Email alert
2021 Vol. 43, No. 3
Display Method:
2021, 43(3): .
Abstract:
2021, 43(3): 509-515.
doi: 10.11999/JEIT200768
Abstract:
Radar detection of Unmanned Aerial Vehicle (UAV) is a big problem. To address this problem, a detection method with dual polarization radar is proposed. First, radar reduces the detection threshold to make the UAV be detected by using the traditional two dimensional cell averaging constant false alarm probability detector in two polarization channels respectively. However, some false targets caused by clutter are also detected by radar at the same time. To eliminate these false targets, the detection results are integrated and the second detection is carried out for the integrated results in two polarization channels respectively. Then, the second detection results in two polarization channels are matched to further eliminate the false targets. Outdoor experiment is carried out. The processing results for the real data demonstrate that the second types of UAV can be effectively detected and the false targets caused by clutter can be eliminated at the same time with the proposed method.
Radar detection of Unmanned Aerial Vehicle (UAV) is a big problem. To address this problem, a detection method with dual polarization radar is proposed. First, radar reduces the detection threshold to make the UAV be detected by using the traditional two dimensional cell averaging constant false alarm probability detector in two polarization channels respectively. However, some false targets caused by clutter are also detected by radar at the same time. To eliminate these false targets, the detection results are integrated and the second detection is carried out for the integrated results in two polarization channels respectively. Then, the second detection results in two polarization channels are matched to further eliminate the false targets. Outdoor experiment is carried out. The processing results for the real data demonstrate that the second types of UAV can be effectively detected and the false targets caused by clutter can be eliminated at the same time with the proposed method.
2021, 43(3): 516-522.
doi: 10.11999/JEIT200229
Abstract:
To solve the de-correlation problem due to Doppler frequency bias for multistatic radar system, a main-lobe jamming suppression algorithm for multistatic radar based on range-Doppler bias offsetting is proposed. First, based on the cross correlation function of jamming signals of different radars, the effect of Doppler frequency bias on the cross correlation function is analyzed. Then, the time delay and Doppler frequency bias can be estimated and offset by maximizing the two dimensional correlation function of time and frequency, in order to suppress the jamming. What’s more, to reduce the amount of calculation, the time delay can be obtained firstly by cross correlation function of signal amplitude before estimating the Doppler frequency bias. Simulation results illustrate that the jamming can be suppressed effectively after offsetting the Doppler frequency bias and signal cancellation processing.
To solve the de-correlation problem due to Doppler frequency bias for multistatic radar system, a main-lobe jamming suppression algorithm for multistatic radar based on range-Doppler bias offsetting is proposed. First, based on the cross correlation function of jamming signals of different radars, the effect of Doppler frequency bias on the cross correlation function is analyzed. Then, the time delay and Doppler frequency bias can be estimated and offset by maximizing the two dimensional correlation function of time and frequency, in order to suppress the jamming. What’s more, to reduce the amount of calculation, the time delay can be obtained firstly by cross correlation function of signal amplitude before estimating the Doppler frequency bias. Simulation results illustrate that the jamming can be suppressed effectively after offsetting the Doppler frequency bias and signal cancellation processing.
2021, 43(3): 523-530.
doi: 10.11999/JEIT200644
Abstract:
In this paper, under the generalized Pareto distributed sea clutter, the Constant False Alarm Rate (CFAR) properties of Cell-Averaging (CA) and Order-Statistic (OS) non-coherent detectors are studied, the false alarm probability formulas of the two non-coherent detectors are derived, and it is found that the two detectors are CFAR with respect to the scale parameter of sea clutter. However, the two detectors do not have CFAR with respect to speckle covariance matrix structure and shape parameter of sea clutter. In order to ensure CFAR detection in the overall scene, the correlated sea clutter is decorrelated via whitening and the detection threshold matching the shape parameters of sea clutter, number of accumulated pulses, and number of reference cells is applied via lookup tables. In this case, the experimental results show that the two non-coherent detectors can ensure CFAR in the overall scene.
In this paper, under the generalized Pareto distributed sea clutter, the Constant False Alarm Rate (CFAR) properties of Cell-Averaging (CA) and Order-Statistic (OS) non-coherent detectors are studied, the false alarm probability formulas of the two non-coherent detectors are derived, and it is found that the two detectors are CFAR with respect to the scale parameter of sea clutter. However, the two detectors do not have CFAR with respect to speckle covariance matrix structure and shape parameter of sea clutter. In order to ensure CFAR detection in the overall scene, the correlated sea clutter is decorrelated via whitening and the detection threshold matching the shape parameters of sea clutter, number of accumulated pulses, and number of reference cells is applied via lookup tables. In this case, the experimental results show that the two non-coherent detectors can ensure CFAR in the overall scene.
2021, 43(3): 531-538.
doi: 10.11999/JEIT200449
Abstract:
Group target tracking is an efficient method to measure the states of airborne flocks. The first step of group target tracking is track initiation, including target clustering and track promotion. The state-of-the-art algorithms require mutual similarity between targets for clustering procedure, and track may be wrongly rejected due to the large residual of equivalent measurement. A modified Bayesian group track initiation algorithm based on algebraic graph theory is proposed. The clustering of measurement sets in surveillance volume is achieved by introducing the algebraic graph theory. The rejection of true track is avoided by modify the definition of classical Bayesian likelihood ratio. Results from actual field tests demonstrate the capability of clustering group targets precisely and promoting group tracks effectively.
Group target tracking is an efficient method to measure the states of airborne flocks. The first step of group target tracking is track initiation, including target clustering and track promotion. The state-of-the-art algorithms require mutual similarity between targets for clustering procedure, and track may be wrongly rejected due to the large residual of equivalent measurement. A modified Bayesian group track initiation algorithm based on algebraic graph theory is proposed. The clustering of measurement sets in surveillance volume is achieved by introducing the algebraic graph theory. The rejection of true track is avoided by modify the definition of classical Bayesian likelihood ratio. Results from actual field tests demonstrate the capability of clustering group targets precisely and promoting group tracks effectively.
2021, 43(3): 539-546.
doi: 10.11999/JEIT200636
Abstract:
In the scenario of multi-target tracking by a radar network system, a Radio Frequency (RF) stealth-based optimal RF resource allocation algorithm in radar network is proposed. Firstly, the Bayesian Cramer-Rao Lower Bound (BCRLB) of target tracking error is used as the target tracking performance index. Secondly, the optimization model is established which includes three optimization variables: radar node selection, dwell time and radiation power. In this model, the objective function is the weighted sum of the dwell time resources and radiation power resources of each radar, the constraint condition can be conclude that the BCRLB must be less than the given threshold and the system RF radiation resources must be between the upper and lower limits. Then, the two-step decomposition method is used to solve the above optimization model. The radar node selection is fixed first, then the interior point method is used to solve the simplified non-convex nonlinear optimization model, and then the Hungarian algorithm is used to determine the best radar node selection mode. The simulation results show that compared with uniform resource allocation algorithm, the proposed algorithm can effectively reduce the RF resource consumption of the radar network and improve the RF stealth performance of the system.
In the scenario of multi-target tracking by a radar network system, a Radio Frequency (RF) stealth-based optimal RF resource allocation algorithm in radar network is proposed. Firstly, the Bayesian Cramer-Rao Lower Bound (BCRLB) of target tracking error is used as the target tracking performance index. Secondly, the optimization model is established which includes three optimization variables: radar node selection, dwell time and radiation power. In this model, the objective function is the weighted sum of the dwell time resources and radiation power resources of each radar, the constraint condition can be conclude that the BCRLB must be less than the given threshold and the system RF radiation resources must be between the upper and lower limits. Then, the two-step decomposition method is used to solve the above optimization model. The radar node selection is fixed first, then the interior point method is used to solve the simplified non-convex nonlinear optimization model, and then the Hungarian algorithm is used to determine the best radar node selection mode. The simulation results show that compared with uniform resource allocation algorithm, the proposed algorithm can effectively reduce the RF resource consumption of the radar network and improve the RF stealth performance of the system.
2021, 43(3): 547-554.
doi: 10.11999/JEIT200595
Abstract:
Vortex ElectroMagnetic (EM) wave with Orbital Angular Momentum (OAM) is widely concerned in radar applications. With vortex EM wave, not only the linear Doppler shift of the target can be observed, but also the angular Doppler shift information can be obtained. Based on the angular Doppler effect, the vortex EM wave radar has the ability to detect the component perpendicular to the radial motion, and can extract the micro-motion features of the spinning target. Firstly, the parametric model of angular Doppler shift in Cartesian coordinate system is established, and the quantitative relationship among vortex EM wave radar, target motion parameters and angular Doppler shift is provided. Then, when the target rotational trajectory is perpendicular to the radar Line Of Sight (LOS), the angular Doppler shift is analyzed, and the micro-motion features of the spinning target are extracted. Finally, experimental results are given to demonstrate the effectiveness of the proposed method and the correctness of the theoretical analyses.
Vortex ElectroMagnetic (EM) wave with Orbital Angular Momentum (OAM) is widely concerned in radar applications. With vortex EM wave, not only the linear Doppler shift of the target can be observed, but also the angular Doppler shift information can be obtained. Based on the angular Doppler effect, the vortex EM wave radar has the ability to detect the component perpendicular to the radial motion, and can extract the micro-motion features of the spinning target. Firstly, the parametric model of angular Doppler shift in Cartesian coordinate system is established, and the quantitative relationship among vortex EM wave radar, target motion parameters and angular Doppler shift is provided. Then, when the target rotational trajectory is perpendicular to the radar Line Of Sight (LOS), the angular Doppler shift is analyzed, and the micro-motion features of the spinning target are extracted. Finally, experimental results are given to demonstrate the effectiveness of the proposed method and the correctness of the theoretical analyses.
2021, 43(3): 555-563.
doi: 10.11999/JEIT200605
Abstract:
For removing non-weather echoes from the data of dual-polarization weather radar, a clutter filtering method based on Spectral Polarimetric Parameters (SPP) is presented in this paper. Unlike the traditional time- or frequency-domain clutter suppression methods, this method retains the weather echoes and mitigates clutter according to their differences in the Range Doppler (RD) domain. Firstly, the SPPs are defined by exploiting the polarimetric features in the RD domain. With the help of the morphological method, the binary masks are generated. Then, with the object-oriented idea, the binary masks are labeled as weather-object masks or clutter-object ones. The spectral width is introduced as an additional parameter to select weather-object masks, all of which are added up to form the clutter suppression filter and hence the complete weather information can be remained. The X- and C-band polarized weather radar data are used to demonstrate the effectiveness of the proposed method. As compared with Moving Double spectral Linear Depolarization Ratio(MDsLDR) filter and the time-domain clutter suppression method using threshold factor, the proposed SPP clutter suppression filter is more effective in retaining relatively weak weather information. Moreover, the SPP filter has low computational complexity and can be applied to real time for dual-polarization weather radars working in either Simultaneous Transmission and Simultaneous Reception (STSR) mode or Alternate Transmission and Simultaneous Reception(ATSR) mode.
