This study presents a convolutional neural network-based drone classification method. The primary criterion for a high-fidelity neural network-based classification is a real dataset of large size and diversity for training. The first goal of the study was to create a large database of micro-Doppler spectrogram images of in-flight drones and birds. Two separate datasets with the same images have been created, one with RGB images and others with greyscale images. The RGB dataset was used for GoogLeNet architecture-based training. The greyscale dataset was used for training with a series of architecture developed during this study. Each dataset was further divided into two categories, one with four classes (drone, bird, clutter and noise) and the other with two classes (drone and non-drone). During training, 20% of the dataset has been used as a validation set. After the completion of training, the models were tested with previously unseen and unlabelled sets of data. The validation and testing accuracy for the developed series network have been found to be 99.6 and 94.4%, respectively, for four classes and 99.3 and 98.3%, respectively, for two classes. The GoogLenet based model showed both validation and testing accuracies to be around 99% for all the cases.

The direction of arrival (DOA) is one of the most important parameters measured by electronic warfare systems. Monopulse direction finding techniques based on amplitude comparison have many advantages because of their simplicity but they generally have limited accuracy. There are many sources of error in such a system. This work focuses on improving the noise error. An algorithm is proposed for error reduction in amplitude comparison-based direction finding systems. The algorithm, based on signal level and DOA, adaptively utilises either adjacent or alternate antennas for the DOA measurement. No additional hardware is required for the implementation of this algorithm. A prototype system is developed using spiral antennas and the performance of the proposed technique is compared against traditional ratio-based and correlation-based amplitude comparison schemes. Simulations and laboratory experiments show that the DOA accuracy can be improved by an average of 36 and 21% as compared to the ratio-based and correlation-based methods, respectively.

The multi-jointed autonomous underwater vehicle (AUV) has higher manoeuvrability and a smaller turning radius in comparison with an AUV with a rigid body. It is able to implement spatially denser sample collection in three-dimensional space in the deep ocean for scientific expeditions. As conventional positioning approaches, such as ultra-short-baseline and long-baseline, cannot be employed to obtain the coordinates of a multi-jointed AUV, because they only work with constant baselines, a novel underwater acoustic positioning system with a single beacon and a varied baseline is proposed. The coordinate transformation based on the angles of rotating joints and the geometry of the multi-jointed AUV is mathematically modelled. Generalised cross-correlation is applied to measure the time delays of sound signals between the beacon and the hydrophones on the AUV. Numerical simulations and experiments in an anechoic water tank were carried out, and the impacts of parameters on the accuracy of the estimated coordinates of the multi-jointed AUV are examined. Accuracies of 0.86% in simulation and 0.2 m (at a range of 30 m) in experiments are demonstrated.

This study investigates the continuous coherent direction-of-arrival (DOA) estimation, and concentrated on developing grid-less sparsity-based methods to Gaussian coloured noise environment. The noise component is greatly suppressed by applying fourth-order cumulants (FOC) due to its blind property to additive Gaussian noise. Two grid-less sparse models are designed separately. The first sparse representation model is built based on the simplified FOC vector, which would effectively reduce the computational complexity. Then the dual atomic norm minimisation algorithms are applied to solve the basis mismatch problem and improve the estimation accuracy. Additionally, a Toeplitz matrix based on FOC vector is constructed. The second sparse model based on this Toeplitz FOC matrix is proposed to implement array aperture extension, which can break through the restriction of maximum signal number and improve resolution. The proposed methods can handle the coherent signals and do not require the signal number as a prior. Numerical simulations demonstrate the outperformance of the proposed methods in estimation precision, computational cost and robustness to coloured noise.

Collocated multiple-input multiple-output radar can track multiple targets simultaneously by transmitting multiple orthogonal beams and adopting the digital beamforming technology. In this scenario, the authors propose a joint power and time width allocation approach, which combines a cognitive tracking model based on the posterior Cramér-Rao lower bound (PCRLB) and the square-root cubature Kalman filter. The aim of the optimisation model is to improve the velocity estimation accuracy by minimising the sum of the PCRLBs of the velocity of multiple targets, which are predicted based on the feedback information from the cognitive tracking model. However, there are two finite working resources in the optimisation model: the total transmit power of multiple beams and the total effective time width of each corresponding signal. The resource allocation problem can be transformed into a non-convex optimisation problem, which can be converted into a standard convex optimisation problem by the linear relationship between the optimal power and the optimal time width. In this way, the joint power and time width allocation scheme is established as an adaptive closed-loop system. Numerical results demonstrate that the velocity tracking accuracy can be improved efficiently by the proposed algorithm.

