grating lobes in stepped-frequency train of LFM pulses’, IEEE Trans. Aerosp. Electron. Syst., 2003, 39, (2), pp. 694– 703 (doi: https://doi.org/10.1109/TAES.2003.1207275) 9Rabideau, D.J.: ‘ Nonlinear synthetic wideband waveforms’. Proc. 2002 IEEE Radar Conf., Long Beach, CA, April 2002, pp. 212– 219 10Maron, D.E.: ‘ Frequency-jumped burst waveforms with stretch processing’. Proc. IEEE 1990 Int. Radar Conf., Arlington, VA, May 1990, pp. 274– 279 11Gladkova, I., Chebanov, D.: ‘Grating lobes suppression in stepped-frequency pulse’, IEEE Trans. Aerosp. Electron. Syst., 2008, 44, (4), pp. 1265– 1275 (doi: https://doi.org/10.1109/TAES.2008.4667708) 12Chen, V.C., Qian, S.: ‘Joint time-frequency transform for radar range-Doppler imaging’, IEEE Trans. Aerosp. Electron. Syst., 1998, 34, (2), pp. 486– 499 (doi: https://doi.org/10.1109/7.670330) 13Liu, Y.M., Meng, H.D., Li, G. et al: ‘Velocity estimation and range shift compensation for high range resolution profiling in stepped-frequency radar’, IEEE Geosci. Remote Sens. Lett., 2000, 7, (4), pp. 791– 795 (doi: https://doi.org/10.1109/LGRS.2010.2047492) 14Cheng, H.Y., Liu, Y.X., Jiang, W.D. et al: ‘A new approach for synthesizing the range profile of moving targets via stepped-frequency waveforms’, IEEE Geosci. Remote Sens. Lett., 2006, 3, (3), pp. 406– 409 (doi: https://doi.org/10.1109/LGRS.2006.873874) 15Zwart, J.P., van der Heiden, R., Gelseman, S. et al: ‘Fast translation invariant classification of HRR range profiles in a zero phase representation’, IEE Proc. Radar Sonar Navig., 2003, 150, (6), pp. 411– 418 (doi: https://doi.org/10.1049/ip-rsn:20030428) 16Zhang, L., Qiao, Z.J., Xing, M.D. et al: ‘High-resolution ISAR imaging with sparse stepped-frequency waveforms’, IEEE Trans. Geosci. Remote Sens., 2011, 49, (11), pp. 4630– 4651 (doi: https://doi.org/10.1109/TGRS.2011.2151865) 17Wang, D.W., Ma, X.Y., Su, Y.: ‘Two-dimensional imaging via a narrowband MIMO radar system with two perpendicular linear arrays’, IEEE Trans. Image Process., 2010, 19, (5), pp. 1269– 1279 (doi: https://doi.org/10.1109/TIP.2009.2039374) 18Wang, D.W., Ma, X.Y., Chen, A.L. et al: ‘High-resolution imaging using a wideband MIMO radar system with two distributed arrays’, IEEE Trans. Image Process., 2010, 19, (5), pp. 1280– 1289 (doi: https://doi.org/10.1109/TIP.2009.2039623) 19Li, J., Stoica, P.: ‘ MIMO radar-diversity means superiority’, in J. Li, P. Stoica (EDs.): ‘ MIMO radar signal processing’ ( John Wiley & Sons Inc., Hoboken, NJ, 2009), pp. 1– 64 20Antonik, P., Wicks, M.C.: ‘ Frequency diverse array radars’. Proc. 2006 IEEE Conf. Radar, Verona, NY, April 2006, pp. 215– 217 21Antonik, P., Wicks, M.C., Griffiths, H.D. et al: ‘ Multi-mission multi-mode waveform diversity’. Proc. 2006 IEEE Conf. Radar, Verona, NY, April 2006, pp. 580– 582 22Wang, W.Q.: ‘Range-angle dependent transmit beampattern synthesis for linear frequency diverse arrays’, IEEE Trans. Antennas Propag., 2013, 61, (8), pp. 4073– 4081 (doi: https://doi.org/10.1109/TAP.2013.2260515) 23Belfiori, F., Van Rossum, W., Hoogeboom, P.: ‘Coherent MUSIC technique for range angle information retrieval – application to a frequency-modulated continuous wave MIMO radar’, IET Radar Sonar Navig., 2014, 8, (2), pp. 75– 83 (doi: https://doi.org/10.1049/iet-rsn.2013.0121) 24Dai, X.Z., Xu, J., Ye, C.M. et al: ‘ Low-sidelobe HRR profiling based on the FDLFM-MIMO radar’. Proc. First Asian and Pacific Conf. Synthetic Aperture Radar, Huangshan, China, November 2007, pp. 132– 135 25Dai, X.Z., Xu, J., Peng, Y.N.: ‘ High resolution frequency MIMO radar’. Proc. 2007 IEEE Radar Conf., Boston, MA, April 2007, pp. 693– 698 26Liu, X.H., Sun, C., Zhuo, J. et al: ‘ High-resolution 2-D imaging using ultra-wideband MIMO sonar’. Proc. IEEE Int. Conf. Signal Processing, Communications and Computing (ICSPCC), Kunming, China, August 2013, pp. 1– 5 27Sun, C.X., Liu, H., Zhuo, J. et al: ‘ High-resolution 2-D sector-scan imaging using MIMO sonar with narrowband LFM pulses’. Proc. OCEANS'13 MTS/IEEE, San Diego, CA, September 2013, pp. 1– 5 28Liu, X.H., Sun, C., Zhuo, J. et al: ‘ Devising MIMO arrays for underwater 3-D short-range imaging’. Proc. OCEANS'12 MTS/IEEE, Hampton Roads, VA, October 2012, pp. 1– 7 29Liu, X.H., Sun, C., Yi, F. et al: ‘Underwater three-dimensional imaging using narrowband MIMO array’, Sci. China Phys. Mech. Astron., 2013, 56, (7), pp. 1346– 1354 (doi: https://doi.org/10.1007/s11433-013-5117-2) 30Liu, X.H., Sun, C., Zhuo, J. et al: ‘High-resolution swath bathymetry using MIMO sonar system’, J. Syst. Eng. Electron., 2014, 25, (5), pp. 760– 768 (doi: https://doi.org/10.1109/JSEE.2014.00088) Citing Literature Volume10, Issue5June 2016Pages 983-991 FiguresReferencesRelatedInformation