For removing non-weather echoes from the data of dual-polarization weather radar, a clutter filtering method based on Spectral Polarimetric Parameters (SPP) is presented in this paper. Unlike the traditional time- or frequency-domain clutter suppression methods, this method retains the weather echoes and mitigates clutter according to their differences in the Range Doppler (RD) domain. Firstly, the SPPs are defined by exploiting the polarimetric features in the RD domain. With the help of the morphological method, the binary masks are generated. Then, with the object-oriented idea, the binary masks are labeled as weather-object masks or clutter-object ones. The spectral width is introduced as an additional parameter to select weather-object masks, all of which are added up to form the clutter suppression filter and hence the complete weather information can be remained. The X- and C-band polarized weather radar data are used to demonstrate the effectiveness of the proposed method. As compared with Moving Double spectral Linear Depolarization Ratio(MDsLDR) filter and the time-domain clutter suppression method using threshold factor, the proposed SPP clutter suppression filter is more effective in retaining relatively weak weather information. Moreover, the SPP filter has low computational complexity and can be applied to real time for dual-polarization weather radars working in either Simultaneous Transmission and Simultaneous Reception (STSR) mode or Alternate Transmission and Simultaneous Reception(ATSR) mode.
2021, 43(3): 564-571.
doi: 10.11999/JEIT200075
Abstract:
Multi-ballistic targets have different translational parameters in midcouse. The former translational compensation methods for single target and group targets with the same translational parameters are no longer applicable. In order to solve this problem, a new method of multi-target translation parameters and micro-motion period estimation is proposed based on high-order ambiguity, delayed conjugate multiplication and time-frequency distribution processing. First, the second-order translational acceleration and micro-motion period are estimated by using the high-order ambiguity. Then, the first-order translational acceleration is estimated by the delayed conjugate multiplication of the echo after the first compensation. Finally, the residual translational velocity is estimated by the weighted accumulation of the time axis based on the difference of the time-frequency image of the echo after the second compensation. The simulation results show the effectiveness of the proposed algorithm for multi-target estimation of different translational parameters.
Multi-ballistic targets have different translational parameters in midcouse. The former translational compensation methods for single target and group targets with the same translational parameters are no longer applicable. In order to solve this problem, a new method of multi-target translation parameters and micro-motion period estimation is proposed based on high-order ambiguity, delayed conjugate multiplication and time-frequency distribution processing. First, the second-order translational acceleration and micro-motion period are estimated by using the high-order ambiguity. Then, the first-order translational acceleration is estimated by the delayed conjugate multiplication of the echo after the first compensation. Finally, the residual translational velocity is estimated by the weighted accumulation of the time axis based on the difference of the time-frequency image of the echo after the second compensation. The simulation results show the effectiveness of the proposed algorithm for multi-target estimation of different translational parameters.
2021, 43(3): 572-579.
doi: 10.11999/JEIT200572
Abstract:
To demonstrate the performance of beam-pattern and clutter distribution for airborne conformal radar due to the bending effect of carrier, the half-uniform circular array and fan-shaped array with the same radius are compared in the array synthesis characteristic, in the case of vertical side-looking. It is found that the edge elements have little contribution on the main-beam gain, alternatively, but enhances the side-lobe level so obvious that the combination property of the radar and aperture efficiency suffers degradation. Simultaneously, the inherent relation between the beam-pattern and the clutter spread is proved that the beam distortion resulting from the curved surface is the core factor on the clutter distribution inhomogeneity, through strict theoretical reduction. On the basis of the previous discussion, the additional structure of manufacture and installation, which would increase the physical dimension, is considered to enlarge the space between the adjacent radiators. In particular, the grating-lobe has appeared on the far-field pattern and made further efforts on the clutter distribution for Conformal Array Radar (CAR), in the condition of the reasonable element interval. One to be noted is that the grating-lobe clutter is folded to the main-lobe clutter region under the specific Pulse Repetition Frequency (PRF). To avoid the foldover of clutter, the PRF of CAR can be optimized and the related simulation is conducted to verify the effectiveness of the proposed method.
To demonstrate the performance of beam-pattern and clutter distribution for airborne conformal radar due to the bending effect of carrier, the half-uniform circular array and fan-shaped array with the same radius are compared in the array synthesis characteristic, in the case of vertical side-looking. It is found that the edge elements have little contribution on the main-beam gain, alternatively, but enhances the side-lobe level so obvious that the combination property of the radar and aperture efficiency suffers degradation. Simultaneously, the inherent relation between the beam-pattern and the clutter spread is proved that the beam distortion resulting from the curved surface is the core factor on the clutter distribution inhomogeneity, through strict theoretical reduction. On the basis of the previous discussion, the additional structure of manufacture and installation, which would increase the physical dimension, is considered to enlarge the space between the adjacent radiators. In particular, the grating-lobe has appeared on the far-field pattern and made further efforts on the clutter distribution for Conformal Array Radar (CAR), in the condition of the reasonable element interval. One to be noted is that the grating-lobe clutter is folded to the main-lobe clutter region under the specific Pulse Repetition Frequency (PRF). To avoid the foldover of clutter, the PRF of CAR can be optimized and the related simulation is conducted to verify the effectiveness of the proposed method.
2021, 43(3): 580-588.
doi: 10.11999/JEIT200655
Abstract:
Considering that the receiving data of Over-The-Horizon (OTH) radar is complicated, a signal analysis method based on various domains and maps is proposed, and the characteristics of sea-clutter, transient interference and radio frequency interference are discussed in this paper. By employing the time, the frequency, the range, the period and the Doppler domains, six kinds of two dimensional matrices or maps are constructed, which forms the Five-Domain-Six-Map (5D6M) concept for the radar receiving data. By using the 5D6M, the signal components of the receiving data and reveal their characteristics can be separated in various domains, including the spectrum and Doppler property of sea-clutter, the power and transient features of transient interference, and the narrowband and Doppler property of radio frequency interference. The characteristic analysis based on the 5D6M can help the radar signal analysis and processing. By taking the sea-clutter as an example, a novel method for detecting the sea-clutter Doppler based on the frequency-Doppler map is proposed.
Considering that the receiving data of Over-The-Horizon (OTH) radar is complicated, a signal analysis method based on various domains and maps is proposed, and the characteristics of sea-clutter, transient interference and radio frequency interference are discussed in this paper. By employing the time, the frequency, the range, the period and the Doppler domains, six kinds of two dimensional matrices or maps are constructed, which forms the Five-Domain-Six-Map (5D6M) concept for the radar receiving data. By using the 5D6M, the signal components of the receiving data and reveal their characteristics can be separated in various domains, including the spectrum and Doppler property of sea-clutter, the power and transient features of transient interference, and the narrowband and Doppler property of radio frequency interference. The characteristic analysis based on the 5D6M can help the radar signal analysis and processing. By taking the sea-clutter as an example, a novel method for detecting the sea-clutter Doppler based on the frequency-Doppler map is proposed.
2021, 43(3): 589-597.
doi: 10.11999/JEIT200127
Abstract:
In order to solve incomplete prior information of radar in non-cooperative electronic countermeasure environment, a novel recognition algorithm named ISNB (Improved Semi-supervised Naïve Bayes) based on the energy cumulant of Choi-Williams Distribution(CWD) is put forward. This algorithm extracts the energy cumulant of Choi-Williams distribution of radar signals as the recognition feature. The energy cumulant of CWD is attained by calculating the accumulations of each frequency sample value with the different time samples. Before this procedure, CWD is processed by base noise reduction. Considering disadvantages of traditional Semi-supervised Naïve Bayes(SNB) which comes from repeated errors in updating sample sets, it uses ISNB to construct classifier, and then completes the recognition of tested sample sets. ISNB selects those samples with high degree of confidence which comes from generated confidence. Theoretical analysis and simulation results show that the proposed method is about 3% higher than the traditional SNB algorithm. Under the same signal-to-noise ratio, this algorithm has higher classification recognition rate and better classification performance than the traditional principal component analysis plus support vector machine.
In order to solve incomplete prior information of radar in non-cooperative electronic countermeasure environment, a novel recognition algorithm named ISNB (Improved Semi-supervised Naïve Bayes) based on the energy cumulant of Choi-Williams Distribution(CWD) is put forward. This algorithm extracts the energy cumulant of Choi-Williams distribution of radar signals as the recognition feature. The energy cumulant of CWD is attained by calculating the accumulations of each frequency sample value with the different time samples. Before this procedure, CWD is processed by base noise reduction. Considering disadvantages of traditional Semi-supervised Naïve Bayes(SNB) which comes from repeated errors in updating sample sets, it uses ISNB to construct classifier, and then completes the recognition of tested sample sets. ISNB selects those samples with high degree of confidence which comes from generated confidence. Theoretical analysis and simulation results show that the proposed method is about 3% higher than the traditional SNB algorithm. Under the same signal-to-noise ratio, this algorithm has higher classification recognition rate and better classification performance than the traditional principal component analysis plus support vector machine.
2021, 43(3): 598-605.
doi: 10.11999/JEIT200210
Abstract:
In order to mitigate the location ambiguity of radar signal in Low Earth Orbit (LEO) dual-satellite Time Difference Of Arrival (TDOA) location system, the ambiguous characteristics of high PRF signal are analyzed, the process of ghost location generating by ambiguous TDOA is introduced, and a method for location ambiguity mitigation by Variation Trendline Matching of TDOA (VTMT) is proposed. If the condition that the residuals between the measured TDOA series and the theoretical TDOA series corresponding to the positioning results approximately obey normal distribution is satisfied, the location where the measured TDOA line is closest to the theoretical TDOA line can be selected as the real position of radar emitter with the Euclidean distance being used to measure the similarity between the two series. Test signal of calibration station validates the effectiveness of the proposed method. Numerical simulations demonstrate that the proposed method can increase the probability of unambiguous location significantly, reduce the requirements of TDOA measurement accuracy and observation time. The proposed method has good prospect for applications to TDOA location of radar signal in long base line location systems.
