This Letter presents a dual polarised monostatic patch antenna with a simple two-tap RF/analogue domain-based self-interference cancellation (SIC) (RF-SIC) circuitry to achieve high interport isolation in a comparatively wider bandwidth for a 2.4 GHz in-band full duplex wireless transceiver. A brief mathematical description for the deployed two-tap RF-SIC is also presented to provide insight into self-interference (SI) suppression in a comparatively wider bandwidth. The implemented antenna with a two-tap RF-SIC circuitry achieves around 80 dB T_x–R_x isolation for a 20 MHz bandwidth in addition to more than 90 dB peak isolation for measurements performed in a lab environment where SI may result from the nearby placed objects. The deployed two-tap RF-SIC circuit can be tuned to effectively cancel out SI caused by reflections of the radio signal from the surroundings. Moreover, an 80 dB isolation bandwidth can be tuned within a 10 dB return loss bandwidth of the implemented dual polarised antenna.

A single Rotman lens fed circular array requires phase shifters to beam steer, whereas a two-cascaded Rotman lens fed circular array can beam form and beam steer by simply exciting a single Rotman lens beam port. In this Letter, the authors show how cascaded Rotman lenses feeding a circular array can use this property in order to beam steer/shape and perform retrodirective action. Experimental validation at 9.3 GHz is given.

Radio frequency (RF) heating is a serious risk for a patient with an elongated implant during MRI scan. The conventional dosimeter of specific absorption rate (SAR) and the recently proposed B _1+rms may not accurately indicate the implant heating. The local B ₁ field around the tip of a lead and its relationship with the RF heating were studied by numerical simulations. It was found that regardless of the route of the lead, the angle or position of placement, there was a strong linear correlation between the averaged square of local B ₁ field and the rise in temperature. The evaluation range had an effect on the linearity. Generally, the local B ₁ field performed better as a dosimeter of RF heating than SAR or B _1+rms at the isocentre.

In this Letter, a novel approach for contactless, low-cost and low-power respiratory sensing is presented, which is based on a commercial LC tank. This sensor is able to measure the variation of losses of induced eddy currents during one respiration period. The developed sensor system can be used in multiple different application fields where low-cost and low-power respiratory sensing is necessary. Examples for that can be infant or patient beds where vital parameters have to be monitored or in other fields where contactless presence check is desired. At first, the theory of operation and requirements for the respiratory sensing applications are explained which is followed by the system design and measurement results.

This Letter presents a wide supply voltage range, ultra-low power, and CMOS-only subthreshold voltage reference. A complementary-to-absolute-temperature () generator implemented by a standard transistor and a high transistor is used to obtain a negative temperature coefficient (TC) voltage. While, a proportional-to-absolute-temperature () generator adopts a unbalanced differential pair to achieve a positive TC voltage. With both generator and generator, the CMOS-only voltage reference circuit is proposed, which is implemented in a standard CMOS process. The simulation results show that the mean output reference voltage is 225.5 mV at room temperature with a supply voltage as low as 0.55 V, meanwhile the output reference voltage achieves a TC of with a temperature range from to . With a supply voltage of 0.55 V and at room temperature, the current consumption is 13.6 nA and the power consumption is only 7.5 nW. The reference circuit correctly works with a supply voltage ranging from 0.55 to 1.8 V, and it obtains the line regulation of 0.0022%/V.

The dark offset which is the reference value of an ambient light sensor is analysed. It varies across the wafer when the temperature is increased because of the leakage current of switch transistors. The difference between before and after wafer revisions shows the clue of leakage current. Modern fabrication processes which change the threshold voltage of switch transistors are surveyed. To explain the discrepancy between test and survey data, some other process issues are discussed.

This Letter introduces a histogram-based time-to-digital-converter (TDC) array for direct time-of-flight depth sensors. The 12-bit TDC array measures the time of flight (TOF) of a light pulse that is detected using an external single-photon avalanche diode (SPAD) array. As SPADs are noisy owing to dark electrons and scattered photons, statistical measurements based on histograms are essential, but require a large memory. In this work, the authors propose a mixed-signal TDC with an integrated histogram generation unit (HGU) that reduces memory requirement significantly as well as filters out invalid TOF readings. In addition, for application of the high-resolution SPAD array, an area-efficient TDC array with high uniformity was implemented by the proposed phase-dependent latching and temporal double sampling schemes. The prototype chip was fabricated using a 180 nm CMOS process, including 8-channel TDCs. The measurements show an integral non-linearity (INL)/a differential nonlinearity (DNL) of 0.76/0.49 least significant bit (LSB) and high uniformity under 0.19 LSB. The HGU was designed off-chip for prior verification, and was post-simulated with the measured TOF from the fabricated chip. Using the histogram-based TDC, the authors could detect an object located at a distance of 3 m accurately while reducing memory requirement by more than 128 times.

