Search

To cluster the data with complex structures, Deep Subspace Clustering Network (DSCN) extracts the subspace relations among non-linear latent features. However, the performance improvement has encountered bottlenecks due to the lack of supervision. Meantime, in multi-view settings, most of DSCN-based methods underestimate the significance of view-fusion, which always adopt simple tactics. To address these issues, we propose a self-supervised model for simultaneous subspace clustering, consensus construction and self-guided learning, named as Self-Guided Deep Multi-view Subspace Clustering Network (SG-DMSC). We utilize DSCN to learn a complex subspace representation for each single-view. Considering their different importance, we design the view-fusion layer to establish the agreement. We construct a novel loss term, the spectral supervisor, so that the consensus can be more clustering-friendly by the self-guidance of pseudo labels. Theoretical support is provided to reflect the validity of this self-guided strategy. An alternate iterative optimization algorithm is presented to handle SG-DMSC. Experiments on real-world datasets confirm its efficacy compared with others.

This paper considers the hybrid beamforming design for wideband OFDM-DFRC system serving multiple users (MUs). The analog beamformer for the whole bandwidth and the digital beamformers for each subcarrier and are jointly optimized by considering flexible performance trade-off between the radar and communication. To deal with the resulting optimization problem, a consensus alternating direction method of multipliers (consensus-ADMM) framework based on the weighted mean-square error minimization (WMMSE) approach is proposed. Numerical simulations are provided to demonstrate the effectiveness of the proposed scheme.

Sufficient real information in generator is a critical point for the generation ability of GAN. However, GAN and its variants suffer from lack of this point, resulting in brittle training processes. In this paper, we propose a novel variant of GAN, Posterior Promoted GAN (P2GAN), which promotes generator with the real information in the posterior distribution produced by discriminator. In our frame-work, different from other variants of GAN, the discriminator maps images to a multivariate Gaussian distribution and extracts real information. The generator employs the real information by AdaIN and a latent code regularizer. Besides, reparameterization trick and pretraining is applied to guarantee a stable training process in practice. The convergence of P2GAN is theoretically proved. Experimental results on typical high-dimensional multi-modal datasets demonstrate that P2GAN has achieved comparable results with the state-of-the-art variants of GAN on unsupervised image synthesis.

In this paper, the problem of transmit beampattern synthesis (i.e., transmit beamforming) in multiple input multiple output (MIMO) radar which deploys one-bit digital-to-analog converts (DACs) is investigated. We aim to design appropriate transmit signal sequences, which are quantized by one-bit DACs, such that the amount of transmit energy can be focused into mainlobe region as much as possible, meanwhile, the leakage power of sidelobe region is minimized. It is shown that these requirements can be simultaneously fulfilled by minimizing the integrated sidelobe to mainlobe ratio (ISMR) of transmit beampattern with discrete binary constraints. According to this concept, we utilize the alternating direction multiplier method (ADMM) framework to solve the resulting nonconvex problem. Simulation results will demonstrate the effectiveness and improved performance of the proposed method.

This paper focuses on the problem of the optimization of nodes positions for sparse large aperture radar (SLAR) system. To overcome the deficiency of low target search efficiency in traditional SLAR system, a search scheme based on grating lobe controlling, in which the SLAR system searches multiple directions simultaneously by utilizing a transmit beampattern with multiple grating lobes, is provided. More specifically, the relationship between the nodes positions and grating lobes is mathematically discussed. On this basis, under the practical node position constraint, a design method of nodes positions for synthesizing a beampattern with desired grating lobes and low sidelobes is presented. Some representative simulations are provided to validate the effectiveness of the proposed scheme.

This paper proposes an optimization scheme for collocated multiple-input-multiple-output (MIMO) radar with a partial low-resolution ADCs architecture. With a criterion of the signal-to-interference plus quantization noise ratio (SIQNR) maximization, we formulate our problem by jointly optimizing the waveform, ADC switch vector and receive filter. Since the resultant problem is nonconvex, an alternating optimization framework (AOF) is proposed. Specifically, the block successive upper-bound minimization (BSUM) framework based on the Dinkelback method is devised to optimize the ADC switch vector and waveform matrix. Moreover, we propose a continuous and concave function to approximate the $l_{0}$ norm to obtain a high-quality sparse solution to the ADC switch vector. Numerical simulations are provided to demonstrate the effectiveness of the proposed schemes.