In order to mitigate the location ambiguity of radar signal in Low Earth Orbit (LEO) dual-satellite Time Difference Of Arrival (TDOA) location system, the ambiguous characteristics of high PRF signal are analyzed, the process of ghost location generating by ambiguous TDOA is introduced, and a method for location ambiguity mitigation by Variation Trendline Matching of TDOA (VTMT) is proposed. If the condition that the residuals between the measured TDOA series and the theoretical TDOA series corresponding to the positioning results approximately obey normal distribution is satisfied, the location where the measured TDOA line is closest to the theoretical TDOA line can be selected as the real position of radar emitter with the Euclidean distance being used to measure the similarity between the two series. Test signal of calibration station validates the effectiveness of the proposed method. Numerical simulations demonstrate that the proposed method can increase the probability of unambiguous location significantly, reduce the requirements of TDOA measurement accuracy and observation time. The proposed method has good prospect for applications to TDOA location of radar signal in long base line location systems.
2021, 43(3): 606-614.
doi: 10.11999/JEIT200685
Abstract:
In most of Synthetic Aperture Radar (SAR) target recognition methods, only the amplitude feature, i.e., intensity of pixels, is used to recognize targets. Nevertheless, due to the speckle noise, only using the amplitude feature will affect the recognition performance. For further improving the recognition performance, in this paper, a novel multi-level feature fusion target recognition method based on deep forest for SAR images is proposed. At First, in the feature extraction step, two kinds of features, i.e., the multi-level amplitude feature and the multi-level Dense Scale-Invariant Feature Transform (Dense-SIFT) feature are extracted. The amplitude feature describes intensity information and the Dense-SIFT feature describes structure information. Furthermore, for each feature, its corresponding multi-level features are extracted to represent target information from local to global. Then, for reflecting target information more comprehensive and sufficient, the multi-level amplitude feature and the multi-level Dense-SIFT feature are jointly utilized profiting from the idea of deep forest. On the one hand, the cascade structure can fusion multi-level amplitude feature and the multi-level Dense-SIFT feature steadily. On the other hand, the deep feature representation can be mined by layer-by-layer feature transformation. Finally, the fusion feature is used to recognize targets. Experiments on the moving and stationary target acquisition and recognition data show that the proposed method is an effective target recognition method, and the recognition performance is robust to the hyper-parameters.
In most of Synthetic Aperture Radar (SAR) target recognition methods, only the amplitude feature, i.e., intensity of pixels, is used to recognize targets. Nevertheless, due to the speckle noise, only using the amplitude feature will affect the recognition performance. For further improving the recognition performance, in this paper, a novel multi-level feature fusion target recognition method based on deep forest for SAR images is proposed. At First, in the feature extraction step, two kinds of features, i.e., the multi-level amplitude feature and the multi-level Dense Scale-Invariant Feature Transform (Dense-SIFT) feature are extracted. The amplitude feature describes intensity information and the Dense-SIFT feature describes structure information. Furthermore, for each feature, its corresponding multi-level features are extracted to represent target information from local to global. Then, for reflecting target information more comprehensive and sufficient, the multi-level amplitude feature and the multi-level Dense-SIFT feature are jointly utilized profiting from the idea of deep forest. On the one hand, the cascade structure can fusion multi-level amplitude feature and the multi-level Dense-SIFT feature steadily. On the other hand, the deep feature representation can be mined by layer-by-layer feature transformation. Finally, the fusion feature is used to recognize targets. Experiments on the moving and stationary target acquisition and recognition data show that the proposed method is an effective target recognition method, and the recognition performance is robust to the hyper-parameters.
2021, 43(3): 615-622.
doi: 10.11999/JEIT200630
Abstract:
To solve the problems of inter-frame registration difficult, unclear shadow characteristics of fast moving targets and high false alarm probability in traditional Video Synthetic Aperture Radar (ViSAR) moving target detection methods, a novel video SAR moving target detection method based on improved Faster Region-based Convolutional Neural Networks (Faster R-CNN) is proposed. Combining with the deep learning algorithm of Faster R-CNN, the new method applies the K-means clustering method to preprocess the length, width and aspect ratio of the anchor box. Besides, the Feature Pyramid Networks (FPN) network architecture is used to detect the ‘bright line’ feature of the video SAR moving targets. Compared with traditional methods, the proposed method has the advantages of simple implementation, high detection probability and low false alarm probability. Finally, the effectiveness of the new method is verified by the measured video SAR data obtained from the Mini-SAR system developed by our project team.
To solve the problems of inter-frame registration difficult, unclear shadow characteristics of fast moving targets and high false alarm probability in traditional Video Synthetic Aperture Radar (ViSAR) moving target detection methods, a novel video SAR moving target detection method based on improved Faster Region-based Convolutional Neural Networks (Faster R-CNN) is proposed. Combining with the deep learning algorithm of Faster R-CNN, the new method applies the K-means clustering method to preprocess the length, width and aspect ratio of the anchor box. Besides, the Feature Pyramid Networks (FPN) network architecture is used to detect the ‘bright line’ feature of the video SAR moving targets. Compared with traditional methods, the proposed method has the advantages of simple implementation, high detection probability and low false alarm probability. Finally, the effectiveness of the new method is verified by the measured video SAR data obtained from the Mini-SAR system developed by our project team.
2021, 43(3): 623-631.
doi: 10.11999/JEIT200292
Abstract:
Due to the echoes of the Inverse Synthetic Aperture Radar (ISAR) imagery are spatially sparse, the conventional convex optimization for the sparse image recovery involves tedious adjustment for the regularization parameter, which seriously limits the accuracy and the convenience of the image formation. In this paper, the unconstrained least absolute shrinkage and selection operator (Lasso) model is introduced for the\begin{document}$ {\ell _1}$\end{document} ![]()
![]()
Due to the echoes of the Inverse Synthetic Aperture Radar (ISAR) imagery are spatially sparse, the conventional convex optimization for the sparse image recovery involves tedious adjustment for the regularization parameter, which seriously limits the accuracy and the convenience of the image formation. In this paper, the unconstrained least absolute shrinkage and selection operator (Lasso) model is introduced for the
2021, 43(3): 632-639.
doi: 10.11999/JEIT200319
Abstract:
Microwave photonics radar generates cross-band signals with large bandwidth, providing a basis for precise electromagnetic characteristics description and accurate identification of targets. Meanwhile, a corresponding electromagnetic model parameter extraction method is urgently required in the case of large-bandwidth and wide-angle. For the same scene, the amount of return signals will increase considerably in cross-band condition comparing with narrow-bandwidth condition. Furthermore, the signals under cross-band condition may exhibit complex range-azimuth coupling. Under such a condition, it is of difficulty to estimate high dimensional physical parameters of scattering centers in the scene from return signals. To solve this problem, a multi-parameter estimation of cross-band SAR scattering centers method is proposed. The Polar Format Algorithm (PFA) and the attributed scattering center model are combined to construct a two-dimensional decoupled wavenumber domain model. With this scattering model, the estimation procedure is transformed into an optimization problem with multiple variables. This complex multi-variable optimization problem is divided into a set of single variable optimization problems by using the Coordinate Descent Algorithm (CDA). The separation effectively reduces dictionary dimensions and estimation complexity. Moreover, the Hooke-Jeeves algorithm is introduced to enhance estimation accuracy in each single variable optimization problem. Consequently, the proposed estimator for scattering parameters is not only efficient, but also accurate. The structure and location of each scattering center can be identified according to the parameter estimation results. Simulation results confirm the validity of the proposed method.
Microwave photonics radar generates cross-band signals with large bandwidth, providing a basis for precise electromagnetic characteristics description and accurate identification of targets. Meanwhile, a corresponding electromagnetic model parameter extraction method is urgently required in the case of large-bandwidth and wide-angle. For the same scene, the amount of return signals will increase considerably in cross-band condition comparing with narrow-bandwidth condition. Furthermore, the signals under cross-band condition may exhibit complex range-azimuth coupling. Under such a condition, it is of difficulty to estimate high dimensional physical parameters of scattering centers in the scene from return signals. To solve this problem, a multi-parameter estimation of cross-band SAR scattering centers method is proposed. The Polar Format Algorithm (PFA) and the attributed scattering center model are combined to construct a two-dimensional decoupled wavenumber domain model. With this scattering model, the estimation procedure is transformed into an optimization problem with multiple variables. This complex multi-variable optimization problem is divided into a set of single variable optimization problems by using the Coordinate Descent Algorithm (CDA). The separation effectively reduces dictionary dimensions and estimation complexity. Moreover, the Hooke-Jeeves algorithm is introduced to enhance estimation accuracy in each single variable optimization problem. Consequently, the proposed estimator for scattering parameters is not only efficient, but also accurate. The structure and location of each scattering center can be identified according to the parameter estimation results. Simulation results confirm the validity of the proposed method.
2021, 43(3): 640-649.
doi: 10.11999/JEIT200648
Abstract:
One of the main purposes for space surveillance is to supervise the movement status of non-cooperative space targets, which is also the prerequisite for further on-orbit operations. Because of the rotation of disabled satellites and space debris, it is necessary to accurately obtain the rotation vector, including the rotation speed and the direction of the rotation axis. This paper proposes a novel estimation method to obtain the rotation vector of space targets, which can be simultaneously used to form the Three-Dimensional (3D) image. Firstly, the three-dimensional position coordinates and the effective rotation vector are obtained by the Interferometric Inverse Synthetic Aperture Radar (InISAR) technology. Then, the total rotation velocity is estimated by the micro-Doppler feature extraction. Finally, the total rotation vector is acquired by combining the effective rotation velocity and the rotation velocity along the radar Line-Of-Sight (LOS). The effectiveness of the proposed method is demonstrated by simulation experiments. Performance analysis shows that the method can provide us with accurate results in both rotation vector estimation and three-dimensional imaging.
One of the main purposes for space surveillance is to supervise the movement status of non-cooperative space targets, which is also the prerequisite for further on-orbit operations. Because of the rotation of disabled satellites and space debris, it is necessary to accurately obtain the rotation vector, including the rotation speed and the direction of the rotation axis. This paper proposes a novel estimation method to obtain the rotation vector of space targets, which can be simultaneously used to form the Three-Dimensional (3D) image. Firstly, the three-dimensional position coordinates and the effective rotation vector are obtained by the Interferometric Inverse Synthetic Aperture Radar (InISAR) technology. Then, the total rotation velocity is estimated by the micro-Doppler feature extraction. Finally, the total rotation vector is acquired by combining the effective rotation velocity and the rotation velocity along the radar Line-Of-Sight (LOS). The effectiveness of the proposed method is demonstrated by simulation experiments. Performance analysis shows that the method can provide us with accurate results in both rotation vector estimation and three-dimensional imaging.