The authors demonstrate two types of non-reciprocal waveform-selective metasurfaces that control the level of reflectance, depending not only on the direction of the incident wave but also on the waveform, namely, on the pulse width. A capacitor-based waveform-selective metasurface exhibits strong reflectance for a short pulse compared to that for a continuous wave (CW) at the same frequency. In contrast, an inductor-based waveform-selective metasurface more strongly reflects a CW than a pulse. The difference between them is demonstrated to be more than 50% but is reduced for a wave propagating in the other direction. The concept of the authors' structures is expected to offer an additional degree of freedom with which to design non-reciprocal devices and applications.

This Letter proposes a robust multi-view spectral clustering approach. It first calculates a normalised graph Laplacian for each single view, and then uses them to recover a shared low-rank Laplacian by the low rank and sparse matrix decomposition. To achieve matrix decomposition, partial sum minimisation of singular values is leveraged to design a novel objective function, which can be optimised by the augmented Lagrangian multiplier algorithm to recover a common normalised graph Laplacian. Accordingly, multi-view clustering results can be obtained by taking spectral clustering on the common Laplacian. Experimental results illustrate its effectiveness over other related approaches.

Multimodal medical image registration remains a challenging problem when strong appearance variations and imprecise alignment exist in images. Previous deep network approaches cannot handle such a high degree of variability and prohibit the use of strong geometric constraints. The authors introduce a novel deep architecture that not only produces image representations that are well-suited for this challenging task, but also leverages knowledge of the geometry constraints for robust registration. By enforcing the representations of different modalities living in a common semantic space, they obtain convolutional features tending to respond to object parts consistently across modality. This yields a unique description for all object fragments and allows the end user to know the model's decision process, whereas most existing models remain unclear and difficult to explain. By using the differentiable spatial transformer to compensate transform, they integrate geometric consensus into the cost function to enable end-to-end model optimisation which has not yet been exploited before. The authors evaluate their method on a very challenging medical image dataset. Experiments demonstrate that the proposed method provides a plausible representation and outperforms state-of-the-art approaches by a significant margin.

Deep learning methods are nowadays considered as state-of-the-art approach in many sophisticated problems, such as computer vision, speech understanding or natural language processing. However, their performance relies on the quality of large annotated datasets. If the data are not well-annotated and label noise occur, such data-driven models become less reliable. In this Letter, the authors present very simple way to make the training process robust to noisy labels. Without changing network architecture and learning algorithm, the authors apply modified error measure that improves network generalisation when trained with label noise. Preliminary results obtained for deep convolutional neural networks, trained with novel trimmed categorical cross-entropy loss function, revealed its improved robustness for several levels of label noise.

The usefulness of applying singular value decomposition (SVD) on triaxial accelerometer data for one-dimensional (1D) convolutional neural network (CNN) based fall and activity recognition is investigated. Three-dimensional reduction methods, namely, SVD, sparse principal component analysis, and kernel principal component analysis, are compared for their effectiveness in extracting useful features for fall and activity recognition. Experiments conducted on three public falls and activities of daily living datasets show that SVD applied to acceleration data coupled with raw acceleration data or acceleration signal magnitude vector exhibited better 1D CNN fall and activity recognition accuracy than those using other principal component analysis based acceleration features.

In this Letter, the authors introduce a fractional angular number transform (FrANT). The new transform corresponds to a finite field version of the complex-valued discrete fractional angular transform and, therefore, its computation requires integer arithmetic only. Differently from other number-theoretic transforms, the possible lengths of an FrANT do not depend on the algebraic structure where it is defined. This provides more flexibility regarding the application of the transform in communications, cryptography and error correcting codes, for instance.

It has been shown that more accurate signal recovery can be achieved with low-coherence dictionaries in sparse signal processing. In this Letter, the authors extend the low-coherence attribute to receptive fields in convolutional neural networks. A new constrained formulation to train low-coherence convolutional neural network is presented and an efficient algorithm is proposed to train the network. The resulting formulation produces a direct link between the receptive fields of a layer through training procedure that can be used to extract more informative representations from the subsequent layers. Simulation results over three benchmark datasets confirm superiority of the proposed low-coherence convolutional neural network over the unconstrained version.