This paper addresses the long-time coherent integration detection problem for a weak maneuvering target, involving linear range migration (LRM), range curvature (RC), and Doppler frequency migration (DFM). To solve the above problems, an efficient coherent integration method based on frequency phase compensation function process (FPCFP) is proposed. By jointly searching the unknown motion parameters in velocity and acceleration domains, the FPCFP algorithm not only can eliminate LRM, RC, and DFM simultaneously, but also can realize the coherent integration. Compared with the existing algorithms, the proposed method achieves a better detection probability in a low signal-to-noise ratio (SNR) environment and ensures high computational efficiency. Several experiments are performed to verify the effectiveness of the presented approach.

This paper investigates a hybrid beamforming design for multi-user dual-function radar-communication (DFRC) system. The hybrid transmit/receive beamformers are designed by maximizing the sum-rate under constraints of power as well as similarity between the designed beamformer and the reference one that has good beampattern property. To solve the nonconvex optimization problem, we devise a two-stage method. To be more specific, in the first stage, we obtain the fully digital beamformer based on the weighted mean-square error minimization (WMMSE) method, in the second stage, we iteratively optimize the digital and analog beamformers to approximate the achieved fully digital beamformer. Numerical simulations are provided to demonstrate the effectiveness of the proposed schemes.

Considering the high sidelobe of the MVDR beamformer, a sparse constraint on beam pattern has been suggested to suppress the sidelobe of adaptive beamformer. Based on the sparse constraint, a modified Capon spectrum is proposed to form a new beamformer to compensate for the lack of robustness of the MVDR beamformer. In proposed method, the constructed penalty matrix gives the minimum penalty on the mainlobe region to keep its shape in existing steering vector error and meanwhile the positions of interferences distinguished from the sidelobe region by the modified Capon spectrum are imposed a larger penalty to obtain a performance of interference suppression as perfect as the MVDR beamformer's. What's more, the sidelobe level in our method is moderately controllable. Finally, an algorithm of the alternating direction method of multipliers (ADMM) is suggested to solve the problem efficiently. Numerical examples show the advantages of the proposed method in adaptive beamforming.

In this paper, we investigate the problem of transmit design for a Dual-Function Radar-Communication (DFRC) system with one-bit DACs. The problem is formulated by minimizing the symbol mean-square error, while ensuring the target localization performance for radar. In order to tackle the nonconvex discrete problem, a new approach based on the alternating direction method of multipliers (ADMM) algorithm is presented, and the closed-form solution of each primal variable in the ADMM framework is provided. Numerical simulations are provided to demonstrate the effectiveness of the proposed schemes.

This paper investigates the problem of hybrid transceiver design for dual-functional radar-communication (DFRC) system. Specifically, we introduce an information embedding scheme for the DFRC system with a subarray structure. The hybrid transmit/receive beamformers are designed by maximizing sum-rate under constraints of power and similarity between the designed bemaformer and the reference one with good beampattern property. Since the formulated problem is difficult to tackle, we propose an alternating optimization method based on the alternating direction method of multipliers (ADMM) framework to obtain the hybrid beamformer. Numerical simulations are provided to demonstrate the effectiveness of the proposed schemes.

In this paper, the problem of transmit beampattern design in multiple-input multiple-output (MIMO) radar with one-bit digital-to-analog converts (DACs) is investigated. The one-bit waveform sequence can be properly designed by minimizing the integrated sidelobe to mainlobe ratio (ISMR) of the transmit beampattern. However, due to the minimum ISMR criterion and discrete constraint, the formulated optimization problem for such a design is nonconvex and thus difficult to tackle directly. To this end, we employ the semidefinite relaxation (SDR) technique to modify the original problem to its convex counterpart. More importantly, the dimension of resulting large-scale semidefinite programming (SDP) problem is greatly reduced via exploiting the special block-spare structure. Simulation results will demonstrate the effectiveness and improved performance of our method.