2021, 43(3): 650-656.
doi: 10.11999/JEIT200620
Abstract:
To address the problem that the aperture-dependence of ‘non-hop-go-hop’ time in the existing imaging algorithm for the multiple receivers Synthetic Aperture Sonar(SAS) is ignored and brings the delay error, a imaging algorithm based on the reference range history is proposed in the paper. Firstly, the shifting relationship between the reference receiver and the other receivers is used to derive the approximated range history of every receiver, which conquer the problem of ignoring aperture-dependence of ‘non-hop-go-hop’ time. Then, after the echo signal of each receiver is shift to the same beam centre range, the undersampling multiple receivers signal can be transformed into the single receiver signal by the azimuth reconstruction, which can be processed by the monostatic imaging algorithm to get imaging result. Finally, the validity of proposed algorithm is tested by the simulation experiments and real data.
To address the problem that the aperture-dependence of ‘non-hop-go-hop’ time in the existing imaging algorithm for the multiple receivers Synthetic Aperture Sonar(SAS) is ignored and brings the delay error, a imaging algorithm based on the reference range history is proposed in the paper. Firstly, the shifting relationship between the reference receiver and the other receivers is used to derive the approximated range history of every receiver, which conquer the problem of ignoring aperture-dependence of ‘non-hop-go-hop’ time. Then, after the echo signal of each receiver is shift to the same beam centre range, the undersampling multiple receivers signal can be transformed into the single receiver signal by the azimuth reconstruction, which can be processed by the monostatic imaging algorithm to get imaging result. Finally, the validity of proposed algorithm is tested by the simulation experiments and real data.
2021, 43(3): 657-664.
doi: 10.11999/JEIT200650
Abstract:
In the case of diving highly squinted Synthetic Aperture Radar (SAR), the existence of range-dependent squint angle, severe range-azimuth coupling, three-Dimensional (3-D) velocity and acceleration produces two-Dimensional (2-D) spatial-variant Range Cell Migration (RCM) and Doppler phases. To accommodate these issues, this paper constructs a 3-D equidistant sphere analytical model to precisely reveal the range-azimuth variant property of the echo. Based on the model, an azimuth-variant residual high-order RCM correction is proposed, and the Frequency Extended NonLinear Chirp Scaling (FENLCS) is rederived to equalize the azimuth-variant Doppler phases. These two methods integrated with SubAperture (SA) processing are adopted to address the aforementioned issues faced by diving highly squinted SAR. Theoretical analysis and simulation results validate that the proposed model is capable of describing the range-azimuth spatial-variance property of echo more precisely, and better imaging performance can be acquired by this algorithm.
In the case of diving highly squinted Synthetic Aperture Radar (SAR), the existence of range-dependent squint angle, severe range-azimuth coupling, three-Dimensional (3-D) velocity and acceleration produces two-Dimensional (2-D) spatial-variant Range Cell Migration (RCM) and Doppler phases. To accommodate these issues, this paper constructs a 3-D equidistant sphere analytical model to precisely reveal the range-azimuth variant property of the echo. Based on the model, an azimuth-variant residual high-order RCM correction is proposed, and the Frequency Extended NonLinear Chirp Scaling (FENLCS) is rederived to equalize the azimuth-variant Doppler phases. These two methods integrated with SubAperture (SA) processing are adopted to address the aforementioned issues faced by diving highly squinted SAR. Theoretical analysis and simulation results validate that the proposed model is capable of describing the range-azimuth spatial-variance property of echo more precisely, and better imaging performance can be acquired by this algorithm.
2021, 43(3): 665-673.
doi: 10.11999/JEIT200285
Abstract:
Generally, the refractive index of atmosphere is simply 1 by default in Synthetic Aperture Radar (SAR) imaging, that is, the ElectroMagnetic (EM) wave velocity is equal to the speed of light in free-space and the atmospheric absorption is ignored. However, the actual absorption may weaken the incident power and variations in the speed of EM waves may cause phase error, thus affecting image reconstruction. In this paper, the influence of wave velocity fluctuation and atmospheric absorption in SAR imagery is analyzed quantitatively. It is theoretically deduced that the atmospheric absorption will lead to amplitude error, which is shown as strength error of the scatterer in the reconstructed image; EM velocity fluctuation will lead to phase error, which is shown as positioning error of the scatterer in the reconstructed image. The correctness of error analysis is verified by simulation experiments. The work in this paper completes further the SAR imaging error analysis, which is beneficial to SAR image interpretation.
Generally, the refractive index of atmosphere is simply 1 by default in Synthetic Aperture Radar (SAR) imaging, that is, the ElectroMagnetic (EM) wave velocity is equal to the speed of light in free-space and the atmospheric absorption is ignored. However, the actual absorption may weaken the incident power and variations in the speed of EM waves may cause phase error, thus affecting image reconstruction. In this paper, the influence of wave velocity fluctuation and atmospheric absorption in SAR imagery is analyzed quantitatively. It is theoretically deduced that the atmospheric absorption will lead to amplitude error, which is shown as strength error of the scatterer in the reconstructed image; EM velocity fluctuation will lead to phase error, which is shown as positioning error of the scatterer in the reconstructed image. The correctness of error analysis is verified by simulation experiments. The work in this paper completes further the SAR imaging error analysis, which is beneficial to SAR image interpretation.
2021, 43(3): 674-682.
doi: 10.11999/JEIT200338
Abstract:
The classical sparse recovery of Inverse Synthetic Aperture Radar (ISAR) imagery obtains the ISAR image by solving the constrained problem of\begin{document}${\ell _{1}}$\end{document} ![]()
![]()
\begin{document}${\ell _{1}}/{\ell _{\rm{F}}}$\end{document} ![]()
![]()
\begin{document}${\ell _{1}}/{\ell _{\rm{F}}}$\end{document} ![]()
![]()
\begin{document}${\ell _{1}}/{\ell _{\rm{F}}}$\end{document} ![]()
![]()
The classical sparse recovery of Inverse Synthetic Aperture Radar (ISAR) imagery obtains the ISAR image by solving the constrained problem of
2021, 43(3): 683-691.
doi: 10.11999/JEIT200568
Abstract:
Considering the problem of low classification accuracy caused by large intra-class differences and high inter-class similarity in remote sensing image scene classification, a discriminative feature representation method based on dual attention mechanism is proposed. Due to the difference in the importance of the features contained in different channels and the significance of different local regions, the channel-wise and spatial-wise attention module are designed, based on the high-level features extracted by the Convolutional Neural Networks. Relying on the ability to extract contextual information, the Recurrent Neural Network is adopted to learn and output the importance weights of different channels and different local regions, paying more attention to the salient features and salient regions, while ignoring non-salience features and regions, to enhance the discriminative ability of feature representation. The proposed dual attention module can be connected to the last convolutional layer of any convolutional neural network, and the network structure can be trained end-to-end. Comparative experiments are conducted on the two public data sets AID and NWPU45. Compared with the existing methods, the classification accuracy has been significantly improved, and the effectiveness of the proposed method can be verified.
Considering the problem of low classification accuracy caused by large intra-class differences and high inter-class similarity in remote sensing image scene classification, a discriminative feature representation method based on dual attention mechanism is proposed. Due to the difference in the importance of the features contained in different channels and the significance of different local regions, the channel-wise and spatial-wise attention module are designed, based on the high-level features extracted by the Convolutional Neural Networks. Relying on the ability to extract contextual information, the Recurrent Neural Network is adopted to learn and output the importance weights of different channels and different local regions, paying more attention to the salient features and salient regions, while ignoring non-salience features and regions, to enhance the discriminative ability of feature representation. The proposed dual attention module can be connected to the last convolutional layer of any convolutional neural network, and the network structure can be trained end-to-end. Comparative experiments are conducted on the two public data sets AID and NWPU45. Compared with the existing methods, the classification accuracy has been significantly improved, and the effectiveness of the proposed method can be verified.
2021, 43(3): 692-699.
doi: 10.11999/JEIT200416
Abstract:
Winter wheat is one of the most important food crops in China. Monitoring the soil moisture over winter wheat covered surface can help to solve the problem of poor harvest of winter wheat and waste of agricultural water due to soil water supply. In order to reduce the influence of winter wheat on radar backscattering coefficient in the process of microwave remote sensing retrieval of soil moisture covered by winter wheat, based on the Synthetic Aperture Radar (SAR) data carried by Sentinel-1 and the MultiSpectral Imager (MSI) data carried by Sentinel-2, combined with the water cloud model, the collaborative inversion of soil moisture over winter wheat mulching surface is carried out. Firstly, based on the MSI data from Sentinel-2, a new vegetation index called Fusion Vegetation Index (FVI) is defined for inversion of winter wheat moisture. Secondly, a semi-empirical soil moisture inversion model based on active and passive remote sensing data is developed to correct the influence of winter wheat on radar backscatter coefficient. Finally, by taking a winter wheat field in Henan Province as the study area, the comparative experiments of soil moisture inversion are carried out under six combinations, which are composed of two vegetation indexes, Normalized Difference Water Index (NDWI) and FVI respectively, and three types of polarization data, VV, VH and VV/VH respectively. Through the experimental results, FVI shows a better performance than NDWI in reducing the influence of winter wheat on radar backscatter coefficient. Meanwhile, among the six inversion combinations, the one of FVI and VV/VH achieves the optimal inversion precision, with a determination coefficient of 0.7642, a Root Mean Square Error of 0.0209 cm3/cm3, and a Mean Absolute Error of 0.0174 cm3/cm3, demonstrating the application potential of the soil inversion model developed in this paper.
Winter wheat is one of the most important food crops in China. Monitoring the soil moisture over winter wheat covered surface can help to solve the problem of poor harvest of winter wheat and waste of agricultural water due to soil water supply. In order to reduce the influence of winter wheat on radar backscattering coefficient in the process of microwave remote sensing retrieval of soil moisture covered by winter wheat, based on the Synthetic Aperture Radar (SAR) data carried by Sentinel-1 and the MultiSpectral Imager (MSI) data carried by Sentinel-2, combined with the water cloud model, the collaborative inversion of soil moisture over winter wheat mulching surface is carried out. Firstly, based on the MSI data from Sentinel-2, a new vegetation index called Fusion Vegetation Index (FVI) is defined for inversion of winter wheat moisture. Secondly, a semi-empirical soil moisture inversion model based on active and passive remote sensing data is developed to correct the influence of winter wheat on radar backscatter coefficient. Finally, by taking a winter wheat field in Henan Province as the study area, the comparative experiments of soil moisture inversion are carried out under six combinations, which are composed of two vegetation indexes, Normalized Difference Water Index (NDWI) and FVI respectively, and three types of polarization data, VV, VH and VV/VH respectively. Through the experimental results, FVI shows a better performance than NDWI in reducing the influence of winter wheat on radar backscatter coefficient. Meanwhile, among the six inversion combinations, the one of FVI and VV/VH achieves the optimal inversion precision, with a determination coefficient of 0.7642, a Root Mean Square Error of 0.0209 cm3/cm3, and a Mean Absolute Error of 0.0174 cm3/cm3, demonstrating the application potential of the soil inversion model developed in this paper.