In this Letter, three types of bearingless switched reluctance machines (BSRMs) with different hybrid stator structures are designed and analysed. First of all, topologies of three BSRMs, the conventional 8/10 pole, 12/14 pole BSRMs and a novel 12/12 pole BSRM with axial split phase inner stator permanent magnet structure, are given. Then, the electromagnetic performances, including the distribution of magnetic fields, suspension force and electromagnetic torque are comprehensively compared by means of 2D and 3D finite-element method (FEM). The results based on theoretical analysis and FEM reveal that the proposed 12/12 pole BSRM would exhibit complete decoupling, shorter magnetic circuit, higher torque (power) density and suspension force than the conventional ones. Further, it is found that four degree-of-freedom suspension of the rotor can be realised by two-phase suspension windings in the proposed 12/12 pole BSRM. Finally, a prototype machine is designed and manufactured to verify the validity of the topology design and electromagnetic performance.

In this Letter, the authors propose a novel fully integrable architecture for a quasi-circulator. It is comprised of an inductorless and wideband structure and off-chip components are not required. This solution is appropriate for use in backscattering systems, such as RFID-tag readers. The proposed quasi-circulator differs from passive off-chip circulators because the power amplifier is embedded and the terminals have non-standard impedances. Thus, they propose a leakage figure as a figure-of-merit, which is related to the isolation of conventional circulators. The structure described herein is applied to a wireless power transfer system with an output power of 25 dBm at 990 MHz.

A method to effectively suppress the spurious modes in a hemispherical resonator is proposed and the suppression contributes to extension of the spurious-free stopband for the hemispherical-resonator-based bandpass filters. The TM_2m1 and TE₁₀₁ modes in a hemispherical resonator are suppressed by interrupting the corresponding surface current via properly slotting the cavity shell, without significantly interfering with the fundamental TM₁₀₁ mode. Resonant modes in a slotted hemispherical (SH) resonator are analysed, and design and implementation of a proof-of-concept X-band fourth-order SH-resonator waveguide filter are presented. The filter is monolithically prototyped by employing metallic additive manufacturing technology. The filter demonstrates passband insertion and returns losses of averagely 0.9 dB and >20 dB, respectively, a small frequency shift of <0.9%, and a spurious-free stopband frequency ratio of >1.8:1 with a stopband rejection over 34 dB at 10.5–16.4 GHz.

The authors realised the first E-band thulium-doped fibre (TDF) amplifier, which includes a three-stage configuration fluoride-based TDF and two filters suppressing amplified spontaneous emission in the S-band.

The inclusion of organic light emitting diodes (OLEDs) in high-end devices, such as TVs and smart-phones, along with the insertion of cameras embedded in daily use devices, provides the opportunity to establish optical camera communication (OCC) systems based on OLED emitters for Internet of Things (IoT). This Letter presents an experimental demonstration of this OLED-based OCC system for IoT. The results suggest that despite the low emitted power of OLED devices, long range links can be established based on OCC. One outdoor and two indoor scenarios are tested, validating BER below for short range and give just for long range links.

A new method of voltage application is proposed to increase the focus range of a liquid crystal lens driven by two voltages. It is possible for the optical power of the liquid crystal lens to change in a wider range, while the root mean square aberration of low level is maintained. The liquid crystal lens is driven by a frequency of the voltages different in positive and negative lens states to obtain its largest optical power in each state. An obvious increase in focus range is realised.

The authors propose and experimentally demonstrate an optical vector network analyser (OVNA) employing optical double-sideband modulation and optical Hilbert transform. The optical device-under-test is characterised with and without optical Hilbert transform. Then, by employing post signal processing, accurate spectral responses on upper and lower sides of the optical carrier are observed. A proof-of-concept experiment for measuring gain and absorption responses of the stimulated Brillouin scattering (SBS) is performed, showing that the proposed OVNA is able to accurately measure the gain and absorption responses of the SBS. The measurement error is smaller than 1%.

In the past, rotating vectors were not commonly used in most modulation strategies for matrix converters (MCs) because their angular positions always change over time, making it difficult to create a repetitive switching pattern. To make them easier to use, the relationship between active vectors and rotating vectors was analysed regarding the output voltage and input current, which are the main objectives of the MC-modulation strategy. Based on a theoretical analysis, an active vector is equivalently replaced with a set of rotating vectors. The proposed idea to apply the rotating vectors was evaluated with the conventional space vector modulation method and showed good input/output waveforms and zero common-mode voltage.

A new cascaded multilevel topology is proposed in this study. It is denominated single DC source quasi-Z-source cascaded multilevel inverter (SS qZS-CMI). The SS qZS-CMI performs at maximum power point tracking of the photovoltaic (PV) array with a single inverter module and each of the inverter modules shares an equal amount of the power of the whole system. The SS qZS-CMI has the capacity to equally regulate the peak voltage on each inverter module by means of the dual control loop of the main inverter module and the shoot-trough state of the auxiliary inverter modules, ensuring the symmetry of the cascaded multilevel inverter. Experimental results verify the proposed topology.

