Iet Radar Sonar and Navigation最新文献

The authors present multiple machine learning-based methods for distinguishing maritime targets from sea clutter. The main goal for this classification framework is to aid future millimetre wave radar system design for marine autonomy. Availability of empirical data at this frequency range in the literature is scarce. The classification and anomaly detection techniques reported here use experimental data collected from three different field trials from three different millimetre wave radars. Two W-band radars operating at 77 and 94 GHz and a G-band radar operating at 207 GHz were used for the field trial data collection. The dataset encompasses eight classes including sea clutter returns. The other targets are boat, stand up paddleboard/kayak, swimmer, buoy, pallet, stationary solid object (i.e. rock) and sea lion. The Doppler signatures of the targets have been investigated to generate feature values. Five feature values have been extracted from Doppler spectra and four feature values from Doppler spectrograms. The features were trained on a supervised learning model for classification as well as an unsupervised model for anomaly detection. The supervised learning was performed for both multi-class and 2-class (sea clutter and target) classification. The classification based on spectrum features provided an 84.3% and 80.1% validation and test accuracy respectively for the multi-class classification. For the spectrogram feature-based learning, the validation and test accuracy for multi-class increased to 93.3% and 88.7% respectively. For the 2-class classification, the spectrum feature-based training accuracies are 88.1% and 86.8%, whereas with the spectrogram feature-based model, the values are 95% and 94.1% for validation and test accuracies respectively. A one class support vector machine was also applied to an unlabelled dataset for anomaly detection training, with 10% outlier data. The cross-validation accuracy has shown very good agreement with the expected anomaly detection rate.

Accurate and robust time delay estimation is crucial for synthetic aperture sonar (SAS) imaging. A two-step time delay estimation method based on displaced phase centre antenna (DPCA) micronavigation has been widely applied in motion estimation and compensation of SASs. However, the existing methods for time delay estimation are not sufficiently robust, which reduces the performance of SAS motion estimation. Deep learning is currently one of the cutting-edge techniques and has brought about a remarkable progress in the field of underwater acoustic signal processing. In this study, a deep learning-based time delay estimation method is introduced to SAS motion estimation and compensation. The subband processing is first applied to obtain ambiguous time delays between adjacent pings from phases of SAS echoes. Then, a lightweight neural network is utilised to construct phase unwrapping. The model of the employed neural network is trained with simulation data and applied to real SAS data. The results of time delay estimation and motion compensation demonstrate that the proposed neural network-based method has much better performance than the two-step and joint-subband methods.

The joint transmit and receive beamforming algorithm has been envisioned to optimise the target's signal-to-interference-plus-noise ratio by co-designing the transmit and receive beamforming vectors simultaneously. However, the traditional design concepts for this algorithm only consider the monostatic radar system, which may not work as well for the multistatic radar system. The authors present a novel transmit and receive beamforming algorithm for the multistatic radar system that includes a common transmitter and multiple receivers. The transmit beamforming vector can directly influence the output signal-to-interference-plus-noise ratio of each radar subsystem. To ensure the balanced performance of the subsystems, the authors propose a weighted sum of the signal-to-interference-plus-noise ratio optimisation problem to co-design the transmit and receive beamforming. The proposed problem is non-convex, and the authors construct an iterative algorithm to solve it using semi-definite relaxation and slack-variable replacement techniques. By using pre-determined weights, the output performance of each radar subsystem can be effectively regulated. The simulation results confirm that the proposed algorithm can ensure better output signal-to-interference-plus-noise ratio for the entire multistatic radar system than the conventional joint transmit and receive beamforming algorithm.

Space-time adaptive processor (STAP) has been widely used for global navigation satellite system (GNSS) anti-jamming receiver due to its good anti-jamming performance. When direction of satellite is unknown, STAP can be implemented based on power inversion (PI) criterion. However, existing space-time PI algorithm will introduce tens to hundreds of degrees biases into carrier phase, and sometimes will even cause cycle slips, which will reduce the success rate of ambiguity resolution, ultimately deteriorating positioning accuracy. A distortion-less carrier phase tracking space-time PI algorithm is proposed. The main novelty is that the proposed algorithm keeps the coefficients of the temporal taps as real values by imposing constraints on the weights of the antenna array. Several experiments are implemented to verify the effectiveness of the proposed algorithm. For comparison, the results of PI algorithm and minimum variance distortion-less response (MVDR) algorithm are shown. Results show that when the number, style, and direction of interferences and the direction of GNSS signal vary, different degrees of biases are introduced into carrier phases for the PI and the MVDR algorithm. However, no bias is introduced into the proposed algorithm. As a result, the effectiveness of the proposed algorithm is verified.