2021, 43(3): 700-707.
doi: 10.11999/JEIT200366
Abstract:
Water segmentation of Synthetic Aperture Radar (SAR) is of great significance in land hydrological monitoring, such as lakes and rivers. Water segmentation accuracy is influenced by the blurring of the boundary between land and water region because of the insufficient resolution of SAR image. Sentinel-1A SAR image is used to study the Duoqing Co in the Tibetan Plateau of China. This paper integrates the enhanced deep residual block, channel attention mechanism and sub-pixel convolution, an enhanced channel attention deep residual network is proposed based on sub-pixel to reconstruct the filtered SAR image, extract the water contour and analyze the accuracy. By comparing the reconstruction results of different super-resolution algorithms and the accuracy of water contour extraction, this algorithm, with great robustness, is obviously better than the traditional method in both reconstruction effect and extraction accuracy.
Water segmentation of Synthetic Aperture Radar (SAR) is of great significance in land hydrological monitoring, such as lakes and rivers. Water segmentation accuracy is influenced by the blurring of the boundary between land and water region because of the insufficient resolution of SAR image. Sentinel-1A SAR image is used to study the Duoqing Co in the Tibetan Plateau of China. This paper integrates the enhanced deep residual block, channel attention mechanism and sub-pixel convolution, an enhanced channel attention deep residual network is proposed based on sub-pixel to reconstruct the filtered SAR image, extract the water contour and analyze the accuracy. By comparing the reconstruction results of different super-resolution algorithms and the accuracy of water contour extraction, this algorithm, with great robustness, is obviously better than the traditional method in both reconstruction effect and extraction accuracy.
2021, 43(3): 708-716.
doi: 10.11999/JEIT200656
Abstract:
Off-grid Direction Of Arrival (DOA) estimation aims to handle the mismatch between the actual DOA and the presumed grid points. For DOAs of closely spaced signals, sparse grid points leads to degradation of accuracy and resolution, although dense grid points can improve the estimation accuracy, it significantly increases the computational burden. To solve this problem, this paper proposes a Sparse Bayesian Learning (SBL) based algorithm for DOA estimation of closely spaced signals, which consists of three steps. Firstly, a novel fixed point iterative method for signal of Laplace priori is derived to pre-estimate the hyper-parameters by maximizing the array’s marginal likelihood function, which results in faster convergence speed compared to other classical SBL algorithms. Secondly, a new grid interpolation method is implemented to optimize a set of grid points, and signal power and noise variance are estimated again to resolve closely spaced DOAs. Finally, an expression of maximum likelihood function with respect to angle is derived to improve the search of the off-grid DOA. Simulation results show that the proposed algorithm has higher accuracy and resolution for closely spaced DOAs with higher computational efficiency compared with other classical algorithms based on SBL.
Off-grid Direction Of Arrival (DOA) estimation aims to handle the mismatch between the actual DOA and the presumed grid points. For DOAs of closely spaced signals, sparse grid points leads to degradation of accuracy and resolution, although dense grid points can improve the estimation accuracy, it significantly increases the computational burden. To solve this problem, this paper proposes a Sparse Bayesian Learning (SBL) based algorithm for DOA estimation of closely spaced signals, which consists of three steps. Firstly, a novel fixed point iterative method for signal of Laplace priori is derived to pre-estimate the hyper-parameters by maximizing the array’s marginal likelihood function, which results in faster convergence speed compared to other classical SBL algorithms. Secondly, a new grid interpolation method is implemented to optimize a set of grid points, and signal power and noise variance are estimated again to resolve closely spaced DOAs. Finally, an expression of maximum likelihood function with respect to angle is derived to improve the search of the off-grid DOA. Simulation results show that the proposed algorithm has higher accuracy and resolution for closely spaced DOAs with higher computational efficiency compared with other classical algorithms based on SBL.
2021, 43(3): 717-726.
doi: 10.11999/JEIT200505
Abstract:
To deal with the problem that the Degree Of Freedom(DOF) of uniform linear array is limited by the number of elements, a new type of coprime array is proposed called Displaced Coprime Array(DCA).It takes use of the conjugate augmented matrix which is formed by the time and space information of the received signal to obtain the equivalent difference and sum co-array and to estimate the Direction Of Arrival(DOA). DCA places the generalized coprime array at a certain distance from the single array element at the coordinate origin so that the elements of the sum co-array and the difference co-array are complemented. As a result, the use of DOF provided by the sum co-array can be maximized. In this paper, the closed-form expressions of the element positions and the placement distance of DCA are given. Then, the performance of the sum co-array and the difference co-array including the continuous elements and the hole positions is theoretically analyzed, the relationship between the two is given and high DOF of DCA is presented. Multiple simulations verify the effectivity of DOA estimation using DCA.
To deal with the problem that the Degree Of Freedom(DOF) of uniform linear array is limited by the number of elements, a new type of coprime array is proposed called Displaced Coprime Array(DCA).It takes use of the conjugate augmented matrix which is formed by the time and space information of the received signal to obtain the equivalent difference and sum co-array and to estimate the Direction Of Arrival(DOA). DCA places the generalized coprime array at a certain distance from the single array element at the coordinate origin so that the elements of the sum co-array and the difference co-array are complemented. As a result, the use of DOF provided by the sum co-array can be maximized. In this paper, the closed-form expressions of the element positions and the placement distance of DCA are given. Then, the performance of the sum co-array and the difference co-array including the continuous elements and the hole positions is theoretically analyzed, the relationship between the two is given and high DOF of DCA is presented. Multiple simulations verify the effectivity of DOA estimation using DCA.
2021, 43(3): 727-734.
doi: 10.11999/JEIT200637
Abstract:
As for long-range target Direction of Arrival (DoA) estimation with benthonic long linear arrays under near-field acoustic interference and space non-stationary noise, a novel joint DoA estimation of subarrays algorithm is proposed, which is based on near-field nulling weight. The long linear array is divided into several high overlap subarrays. The near-field nulling weight is calculated for subarrays to eliminate the effect of the near-field interference on target detection. The spatial frequency variance weighted joint DoA method is presented based on the little discrepancy of the subarrays’ DoA estimates to repress the space nonuniform noise and detect the target in a long distance. The maximum value of spatial spectrum is random in the frequencies without the target because of the space non-stationary noise. The simulation results show that compared with long-linear-array conventional beamforming, long-linear-array conventional beamforming based on near-field nulling weight and long-linear-array multiple signal classification based on near-field nulling weight method, this proposed technique can effectively reduce the background level of spatial spectrum (over 60 dB), improve signal to noise ratio (above 15 dB) and has high spatial resolution, with better engineering application value.
As for long-range target Direction of Arrival (DoA) estimation with benthonic long linear arrays under near-field acoustic interference and space non-stationary noise, a novel joint DoA estimation of subarrays algorithm is proposed, which is based on near-field nulling weight. The long linear array is divided into several high overlap subarrays. The near-field nulling weight is calculated for subarrays to eliminate the effect of the near-field interference on target detection. The spatial frequency variance weighted joint DoA method is presented based on the little discrepancy of the subarrays’ DoA estimates to repress the space nonuniform noise and detect the target in a long distance. The maximum value of spatial spectrum is random in the frequencies without the target because of the space non-stationary noise. The simulation results show that compared with long-linear-array conventional beamforming, long-linear-array conventional beamforming based on near-field nulling weight and long-linear-array multiple signal classification based on near-field nulling weight method, this proposed technique can effectively reduce the background level of spatial spectrum (over 60 dB), improve signal to noise ratio (above 15 dB) and has high spatial resolution, with better engineering application value.
An Algebraic Solution for Single-Observer Passive Coherent Location Using DOA-TDOA-FDOA Measurements
2021, 43(3): 735-744.
doi: 10.11999/JEIT200470
Abstract:
To achieve the target localization using single-observer receiving multiple external illuminators, an algebraic solution based on two-step Weighted Least Squares (2WLS) is proposed to find the target position and velocity from Direction Of Arrival (DOA), Time Difference Of Arrival (TDOA), and Frequency Difference Of Arrival (FDOA) measurements. In the first WLS step, the DOA, TDOA, and FDOA measurements are pseudo-linearized by introducing additional parameters and a WLS minimization is used to obtain an rough estimate of target position and velocity; Then in the second WLS step, the relationship between the additional parameters and the target location parameters is utilized to form another set of linear equations, from which the final accurate estimate of target position and velocity are obtained by using WLS minimization again. The Cramer-Row Lower Bound (CRLB) for DOA-TDOA-FDOA-based target position and velocity estimation are derived. Theoretical accuracy analysis and simulation results indicate that the proposed solution can achieve the CRLB at sufficiently small measurement noise levels.
To achieve the target localization using single-observer receiving multiple external illuminators, an algebraic solution based on two-step Weighted Least Squares (2WLS) is proposed to find the target position and velocity from Direction Of Arrival (DOA), Time Difference Of Arrival (TDOA), and Frequency Difference Of Arrival (FDOA) measurements. In the first WLS step, the DOA, TDOA, and FDOA measurements are pseudo-linearized by introducing additional parameters and a WLS minimization is used to obtain an rough estimate of target position and velocity; Then in the second WLS step, the relationship between the additional parameters and the target location parameters is utilized to form another set of linear equations, from which the final accurate estimate of target position and velocity are obtained by using WLS minimization again. The Cramer-Row Lower Bound (CRLB) for DOA-TDOA-FDOA-based target position and velocity estimation are derived. Theoretical accuracy analysis and simulation results indicate that the proposed solution can achieve the CRLB at sufficiently small measurement noise levels.
2021, 43(3): 745-751.
doi: 10.11999/JEIT200541
Abstract:
In order to realize the fast direction estimation of underwater targets under the conditions of less snapshot and low SNR, a sparse decomposition model of vector hydrophone array direction estimation is established. The real value conversion technique is used to convert the complex direction matrix into the real number field, so as to reconstruct the sparse signal matrix using the SL0 algorithm to obtain the orientation estimation result. The SL0 algorithm is improved, the Compound Inverse Proportional Function (CIPF) with better convergence is used as a smoothing function, and a weighted method is proposed which can promote sparsity, the weighted method is used to correct the problem that the norm as the initial iteration value deviates far from the sparse solution to increase the speed of azimuth estimation. The simulation verifies that the proposed algorithm can achieve better performance than the traditional subspace algorithm under the conditions of low snapshot and low SNR, and improve the speed of bearing estimation while ensuring performance.