Forward-looking radars are essential sensors for many fast moving platforms such as fighter aircraft and missiles. Despite its multiple benefits, noise radar has not yet been used for this task because of the low Doppler tolerance of its waveform and the associated need for computational intense Doppler compensation. This Letter presents the very first experimental results of a continuous-emission noise radar whose antennas are oriented along the direction of proper motion of a car operating as a radar platform at a significant speed. To the best of the authors' knowledge, no experiments of this kind have been published to date. Very likely, the results presented here are the first ones showing moving target detection by a noise radar on a fast moving platform.

A reweighted nuclear norm minimisation algorithm by considering the inherent rank sparsity of the submatrix block for direction-of-arrival (DOA) estimation with unknown mutual coupling in a uniform linear array (ULA) is proposed. A novel block overcomplete dictionary is first derived by parameterising the steering vector to avoid the unknown mutual coupling effect. Then, in order to take advantage of the inherent rank sparsity of the submatrix block, a reweighted nuclear norm minimisation algorithm is introduced, and the weighted matrix can be designed by the Capon spectrum for enhancing the sparsity of scheme. Finally, the DOA estimates can be obtained by the spatial spectrum of the reconstructed matrix. Simulation results verify that the proposed method has distinct advantages over state-of-the-art methods.

LiDAR matching between real-time point clouds and pre-built points map is a popular approach to provide accurate localisation service for autonomous vehicles. However, the performance is severely deteriorated in dense traffic scenes. Unavoidably, dynamic vehicles introduce additional uncertainty to the matching result. The main cause is that the pre-built map can be blocked by the surrounding dynamic vehicles from the view of LiDAR of ego vehicle. A novel uncertainty of LiDAR matching (ULM) estimation method aided by the dynamic vehicle (DV) detection and high definition map is proposed in this Letter. Compared to the conventional Hessian matrix-based ULM estimation approach, the proposed method innovatively estimates the ULM by modelling surrounding DV. Then the authors propose to correlate the ULM with the detected DV and convergence feature of matching algorithm. From the evaluated real-data in an intersection area with dense traffic, the proposed method has exhibited the feasibility of estimating the ULM accurately.

A reverse blocking insulated gate bipolar transistor (IGBT) with partially narrow mesa (PNM) structure and trench collectors is proposed. An n-type carrier stored (n-cs) layer is introduced between two trench gates with the bulgy bottom, forming the PNM structure, which nicely shields high electric field from the n-cs layer at the forward blocking state. Hence, a highly doped n-cs layer can be used to enhance the carrier storage effect in the n-drift region at on-state to reduce the on-state voltage [V _CE(sat)]. Besides, an n-buffer layer is introduced between two trench collectors at the back side, acting as a field-stop (FS) layer at forwarding blocking state. At the reverse blocking state, the trench collectors shield high electric field from the n-buffer region and the n-cs layer acts as an FS layer, which ensures a high reverse breakdown voltage. Numerical simulation results show that the 1.4 kV proposed reverse blocking-IGBT can use a 38% thinner n-drift region and obtain a 45% lower V _CE(sat) than the conventional counterpart.

Among the various non-destructive testing and evaluation methods, infrared thermography gained its importance due to its fast, whole field, remote and quantitative evaluation capabilities for inspection of various materials. Being an optimum technique in terms of usage of low peak power heat sources in a moderate experimentation time, frequency-modulated thermal wave imaging (FMTWI) plays a vital role in the infrared thermographic community. The noise rejection capabilities of the proposed pulse compression favourable FMTWI by using principal component analysis as a post-processing technique are highlighted. The proposed scheme has been tested on a mild-steel sample having sub-surface flat bottom hole defects located inside the test specimen at various depths. It is clear from the obtained results that the reconstructed pulse (main lobe) concentrates much of the imposed energy into a narrow duration, which enhances the defect detection sensitivity and resolution in order to visualise the sub-surface defects with higher signal-to-noise ratio.

This Letter proposes two methods for the Gaussian-tailed zero-memory non-linearity (GZMNL) design in symmetric -stable noise. The optimal GZMNL is designed by maximising the efficacy via a derivative-free method. The efficient GZMNL is designed by polynomial fitting with the derived coefficients. Simulated results demonstrate that the GZMNL designs are nearly as optimal as the locally optimal detector, with an advantage of closed-form formulas.

In this Letter, the authors propose a low-complexity search method for carrier frequency offset (CFO) estimation in generalised frequency division multiplexing (GFDM). The proposed technique does not have any limitations on CFO acquisition range while providing an accurate estimate. Compared with the existing solutions in the literature with the lowest complexity, the proposed technique brings at least an order of magnitude complexity reduction without any performance penalty. Finally, the numerical results and comparisons with the existing literature in terms of performance and complexity attest the efficacy of the proposed method.