Aiming at the problem of accurately detecting targets in real time from the 3D Lidar point cloud, a target detection algorithm based on the features of the scanning line segment is proposed. Firstly; the scanning lines are divided according to the Euclidean distance and vertical height difference between adjacent points. Then analyzes the characteristics of the adjacent line segments, such as pitch, slope, and height difference. Finally, line segments are marked as the ground segments, slope segments and obstacle segments by using Markov random field. The comparison experiments show that this algorithm can accurately segment the ground efficiently.

This paper focuses on the problem of coexistence between a colocated multiple-in multiple-out (MIMO) radar and down-link communication systems. We propose an iterative algorithm to minimize the Cramer-Rao Bound (CRB) on the direction of arrival (DOA) of a target, accounting for the energy and similarity constraints. More Specifically, at each iteration of this algorithm, the radar waveform is obtained with the aid of the alternating direction method of multipliers (ADMM) algorithm, and then the communication weights are obtained by exploiting the block successive upper-bound minimization method (BSUM). Finally, numerical results are presented to evaluate the effectiveness of the proposed algorithm.

In this paper, we investigate the problem of transmit beam-pattern design for MIMO radar with one-bit DACs. Based on the Fast Fourier Transform (FFT), we propose an alternating minimization (AM) method to minimize the weighted squared-error (WSE) between the designed beampattern and a given one. In particular, we propose an approximation approach to achieve a high-quality solution with one-bit constraint. Finally, numerical simulations demonstrate that the effectiveness of the proposed method.

In this paper, we focus on the waveform optimization problem of transmit beampattern design for a multiple-input multiple-output (MIMO) radar with low matched filtering range sidelobes (MFRS). The waveform is designed by minimizing a weighted summation of the transmit beampattern matching errors and the MFRS level under constant modulus constraint. Since the resulting problem is NP-hard, some auxiliary variables are introduced to recast it as a bi-convex quadratic problem, and then the alternating direction method of multipliers (ADMM) algorithm is applied to efficiently solve it. Finally, simulation results are presented to validate our approach.

This paper deals with grating-lobe clutter suppression problem by designing the transmit beamforming (TB) in the uniform subarray (US) configuration. In order to suppress the grating-lobe clutter of the main-lobe area in the conventional space-time adaptive processing (STAP), we propose an appropriate TB in an active array. Firstly, we utilize the information of grating-lobe locations and the respective clutter power to construct a transmit covariance matrix (TCM). Then, the TB can be designed resorting to the minimum variance distortionless response (MVDR) algorithm with the reconstructed TCM. Numerical results illustrate that the proposed method can mitigate the grating-lobe clutter effectively with a higher output signal-to-clutter-plus-noise ratio (SCNR) than traditional STAP.

This paper investigates the design of constant modulus waveforms with low auto- and cross-correlation sidelobes for multiple-input multiple-output (MIMO) radar. Unlike existing methods which indirectly solve an “almost equivalent” problem to the original one, we propose a new method to directly deal with the problem. Specifically, in the proposed algorithm, the fourth-order problem is first converted into a bi-quadratic problem via introducing an auxiliary primal variable, and then a primal-dual framework, named as “nonlinear alternating direction method of multipliers (nonlinear-ADMM)”, is derived to handle the modified problem. Moreover, the solution of each primitive variable can be computed efficiently. Numerical simulations are carried out to assess the performance of the proposed algorithm in terms of the WISL metric, auto- and cross- correlation properties and computational complexity.

This paper investigates the problem of the spectrally compatible waveform design for colocated multiple-input multiple-output (MIMO) radar transmit beampattern formation under peak to average power ratio (PAR) and similarity constraints. The waveform is designed by minimizing a weighted summation of the beampattern integrated sidelobe to mainlobe ratio (ISMR) and waveform energy over the space-frequency bands. Since the resulting problem is NP-hard, an approximated primal-dual algorithm inspired by the alternating direction method of multipliers (ADMM) algorithm is proposed. Moreover, the closed-form solution of each primal variable is provided. Numerical simulations are provided to demonstrate the effectiveness of the proposed approach.