In order to realize the fast direction estimation of underwater targets under the conditions of less snapshot and low SNR, a sparse decomposition model of vector hydrophone array direction estimation is established. The real value conversion technique is used to convert the complex direction matrix into the real number field, so as to reconstruct the sparse signal matrix using the SL0 algorithm to obtain the orientation estimation result. The SL0 algorithm is improved, the Compound Inverse Proportional Function (CIPF) with better convergence is used as a smoothing function, and a weighted method is proposed which can promote sparsity, the weighted method is used to correct the problem that the norm as the initial iteration value deviates far from the sparse solution to increase the speed of azimuth estimation. The simulation verifies that the proposed algorithm can achieve better performance than the traditional subspace algorithm under the conditions of low snapshot and low SNR, and improve the speed of bearing estimation while ensuring performance.
2021, 43(3): 752-757.
doi: 10.11999/JEIT200582
Abstract:
In mobile OFDM underwater acoustic communication systems, the compressed sensing-based sparse channel estimation methods suffer from high computational complexity, which is not suitable for real-time communication. To solve this problem, this paper proposes a Generalized Path Identification (GPI) algorithm for estimating uniform Doppler distorted channel. This scheme first constructs equivalent transmitted symbols using Doppler spread matrices, and thus the channel is converted into an equivalent linear time-invariant one. Then the GPI algorithm is utilized to estimate the channel parameters. Furthermore, the GPI algorithm is extended to Turbo receivers to iteratively improve the channel estimation accuracy. Simulation results show that the performance of the proposed method is better than that of the conventional path identification algorithm, and is close to the Orthogonal Matching Pursuit (OMP) algorithm. Its computational complexity, however, is much lower than OMP algorithm.
In mobile OFDM underwater acoustic communication systems, the compressed sensing-based sparse channel estimation methods suffer from high computational complexity, which is not suitable for real-time communication. To solve this problem, this paper proposes a Generalized Path Identification (GPI) algorithm for estimating uniform Doppler distorted channel. This scheme first constructs equivalent transmitted symbols using Doppler spread matrices, and thus the channel is converted into an equivalent linear time-invariant one. Then the GPI algorithm is utilized to estimate the channel parameters. Furthermore, the GPI algorithm is extended to Turbo receivers to iteratively improve the channel estimation accuracy. Simulation results show that the performance of the proposed method is better than that of the conventional path identification algorithm, and is close to the Orthogonal Matching Pursuit (OMP) algorithm. Its computational complexity, however, is much lower than OMP algorithm.
2021, 43(3): 758-766.
doi: 10.11999/JEIT200611
Abstract:
Single-Carrier Frequency-Domain Equalization (SC-FDE) is one of the key techniques to achieve high data-rate transmissions in underwater acoustic communications. However, the periodically inserted prefixes or suffixes, in forms of cyclic-prefixes or guard intervals in transmission sequence, reduce the bandwidth efficiency and data rate. A SC-FDE technique without prefix or suffix is proposed based on the Time Reversal (TR) for underwater acoustic communications. At the transmitter side, the single-carrier signal without prefix or suffix is transmitted. At the receiver side, the suffix is reconstructed in each block before canceling the intersymbol interference with FDE. The suffix reconstruction, which is the core of receiver, contains three main procedures: the TR processing, interference cancelation, and suffix reconstruction. First, the TR processing converts receptions from an array of receiving elements into a single composite signal. The equivalent impulse response (namely the q-function) has a stable peak, which avoids the noise amplification. Next, the interference cancelation is conducted to remove the InterBlock Interference (IBI) from the previous block and the ACausal Interference (ACI) from the current block. Afterward, suffix reconstruction is achieved by calculating the multipath arrivals of each symbol, which are obtained from the causal part of the q-function. Experimental results validate the effectiveness of this technique and its superiority over the SC-FDE technique without prefix or suffix in the current literature.
Single-Carrier Frequency-Domain Equalization (SC-FDE) is one of the key techniques to achieve high data-rate transmissions in underwater acoustic communications. However, the periodically inserted prefixes or suffixes, in forms of cyclic-prefixes or guard intervals in transmission sequence, reduce the bandwidth efficiency and data rate. A SC-FDE technique without prefix or suffix is proposed based on the Time Reversal (TR) for underwater acoustic communications. At the transmitter side, the single-carrier signal without prefix or suffix is transmitted. At the receiver side, the suffix is reconstructed in each block before canceling the intersymbol interference with FDE. The suffix reconstruction, which is the core of receiver, contains three main procedures: the TR processing, interference cancelation, and suffix reconstruction. First, the TR processing converts receptions from an array of receiving elements into a single composite signal. The equivalent impulse response (namely the q-function) has a stable peak, which avoids the noise amplification. Next, the interference cancelation is conducted to remove the InterBlock Interference (IBI) from the previous block and the ACausal Interference (ACI) from the current block. Afterward, suffix reconstruction is achieved by calculating the multipath arrivals of each symbol, which are obtained from the causal part of the q-function. Experimental results validate the effectiveness of this technique and its superiority over the SC-FDE technique without prefix or suffix in the current literature.
2021, 43(3): 767-772.
doi: 10.11999/JEIT200629
Abstract:
Sallow water acoustic channel is severely affected by time-space variation, which destroys the robustness of underwater acoustic communication system. By introducing manifold learning in the analysis of high dimensional underwater environment and channel equalization processing, a novel underwater acoustic communication algorithm is presented. By establishing the mapping between environment parameter space and signal space, several physical restrictions can be posed on non-linear manifold learning algorithm. Moreover, the sparse property can reduce the dimension of underwater acoustic channel in order to exclude high dimensional non-linear noise from channel time-space variation. Both sound field analysis and shallow water experimental data verify the validity and the robustness of the proposed algorithm.
Sallow water acoustic channel is severely affected by time-space variation, which destroys the robustness of underwater acoustic communication system. By introducing manifold learning in the analysis of high dimensional underwater environment and channel equalization processing, a novel underwater acoustic communication algorithm is presented. By establishing the mapping between environment parameter space and signal space, several physical restrictions can be posed on non-linear manifold learning algorithm. Moreover, the sparse property can reduce the dimension of underwater acoustic channel in order to exclude high dimensional non-linear noise from channel time-space variation. Both sound field analysis and shallow water experimental data verify the validity and the robustness of the proposed algorithm.
2021, 43(3): 773-780.
doi: 10.11999/JEIT200540
Abstract:
In view of the problem that the performance of the Adaptive Line-spectrum Enhancement (ALE) algorithm drops sharply under low signal-to-noise ratio, a two-level ALE algorithm is proposed. Based on the ordinary ALE algorithm, the Fourier transform of the output signal and the error signal is used as the input of the second-level ALE to enhance further the line-spectrum. The noise suppression gate is used to perform noise suppression preprocessing on the signal, and then combining with the proposed two-level ALE algorithm, an algorithm of two-level ALE based on noise suppression gate is proposed to improve further the signal-to-noise ratio and enhance the line-spectrum. Simulation and sea trial data show that the performance of the algorithm is greatly improved compared with the ordinary ALE algorithm.
In view of the problem that the performance of the Adaptive Line-spectrum Enhancement (ALE) algorithm drops sharply under low signal-to-noise ratio, a two-level ALE algorithm is proposed. Based on the ordinary ALE algorithm, the Fourier transform of the output signal and the error signal is used as the input of the second-level ALE to enhance further the line-spectrum. The noise suppression gate is used to perform noise suppression preprocessing on the signal, and then combining with the proposed two-level ALE algorithm, an algorithm of two-level ALE based on noise suppression gate is proposed to improve further the signal-to-noise ratio and enhance the line-spectrum. Simulation and sea trial data show that the performance of the algorithm is greatly improved compared with the ordinary ALE algorithm.
2021, 43(3): 781-787.
doi: 10.11999/JEIT200383
Abstract:
The silent location algorithm for underwater sensor network is a time-synchronization free algorithm that can serve multiple users. No sound is needed of the target node in the whole process which make the algorithm secluded and extended. In this paper, a silent location algorithm for underwater sensor network based on ray-tracing technology is proposed. The problem of location blind area in the existing methods is solved by introducing the Gauss-Newton method into the algorithm proposed. In view of the uneven distribution of sound velocity in water, the ray-tracing technology is integrated into the iterative process to correct the positioning error caused by the bending of the sound ray. At the same time, considering the situation that the beacon node array may not be good in the practical application, an improved Tikhonov regularization method is adopted to control the regularization parameters according to the feedback of the iterative effect, which can compensate the influence of the singular Jacobian matrix on the objective function. The effectiveness of the proposed algorithm in this paper is verified by simulation analysis.
The silent location algorithm for underwater sensor network is a time-synchronization free algorithm that can serve multiple users. No sound is needed of the target node in the whole process which make the algorithm secluded and extended. In this paper, a silent location algorithm for underwater sensor network based on ray-tracing technology is proposed. The problem of location blind area in the existing methods is solved by introducing the Gauss-Newton method into the algorithm proposed. In view of the uneven distribution of sound velocity in water, the ray-tracing technology is integrated into the iterative process to correct the positioning error caused by the bending of the sound ray. At the same time, considering the situation that the beacon node array may not be good in the practical application, an improved Tikhonov regularization method is adopted to control the regularization parameters according to the feedback of the iterative effect, which can compensate the influence of the singular Jacobian matrix on the objective function. The effectiveness of the proposed algorithm in this paper is verified by simulation analysis.
2021, 43(3): 788-795.
doi: 10.11999/JEIT200704
Abstract:
To better understand and utilize the acoustic field in shallow sea, a theoretical method that can give the full-wave solution is proposed, then the complex integral expression of the acoustic field is given. The complex integral fraction is solved in the complex plane, and the components of the acoustic field in shallow sea are obtained. The high-order staggered grid finite difference method is also used to numerically simulate the acoustic field in shallow sea, showing the wave field structure and spatial energy distribution. Results show that the acoustic field in shallow sea includes discrete spectrum and continuous spectrum; The discrete spectrum includes normal waves and Scholte wave, and the continuous wave includes lateral waves; The amplitudes of normal waves and Scholte wave are inversely proportional to the root of horizontal propagation distance, and the amplitude of lateral wave is inversely proportional to the power of horizontal propagation distance; The shallower the sea water, the lower the frequency and the greater the depth of the sound source, the less energy in the sea water will be. The energy radiated by the acoustic source is mainly propagated in the form of Scholte wave, and the energy is mostly concentrated at the seabed interface.