In view of the increasing number of space objects, comprehensive high-quality space surveillance becomes ever more important. Radar is a powerful tool that, in addition to detection and tracking of objects, also enables spatially high-resolution imaging independent of daylight and most weather conditions. Together with the technique of Inverse Synthetic Aperture Radar (ISAR), very high-resolution and distance-independent two-dimensional images can be obtained. However, advanced high-performance radar imaging of space objects is a complex and demanding task, touching many technological and signal processing issues. Therefore, besides theoretical work, the Microwaves and Radar Institute of German Aerospace Center (DLR) has developed and constructed an experimental radar system called IoSiS (Imaging of Satellites in Space) for basic research on new concepts for the acquisition of advanced high-resolution radar image products of objects in a low earth orbit. Based on pulse radar technology, which enables precise calibration and error correction, IoSiS has imaged space objects with a spatial resolution in the centimetre range, being novel in public perception and accessible literature. The goal of this paper is therefore to communicate and illustrate comprehensively the technological steps for the construction and successful operation of advanced radar-based space surveillance. Besides the basic description of the IoSiS system design this paper outlines primarily useful theory for ISAR imaging of objects in space, together with relevant imaging parameters and main formulae. All relevant processing steps, necessary for very high-resolution imaging of satellites in practice, are introduced and verified by simulation results. Finally, a unique measurement result demonstrates the practicability of the introduced processing steps and error correction strategies.

The authors study the problem of compressive target detection in a single-frequency network (SFN)-based multistatic passive radar system (MS-PRS) consisting of multiple illuminators of opportunity (IOs) and one receiver. Firstly, a generalised likelihood ratio test (GLRT)-based SFN-based compressive subspace detector (SFN-CSD) is derived by exploiting the sparsity of the target echoes for the case of known noise variance. When the noise variance is unknown, an SFN-based unknown-noise (UN) compressive subspace detector is proposed, referred to as the SFN-UNCSD. Moreover, closed-form expressions of the probability of false alarm and detection of the proposed detectors are deriived. It is proved that the SNF-UNCSD has a constant false alarm rate (CFAR) property. Finally, numerical simulations are conducted to verify the theoretical analysis and illustrate the performance of the proposed detector relative to several benchmark detectors.

The presented paper is further focused on the presentation and subsequent assessment of utilising a proposed Neural Network (NN) with simple architecture in the role of a signal preprocessing algorithm for the Constant False Alarm Rate detector and the fixed threshold detector applied on a Range-Doppler (RD) map with the aim of radar clutter impact reduction and minimisation of processing time. Based on a comparison of all tested algorithm results, it is possible to state that utilising the proposed NN with simple architecture led to reducing the impact of radar clutter when detecting radar targets on RD maps created from provided datasets. Comparing the mean processing time t_mean values of all tested algorithms, the authors can state that employing the proposed NN in combination with the fixed threshold detector led to a significant improvement in the computation time needed for processing one RD map while preserving the suppression of radar clutter and detection of the radar targets.

The authors deal with the problem of the design and manufacture of a high-resolution FMCW radar for drone detection and classification. The difficulties of the problem are discrete clutter reduction and spurious phase noise mitigation. The discrete clutter is due to the reflected signals from land vehicles, birds etc., while the spurious phase noise is inherent in the radar signal due to the phase-locked loop component and leakage between the transmitting and receiving paths. Both spurious phase noise and clutter will increase the system noise level and hence reduce the probability of detection of small targets such as drones and induce false alarms on the radar screen. In order to reduce discrete clutter, the authors propose a method to separate a drone from discrete clutter based on the design of the radar system parameters for a drone and its propeller detection, target's Doppler dispersion and moving characteristics. For spur mitigation, a method that focuses on the design of the isolation coefficient between transmitting and receiving paths to decrease the power of spurs below the minimum power requirement at the input of the analogue-to-digital converter is introduced. The results were applied by the authors to the development and manufacture of a radar with the given specifications for drone detection and classification. Different laboratory and field tests show that the spurs are mitigated and the drones are separated from discrete clutter with a range and accuracy better than the one recently published.

In the current complex situations of electronic intelligence (ELINT), the authors present a radar emitter structure (RES) identification method based on deep learning at a new level to address the issue of incomplete cognitive information. Firstly, due to the fact that existing simulation data cannot fully reflect the structure features of the entire radar emitter, the structure feature-level RES model is built using direct digital synthesiser (DDS) technology and radio frequency (RF) simulation platform. Afterwards, considering that the structure features are reflected in the frequency domain, a stacked frequency sparse auto encoder (sFSAE) network is constructed by adding a constraint term in frequency domain to the loss function of sparse auto encoder (SAE). Using deep learning to extract structure features with constraints in different domains is instructive for feature extraction techniques under variable operating parameters. Finally, the extracted structure features are input into the Softmax classifier to perform the identification from the radar signal to the RES. The experimental results show that the proposed method has high generalisation ability and robustness under different modulation types, different operating parameters and different signal to noise ratio (SNR). It also has a high identification rate even for untrained modulated signals.

Due to the thinner resolution range of broadband radar, ship recognition issues arise such that minor fluctuations within the targeted area significantly affect the high-resolution range profile (HRRP) of ships. Especially in the presence of reflector decoys around the surroundings of a ship, the HRRP of mixed targets might take a vastly different shape than of single ship, which makes it difficult to capture the effective features for ship identification. This article proposes a novel radar target recognition model based on parallel neural networks. The framework of this model consists of two stages: the data preprocessing and the parallel neural network. The data preprocessing stage effectively solves the sensitivity issue of HRRP and maps one-dimensional HRRP into a two-dimensional image. The second stage employs CNN and bidirectional LSTM to extract overall envelope features and temporal features, respectively. The parallel features are then processed by the Squeeze Excitation (SE) block to enhance critical information. The experimental results, based on HRRP data from mixed targets of ships and reflector decoys, demonstrate that the proposed model outperforms other methods in recognition performance and is quite robust against small sample sets, high noise, and large amounts of decoy jamming.