One of the main shortcomings of received signal strength-based indoor localisation techniques is the labour and time cost involved in acquiring labelled ‘ground-truth’ training data. This training data is often obtained through fingerprinting, which involves visiting all prescribed locations to capture sensor observations throughout the environment. In this work, the authors present a helmet for localisation optimisation (H4LO): a low-cost robotic system designed to cut down on said labour by utilising an off-the-shelf light detection and ranging device. This system allows for simultaneous localisation and mapping, providing the human user with accurate pose estimation and a corresponding map of the environment. The high-resolution location estimation can then be used to train a positioning model, where received signal strength data is acquired from a human-worn wearable device. The method is evaluated using live measurements, recorded within a residential property. They compare the groundtruth location labels generated automatically by the H4LO system with a camera-based fingerprinting technique from previous work. They find that the system remains comparable in performance to the less efficient camera-based method, whilst removing the need for time-consuming labour associated with registering the user's location.

The transfer alignment accuracy between the master inertial navigation system (MINS), the slave inertial navigation system (SINS) and the online calibration capability of SINS are the important factors that affect the system performance. The traditional Kalman filtering-based transfer alignment method greatly depends on the prior information. The alignment and calibration based on optimisation-based alignment are preliminarily achieved without the prior information, but many deductions are required to achieve an optimal solution. In this study, a novel method aimed at transfer alignment and bias calibration is proposed, which is based on singular value decomposition by the inertial measurement vector matching equation to construct the least-squares solution. The performance of the system is analysed from the transfer alignment accuracy and the inertial instrument error estimation, respectively. The simulation and flight experiment results demonstrate that the proposed method is concise and effective without complex deduction and calculation. The relative attitude angle between MINS and SINS can be estimated accurately as well as the bias of SINS.

In this research, the authors extract features from intermediate frequency band radar signals in the time–frequency domain for classification. The extracted features are classified via support vector machine and K-nearest neighbour classifiers. They show the accuracy of classification is above 99% for different classes of radar signals except for frequency shift keying signal with accuracy 83% in negative signal-to-noise ratio (SNR). To identify the radars with the same class, the classification accuracy is 91% for SNR between 5 to 15 dB and 64% in the worst case for SNR between −1 to 10 dB. The proposed method is compared with some methods based on the empirical mode decomposition (EMD), cumulant and Zhao Atlas Mark Distribution (ZAMD). The results show that the classification error in the proposed method is less than that of EMD method 55% in the best case and 9% in the worst case. The performance of the cumulant-based method is weaker than that of the proposed method in common designed scenarios becoming almost similar only in one scenario. The ZAMD-based method could only distinguish the signals with different modulations in high SNR while it is unable to classify the signals with the same modulation but different parameters.

The authors present a non-traditional, parametric method for velocity, angle, and range estimation in a frequency diverse array radar. Unlike the traditional beamforming techniques, the proposed scheme transmits an omni-directional sinusoid regardless of the target locations. They propose a simple sampling strategy, which eliminates the need for employing a bank of bandpass filters at the receiver. Under the proposed sampling scheme the received data follows a convenient low rank model. They exploit this model to design a fast and accurate parametric estimation algorithm. Their velocity and range estimation steps employ known spectral analysis techniques. For angle estimation, they propose a new grid-less sparse recovery algorithm. The resulting methods are applicable to any arbitrary array geometry. Furthermore, they propose an efficient method to mitigate jamming. They also provide necessary guidelines to avoid ambiguity and achieve the desired resolution performance. The Cramér–Rao lower bound for the estimation problem is derived. The utility of the proposed method is demonstrated via numerical simulation results.

To deal with problems of uncertain modulations and multiple pulse widths in pulse waveforms (PWs) during the identifying procedure, a novel specific emitter identification (SEI) method based on PW images (PWIs) and convolutional neural network is proposed. In the method, a more accurate signal model is built with considering the rising, steady and falling part of the whole PW based on actual radar pulse signals. PWI achieves transforming time-domain waveforms to 2D binary images as an SEI analysis feature. To match the PWI feature, a convolutional neural network with the small convolutional kernel is designed to extract the subtle features and finish the supervised training. By tuning the parameters of the convolutional neural network, it completes a balance of consuming time and identifying accuracy. Simulations and experiments indicate that the proposed method outperforms the existed methods on identifying radar individuals with uncertain modulations and multiple pulse widths in the intercepted pulse signals.

This study presents a radial basis function (RBF) aided extended Kalman filter (EKF) (namely, novel RBFEKF: NRBFEKF) to improve attitude estimation solutions in GPS-Denied environments. The NRBFEKF has been developed and applied for attitude estimation using only the outputs of strap-down IMU (gyroscopes and accelerometers) and strap-down magnetometer. In general, neural networks have the capability to map input–output relationships of a system without a-priori knowledge about them. A properly designed RBF neural network is able to learn and extract complex relationships given enough training. Furthermore, if there is a platform with inputs, outputs and many sensors, the RBF is able to adapt all the changes of sensors output. The RBFEKF, which is based on EKF aided by RBF network is validated in Matlab environment using simulated trip data and real data acquired during an UAV's trip. The RBFEKF has increased the accuracy of attitude estimation compared to typical EKF. In addition, the RBF is trained to map the vehicle manoeuvre for tuning measurement noise covariance matrix. Simulation results show that estimated measurement noise covariance matrix is closed to the nominal values in cruise flight (stationary phase), while in non-stationary phase the trained RBF neglects measurements from accelerometers, where accelerometer measurement model is not valid during this phase.