To better understand and utilize the acoustic field in shallow sea, a theoretical method that can give the full-wave solution is proposed, then the complex integral expression of the acoustic field is given. The complex integral fraction is solved in the complex plane, and the components of the acoustic field in shallow sea are obtained. The high-order staggered grid finite difference method is also used to numerically simulate the acoustic field in shallow sea, showing the wave field structure and spatial energy distribution. Results show that the acoustic field in shallow sea includes discrete spectrum and continuous spectrum; The discrete spectrum includes normal waves and Scholte wave, and the continuous wave includes lateral waves; The amplitudes of normal waves and Scholte wave are inversely proportional to the root of horizontal propagation distance, and the amplitude of lateral wave is inversely proportional to the power of horizontal propagation distance; The shallower the sea water, the lower the frequency and the greater the depth of the sound source, the less energy in the sea water will be. The energy radiated by the acoustic source is mainly propagated in the form of Scholte wave, and the energy is mostly concentrated at the seabed interface.
2021, 43(3): 796-802.
doi: 10.11999/JEIT200744
Abstract:
Imaging method based on motion compensation for forward-looking scanning sonar is proposed. By using backward mapping, sector imaging region in the Cartesian coordinate system is converted into polar coordinates. Then two-dimension interpolation is employed in polar coordinates, and the amplitude is extracted and converted into image format. The vehicle motion model is established and analyzed, and vehicle motion is transformed into image domain motion. Real-time motion compensation for forward-looking sonar image is made eliminate radial errors. The forward-looking sonar data of the “Deep Sea Warrior” is analyzed. The maximum radial error of the single image is approximately 19% of the range. The simulation and experimental results show that this algorithm compensates the radial error of the image caused by the motion, which can represent the target information accurately.
Imaging method based on motion compensation for forward-looking scanning sonar is proposed. By using backward mapping, sector imaging region in the Cartesian coordinate system is converted into polar coordinates. Then two-dimension interpolation is employed in polar coordinates, and the amplitude is extracted and converted into image format. The vehicle motion model is established and analyzed, and vehicle motion is transformed into image domain motion. Real-time motion compensation for forward-looking sonar image is made eliminate radial errors. The forward-looking sonar data of the “Deep Sea Warrior” is analyzed. The maximum radial error of the single image is approximately 19% of the range. The simulation and experimental results show that this algorithm compensates the radial error of the image caused by the motion, which can represent the target information accurately.
2021, 43(3): 803-808.
doi: 10.11999/JEIT200584
Abstract:
Time Delay Estimation(TDE) is an important research subject in underwater acoustic field. Passive direction-finding and passive location based on TDE are important brunches in passive location of underwater acoustic target. At present, the common TDE methods include Normalized Cross Correlation(NCC) method, Generalized Cross Correlation(GCC) method and Least Mean Square(LMS) method. Different from the above methods, a new method of TDE is designed by using Cross-power Spectrum Phase(CSP) in this paper. This method performs Fourier transform of CSP and uses the transform domain to estimate the CSP slope to get the time delay. This method can realize multi-target resolution and eliminate the ambiguity and background fluctuation. Simulation results show that this proposed method is better than NCC and GCC.
Time Delay Estimation(TDE) is an important research subject in underwater acoustic field. Passive direction-finding and passive location based on TDE are important brunches in passive location of underwater acoustic target. At present, the common TDE methods include Normalized Cross Correlation(NCC) method, Generalized Cross Correlation(GCC) method and Least Mean Square(LMS) method. Different from the above methods, a new method of TDE is designed by using Cross-power Spectrum Phase(CSP) in this paper. This method performs Fourier transform of CSP and uses the transform domain to estimate the CSP slope to get the time delay. This method can realize multi-target resolution and eliminate the ambiguity and background fluctuation. Simulation results show that this proposed method is better than NCC and GCC.
2021, 43(3): 809-816.
doi: 10.11999/JEIT200634
Abstract:
Marine environment noise and reverberation interference are serious, and the poor target separability is the bottleneck problem in active sonar target classification and recognition. In order to solve this problem, based on the echo signal model of active sonar target and the principle of FRactional Fourier Transform (FRFT), this paper deduces the multi-order FRFT domain feature representation form, establishes the FRFT domain sparse representation model, and proposes a method to classify the sparse representation of active sonar targets with multi-order FRFT domain feature fusion. The method achieves the purpose of suppressing noise and reverberation interference through the energy aggregation of FRFT and removing the residual of sparse decomposition; Through the fusion of multi-order FRFT domain features, the separability of targets is improved, and the active sonar target classification with low SNR is realized. Experimental results show that the classification accuracy of the proposed method can reach more than 90% when the SNR is about 0 dB.
Marine environment noise and reverberation interference are serious, and the poor target separability is the bottleneck problem in active sonar target classification and recognition. In order to solve this problem, based on the echo signal model of active sonar target and the principle of FRactional Fourier Transform (FRFT), this paper deduces the multi-order FRFT domain feature representation form, establishes the FRFT domain sparse representation model, and proposes a method to classify the sparse representation of active sonar targets with multi-order FRFT domain feature fusion. The method achieves the purpose of suppressing noise and reverberation interference through the energy aggregation of FRFT and removing the residual of sparse decomposition; Through the fusion of multi-order FRFT domain features, the separability of targets is improved, and the active sonar target classification with low SNR is realized. Experimental results show that the classification accuracy of the proposed method can reach more than 90% when the SNR is about 0 dB.
2021, 43(3): 817-825.
doi: 10.11999/JEIT200660
Abstract:
Traditional Orthogonal Matching Pursuit (OMP) method needs high oversampling factor and computational overhead to estimate off-grid path delays in underwater acoustic Orthogonal Frequency Division Multiplexing (OFDM) systems. In this paper, the idea of path compensation is introduced from multiple linear fitting theory, and an improved OMP path delay estimation method based on path compensation is proposed to reduce the energy leaking from off-grid paths to its surrounding grids. The compensation distance is used to evaluate compensation effect. The improved algorithm can improve the estimation performance by appropriate compensation distance without increasing the oversampling factor. Compared with the traditional OMP method, the proposed algorithm has lower computational complexity but better estimation performance. The results of simulations and sea trial data decoding show the superiority of the proposed method.
Traditional Orthogonal Matching Pursuit (OMP) method needs high oversampling factor and computational overhead to estimate off-grid path delays in underwater acoustic Orthogonal Frequency Division Multiplexing (OFDM) systems. In this paper, the idea of path compensation is introduced from multiple linear fitting theory, and an improved OMP path delay estimation method based on path compensation is proposed to reduce the energy leaking from off-grid paths to its surrounding grids. The compensation distance is used to evaluate compensation effect. The improved algorithm can improve the estimation performance by appropriate compensation distance without increasing the oversampling factor. Compared with the traditional OMP method, the proposed algorithm has lower computational complexity but better estimation performance. The results of simulations and sea trial data decoding show the superiority of the proposed method.
2021, 43(3): 826-833.
doi: 10.11999/JEIT200446
Abstract:
The small moving target detection in complex underwater environment is complicated due to the weak target signal strength and low signal-to-clutter ratio. A Track-Before-Detect (TBD) algorithm based on subspace projection is proposed to solve these problems. A sequence motion track fragment is extracted from the original data, and then projected from the 3D space-time onto the 2D subspace. The morphological features in 2D subspace are applied to preliminary screening to remove most of the clutters and locate the local motion areas of the target. The 3D space-time track is reconstructed from 2D subspace in these local motion areas. During the above 3D track backtracking process, the motion continuity characteristics are also extracted to further remove the clutters and select the effective target track fragments. Through the above-mentioned hierarchical processing, a fast and high-precision target track fragment detection algorithm is achieved. By combing this track fragment detection algorithm with the foreground detection and Hierarchical Agglomerative Clustering (HAC) based long-time track detection algorithms, a complete TBD scheme for small moving targets detection is constructed. The accuracy and speed of this detection scheme are verified on the real sonar image data.
The small moving target detection in complex underwater environment is complicated due to the weak target signal strength and low signal-to-clutter ratio. A Track-Before-Detect (TBD) algorithm based on subspace projection is proposed to solve these problems. A sequence motion track fragment is extracted from the original data, and then projected from the 3D space-time onto the 2D subspace. The morphological features in 2D subspace are applied to preliminary screening to remove most of the clutters and locate the local motion areas of the target. The 3D space-time track is reconstructed from 2D subspace in these local motion areas. During the above 3D track backtracking process, the motion continuity characteristics are also extracted to further remove the clutters and select the effective target track fragments. Through the above-mentioned hierarchical processing, a fast and high-precision target track fragment detection algorithm is achieved. By combing this track fragment detection algorithm with the foreground detection and Hierarchical Agglomerative Clustering (HAC) based long-time track detection algorithms, a complete TBD scheme for small moving targets detection is constructed. The accuracy and speed of this detection scheme are verified on the real sonar image data.
2021, 43(3): 834-841.
doi: 10.11999/JEIT200570
Abstract:
Autonomous Underwater Vehicle (AUV) localization is one of the main methods to locate underwater targets in large areas. Considering the defects of single AUV positioning, such as long positioning period, low positioning coverage area and large accumulation of positioning errors for a long time, a cooperative target localization method based on double orthogonal moving AUVs is proposed. Each AUV locates itself through its own Inertial Navigation System (INS) and Doppler log. Localization process is achieved through the measurement of time difference of arrive between the target and AUV in the course of multiple movements. This method requires two orthogonal moving AUVs with a relative heading angle of 90°to achieve one positioning process by communicating at least two times. Compared with the traditional single mobile sensor positioning method, the proposed method requires shorter positioning period and lower synchronization requirements. Experimental results show that the positioning accuracy of this method is significantly improved, while the effective positioning area is enlarged, and the influence on AUV position error is lower in the long time positioning process.
Autonomous Underwater Vehicle (AUV) localization is one of the main methods to locate underwater targets in large areas. Considering the defects of single AUV positioning, such as long positioning period, low positioning coverage area and large accumulation of positioning errors for a long time, a cooperative target localization method based on double orthogonal moving AUVs is proposed. Each AUV locates itself through its own Inertial Navigation System (INS) and Doppler log. Localization process is achieved through the measurement of time difference of arrive between the target and AUV in the course of multiple movements. This method requires two orthogonal moving AUVs with a relative heading angle of 90°to achieve one positioning process by communicating at least two times. Compared with the traditional single mobile sensor positioning method, the proposed method requires shorter positioning period and lower synchronization requirements. Experimental results show that the positioning accuracy of this method is significantly improved, while the effective positioning area is enlarged, and the influence on AUV position error is lower in the long time positioning process.