This paper discusses the effect of range migration compensation error on extended target fusion detection, under the distributed high-speed moving platforms. We analyze the linear range migration caused by platforms. More precisely, we consider the situation that, when there exists the range migration compensation error introduced by the Inertial Navigation System (INS) of the platforms. Due to the particularity of extended target, we analyze the performance of fusion detection with the three kinds of target fluctuating models. The influences of compensation error on fusion detection are discussed via simulation results.

This paper investigates the problem of joint optimization of waveform covariance matrix and antenna position for multiple-input multiple-output (MIMO) radar transmit beampattern design. Since the formulated optimization problem of minimization of the mean-square error (MSE) between the designed transmit beampattern and the desired beampattern with respect to waveform covariance matrix R and antenna position p is NP-hard, an iterative method is devised to tackle the resulting problem. Specifically, we first optimize R by the semidefinite quadratic programming (SQP) algorithm for a fixed p, and then update p using the alternating direction method of multipliers (ADMM) method for a fixed R. Finally, numerical simulations demonstrate that the effectiveness of the proposed method.

In this paper, we present a resource-awareness decentralized colocated multiple-input multiple-output radar network for multiple targets tracking. Specifically, we develop a distributed particle filter which incorporates the scheduling of radars. Taking posterior Cramer-Rao lower bound as performance metric, we model the scheduling and power allocation problem of the radar network as a sparse optimization problem. Making use of the sparsity of the solution, the scheduling and power allocation performs jointly. We investigate the decomposable structure of the optimization problem and conduct a decentralized technique to solve it iteratively. We verify the effectiveness of the proposed decentralized radar network with the joint scheduling and power allocation (JSPA) scheme by simulation. The resource utilization is adaptively adjusted according to the states of targets and the performance demand.

This paper investigates the problem of the spectrally compatible waveform design for multiple-in multiple-out (MIMO) radar transmit beampattern formation, subject to peak-to-average-power ratio (PAR) and waveform similarity constraints. Since the formulated optimization problem of minimizing the Integrate Sidelobe Level (ISL) and the waveform energy of the stop-band frequencies is NP-hard, an auxiliary variable is first introduced to modify the non-convex problem into a bi-quasiconvex problem, and then the approximated alternating directions method of multipliers (A-ADMM) algorithm is proposed to tackle the resulting bi-quasiconvex problem. Finally, numerical results are presented to evaluate the effectiveness of the proposed algorithm.

In this paper, a novel approach is proposed to design the transmit waveform for multiple-in multiple-out (MIMO) radar. In practical applications, we usually need to design the transmit beam pattern to approximate a given transmit beam pattern and minimize the cross-correlation sidelobes via designing transmit waveforms for MIMO radar. In this work, we propose a one-step method to obtain the transmit waveform. We first formulate our problem as a nonconvex fourth order polynomial optimization problem subject to the energy constraint, and then we propose a distributed convex optimization technique known as alternating direction method of multipliers (ADMM) algorithm to sovle it. Finally, numerical results are presented to demonstrate the effectiveness of our devised approach.

In this paper, we investigate the location and velocity joint estimattion performance for distributed radars on moving platforms, each platform is equipped with phased array radar. The distributed radars system can be inplemented as non-coherent MIMO radar. Firstly, this paper presents maximum likelihood estimation of location and velocity, and then the Cramer-Rao bound for location and velocity joint estimation are derived. We analyze the influnce of the signal bandwith and observation time on estimation performance. Simulation results are provided to verify the correctness of Cramer-Rao bound derivation.

The paper considers the problem of how to simultaneously estimate the direction-of-arrival (DOA) of a source signal and the phase error of a uniform linear array (ULA). We introduce a new method to estimate those parameters for a partly calibrated ULA. As its name suggests, partly calibrated ULA means a array with some of its sensor elements have been calibrated. In particular, our method can be well executed when there is only one calibrated sensor. Firstly, the proposed method reconstructs some sets of data from the array snapshots. According to the subspace theory, a set of linear combination associated with DOA and array phase errors can be extracted with the help of eigenvalue decomposition (EVD). The unknown DOA and sensor phase erros can thus be described by a set of linear equation and its solution can be obtained by least squares (LS) approach. Different from other partily calibrated array based methods, this method can even work well under the condition that the calibrated element is not consecutively located round the reference one. The proposed method is both simple and effective, and simulation comparisons with other algorithms validate its effectiveness.