Attitude estimation using multi-antenna Global Navigation Satellite System (GNSS) has caught broad attention in recent years. The methods of indirectly calculating attitude parameters based on baseline vectors have been widely used. However, the accuracy of such methods needs to be further improved. An algorithm of directly estimating attitude parameters based on adaptive Kalman filtering (AKF) is proposed. The high-precision phase measurements are used and the system noise covariance matrix corresponding to Euler angles and angle rates is adaptively adjusted according to the maximum a posteriori estimation principle. To decrease the linearisation errors and speed the filtering convergence, a switching strategy is implemented. Namely, when the number of fixed ambiguities is equal to or more than three, the authors deem that the attitude estimate is sufficiently accurate, hence the initial state vector is constructed to start filtering. Otherwise, the attitude parameters are indirectly obtained. The static and simulation experiments are carried out. A comparative study is implemented with the two methods of indirectly calculating attitude parameters. The static and simulation results demonstrate that the proposed method has a competent performance in both accuracy and fixing success rate. The simulation results also show that the adaptability of the AKF method is nice.

In this study, the problem of the low probability of intercept (LPI) performance optimisation for a joint radar-communications system (JRCS) is investigated, which can simultaneously estimate surveillance channel parameters from the target returns and decode the received communications signals. The primary purpose of the proposed LPI performance optimisation strategy is to improve the LPI performance of a JRCS by optimising the energy spectral density of radar waveform design and the communications power allocation while maintaining a predetermined mutual information threshold for parameter estimation and a certain communications data rate for information delivery. The traditional isolated sub-band (TISB) situation, radar isolated sub-band (RISB) situation, and communications isolated sub-band (CISB) situation are analysed. Subsequently, the technique of Lagrange multipliers and the Karush–Kuhn–Tuckers optimality conditions are employed to solve the resulting optimisation problems. Moreover, the successive interference cancellation method is adopted to process the received radar-communications compound signal. Finally, several numerical simulations are conducted to demonstrate the theoretical calculations and to validate the effectiveness of the proposed LPI performance optimisation strategy. It is also shown that the achievable LPI performance in RISB and CISB situations is much better than that in TISB situation.

This study presents a technique for removing the inevitable harmonic distortions and intermodulation products that primary frequency-modulated continuous wave (FMCW) radars inherently generate in the baseband. The required mixers and baseband amplifiers are non-linear devices that introduce signal distortions, depending on the positions of targets in the field of view. Target distances are directly coupled to their corresponding baseband frequencies. The mitigation strategy presented in this study involves changing the resulting baseband frequencies of targets by varying the mixer deramping frequency input. After applying a processing algorithm that involves spectral rotation and combination, harmonic contributions can be suppressed to a high degree, while target-related signals remain the same. After only four iterations, a gain of 30 dB in spurious free dynamic range can be achieved that can even be improved with more time on target. This behaviour is not only observed in simulations, but is also proven with measurements using an FMCW radar, working at a 3 GHz carrier frequency. The algorithm is applicable to radar systems that work in any frequency band.

In this study, a forward model for radio frequency tomography (RFT) is improved to remove the effect of strong sidelobes from dominant scatterers in the region of interest. This approach uses a ‘suppression algorithm (SA)’ to remove the effect of strong sidelobes on weak targets. SA is used to remove the effect of these strong sidelobes using the information from the dyadic contrast function (DCF). DCF is analysed in order to remove the effects of strong sidelobes generated by dominant cells in the measurement domain. The eigenvalues and eigenvectors for dominant cells are obtained to remodel the strong cells as a secondary source in the measurement scene. Furthermore, to simplify the inversion problem in the new forward model, iterative reconstruction algorithms is considered. Subsurface multiplicative algebraic reconstruction technique as additive technique is proposed to solve the new forward model RFT with less computing power and memory. The presented algorithm has been verified using simulated RFT data, generated by the computational electromagnetic software FEKO, for regular and irregular targets scenarios. The proposed research shows that using information from the DCF it is possible to obtain high quality imagery of buried weak targets in RFT.

To enhance the modulation recognition performance of emitter signals under low signal-to-noise ratio (SNR), a recognition system based on secondary time–frequency distribution, discriminative projection, and collaborative representation is proposed. Firstly, a novel time–frequency processing method, including sparse-domain noise reduction and secondary feature extraction, is proposed to reduce noise interference and information redundancy in time–frequency images. In this way, secondary time–frequency distribution with high stability and detailed representation is obtained. Then, the classifier based on discriminative projection and collaborative representation was designed to enhance the ability of low-dimensional representation and between-class discrimination, which optimised using the mini-batch random gradient descent method. As shown in the simulation, the overall average recognition success rate of this system aiming at eight types of emitter signals reaches 95.6% at the SNR of −8 dB. Results of simulation and analysis indicate the superiority of the proposed classification system in terms of robustness, timeliness, and adaptability.