2021, 43(3): 842-849.
doi: 10.11999/JEIT200649
Abstract:
In view of enhancing the time delay estimation resolution for the target echo in a complex shallow-water environment, thus improving the target detection ability of the active sonar system. A high-resolution time delay estimation technique is proposed to detect the underwater target based on sparse representation theory and deconvolution framework. Firstly, the Toeplitz operator is introduced here to construct a dictionary matrix using the various time delayed replicing of the transmitting signal. The estimated time-delay value can be found in the desired sparse vector solution. Secondly, the Alternating Direction Method of Multipliers (ADMM) is implemented to calculate the optimal solution globally. Thirdly, the reweighted iteration approach is explored to control the regularization parameter, thus suppressing the impact of the multipath channel. The arrival time of the echo can be decoupled to obtain a high-resolution time delay result. The simulated and experimental data verify that the proposed deconvolution-based time delay estimation technique can be used to detect the underwater target in shallow-water acoustic multipath channels. The resolution of the estimated time-delay result can achieve 0.056 ms.
In view of enhancing the time delay estimation resolution for the target echo in a complex shallow-water environment, thus improving the target detection ability of the active sonar system. A high-resolution time delay estimation technique is proposed to detect the underwater target based on sparse representation theory and deconvolution framework. Firstly, the Toeplitz operator is introduced here to construct a dictionary matrix using the various time delayed replicing of the transmitting signal. The estimated time-delay value can be found in the desired sparse vector solution. Secondly, the Alternating Direction Method of Multipliers (ADMM) is implemented to calculate the optimal solution globally. Thirdly, the reweighted iteration approach is explored to control the regularization parameter, thus suppressing the impact of the multipath channel. The arrival time of the echo can be decoupled to obtain a high-resolution time delay result. The simulated and experimental data verify that the proposed deconvolution-based time delay estimation technique can be used to detect the underwater target in shallow-water acoustic multipath channels. The resolution of the estimated time-delay result can achieve 0.056 ms.
2021, 43(3): 850-856.
doi: 10.11999/JEIT200315
Abstract:
To solve the problems of time-varying underwater acoustic channel estimation and equalization, an estimation and equalization algorithm of time-varying underwater acoustic channel based on Superimposed Training (ST) and Low-complexity Turbo Equalization (LTE) in frequency domain (ST-LTE) is proposed. Based on the ST scheme, the training sequence and symbols are linearly superimposed to make the channel information of the training sequence and symbols consistent; Based on the least square algorithm, channel estimation is performed. Based on the interference elimination technique of training sequence in frequency domain, the interference of training sequence on symbols is eliminated in frequency domain; Based on the Linear Minimum Mean Square Error (LMMSE) equalization algorithm in frequency domain, the low-complexity channel equalization (symbol estimation) is realized by the calculation of prior, posterior, extrinsic mean and variance; Based on the Turbo equalization algorithm, soft reconstruction of superimposed training and update of channel estimation are conducted, the information exchange between equalizer and decoder is also carried out and the performance of channel equalization is extremely improved by using coding redundancy information. Simulation, static communication experiment in a pool (communication frequency is 12 kHz, bandwidth 6 kHz, the sampling frequency 96 kHz, the transmission rate of symbols 4.8 ksym/s and the power ratio of the training sequence on symbols 0.25:1) and moving communication experiment in Jiaozhou Bay (communication frequency is 12 kHz, bandwidth 6 kHz, the sampling frequency 96 kHz, the transmission rate of symbols 3 ksym/s and the power ratio of the training sequence on symbols 0.25:1) are carried out and simulation and experimental results verify the effectiveness of the proposed algorithm.
To solve the problems of time-varying underwater acoustic channel estimation and equalization, an estimation and equalization algorithm of time-varying underwater acoustic channel based on Superimposed Training (ST) and Low-complexity Turbo Equalization (LTE) in frequency domain (ST-LTE) is proposed. Based on the ST scheme, the training sequence and symbols are linearly superimposed to make the channel information of the training sequence and symbols consistent; Based on the least square algorithm, channel estimation is performed. Based on the interference elimination technique of training sequence in frequency domain, the interference of training sequence on symbols is eliminated in frequency domain; Based on the Linear Minimum Mean Square Error (LMMSE) equalization algorithm in frequency domain, the low-complexity channel equalization (symbol estimation) is realized by the calculation of prior, posterior, extrinsic mean and variance; Based on the Turbo equalization algorithm, soft reconstruction of superimposed training and update of channel estimation are conducted, the information exchange between equalizer and decoder is also carried out and the performance of channel equalization is extremely improved by using coding redundancy information. Simulation, static communication experiment in a pool (communication frequency is 12 kHz, bandwidth 6 kHz, the sampling frequency 96 kHz, the transmission rate of symbols 4.8 ksym/s and the power ratio of the training sequence on symbols 0.25:1) and moving communication experiment in Jiaozhou Bay (communication frequency is 12 kHz, bandwidth 6 kHz, the sampling frequency 96 kHz, the transmission rate of symbols 3 ksym/s and the power ratio of the training sequence on symbols 0.25:1) are carried out and simulation and experimental results verify the effectiveness of the proposed algorithm.
2021, 43(3): 857-864.
doi: 10.11999/JEIT200657
Abstract:
In order to obtain higher resolution and avoid the failure of Two-Dimensional (2D) of Direction-Of-Arrival (DOA) estimation, One-Dimensional (1D) spatial DOA estimation method, vertical DOA estimation via Vernier method and horizontal DOA estimation method via minimum angle theorem are proposed. First, covariance matrices are constructed based on various subarrays to alleviate the failure of 2D model, and the Khatri-Rao product is adopted to extend the virtual array aperture. Second, the extended observation models and corresponding array steer vector are exploited for 2D DOA estimation. Compared with the steer vector of the original array, the number of virtual array elements is doubled, and thus the array aperture is extended. Simulation results show that the proposed method has better resolution and lower RMSE performance in 2D DOA estimation problem compared with the Single Measured Vector Beamforming method. The tank experiment further verifies the engineering practicability of the proposed method.
In order to obtain higher resolution and avoid the failure of Two-Dimensional (2D) of Direction-Of-Arrival (DOA) estimation, One-Dimensional (1D) spatial DOA estimation method, vertical DOA estimation via Vernier method and horizontal DOA estimation method via minimum angle theorem are proposed. First, covariance matrices are constructed based on various subarrays to alleviate the failure of 2D model, and the Khatri-Rao product is adopted to extend the virtual array aperture. Second, the extended observation models and corresponding array steer vector are exploited for 2D DOA estimation. Compared with the steer vector of the original array, the number of virtual array elements is doubled, and thus the array aperture is extended. Simulation results show that the proposed method has better resolution and lower RMSE performance in 2D DOA estimation problem compared with the Single Measured Vector Beamforming method. The tank experiment further verifies the engineering practicability of the proposed method.
2021, 43(3): 865-872.
doi: 10.11999/JEIT200672
Abstract:
The output Signal-to-Noise Ratio(SNR) of conventional broadband energy detection is reduced in multi-target and strong interference environment, and the detection performance is great reduced. In order to solve the problem, a broadband detection method based on space-frequency joint optimal filtering which combines subarray STeered Minimum Variance(STMV) broadband adaptive beamforming with Eckart filtering is proposed. Firstly, the spatial adaptive processing is carried out by subarray STMV beamforming, and the optimal filtering is realized in spatial domain by using the interference suppression ability of adaptive beamforming. Then, the power spectrum of signal and noise is estimated by maximum likelihood estimation, and Eckart filter is constructed to assign different weights to the output of adaptive beamforming to maximize the output SNR in frequency domain. The influence of spatial sidelobe interference and noise in the frequency band are reduced to make the output SNR maximum by the proposed method. The broadband detection ability of the target can be effectively improved and the broadband detection performance of passive sonar is also improved. The simulation and experimental data processing results verifiy the effectiveness of the method.
The output Signal-to-Noise Ratio(SNR) of conventional broadband energy detection is reduced in multi-target and strong interference environment, and the detection performance is great reduced. In order to solve the problem, a broadband detection method based on space-frequency joint optimal filtering which combines subarray STeered Minimum Variance(STMV) broadband adaptive beamforming with Eckart filtering is proposed. Firstly, the spatial adaptive processing is carried out by subarray STMV beamforming, and the optimal filtering is realized in spatial domain by using the interference suppression ability of adaptive beamforming. Then, the power spectrum of signal and noise is estimated by maximum likelihood estimation, and Eckart filter is constructed to assign different weights to the output of adaptive beamforming to maximize the output SNR in frequency domain. The influence of spatial sidelobe interference and noise in the frequency band are reduced to make the output SNR maximum by the proposed method. The broadband detection ability of the target can be effectively improved and the broadband detection performance of passive sonar is also improved. The simulation and experimental data processing results verifiy the effectiveness of the method.
2021, 43(3): 873-880.
doi: 10.11999/JEIT200638
Abstract:
The underwater positioning system is the essential and necessary equipment for modern deep-sea operations. Accurate time delay estimation is the basis for achieving high-precision localization. However, the performance of time delay estimation will decrease due to the long-distance transmission and strong clutter. To solve this problem, a novel interference suppression method based on the subspace theory is proposed. The dimension of the subspace is estimated using Bayesian information criterion at first, and the probability density function under different assumptions are then derived by estimating the unknown parameters to construct the generalized likelihood ratio. Finally, the most suitable subspace may be estimated using the optimal matching generalized likelihood ratio detection method. The projection operator is constructed to linearly project the received data, suppressing the interference and noise as a result of improving the accuracy of time delay estimation. The simulation results show that the proposed method can effectively suppress the influence of the clutter and improve the estimation accuracy of the time-delay.
The underwater positioning system is the essential and necessary equipment for modern deep-sea operations. Accurate time delay estimation is the basis for achieving high-precision localization. However, the performance of time delay estimation will decrease due to the long-distance transmission and strong clutter. To solve this problem, a novel interference suppression method based on the subspace theory is proposed. The dimension of the subspace is estimated using Bayesian information criterion at first, and the probability density function under different assumptions are then derived by estimating the unknown parameters to construct the generalized likelihood ratio. Finally, the most suitable subspace may be estimated using the optimal matching generalized likelihood ratio detection method. The projection operator is constructed to linearly project the received data, suppressing the interference and noise as a result of improving the accuracy of time delay estimation. The simulation results show that the proposed method can effectively suppress the influence of the clutter and improve the estimation accuracy of the time-delay.
