This Letter presents a 2 GHz 130 nm CMOS absorptive on–off keying (OOK) modulator. The loop matching technique is proposed to realise wideband matching such that fast modulation envelope transient can be achieved. The parallel LC network is adopted to reduce the impedance variation between the ON and OFF states of the modulator. The CMOS OOK modulator is capable of delivering the modulation data rate of 734 Mbps, leading to the data-to-carrier-ratio of 36.7%.

Inspired by the recent success of deep learning (DL) approaches in computer vision domain, this Letter proposes a framework to encode the sensor data into an image representation for the activity recognition task. The signal from sensors is encoded based on the Gramian Angular Field. The encoding technique increases the dimension of the data, captures a local temporal relationship in terms of temporal correlation between time intervals on the geometric interpretation, and can be easily applied to the pre-trained DL architecture. The proposed framework is examined with respect to six popular sensor-based activity recognition datasets. Using the authors’ framework, the results show that their approach outperforms most of the state-of-the-art approaches.

Loop closure detection is an indispensable component to reduce the error accumulation in localisation applications, however, it is hard to be implemented as a real-time application with high accuracy on the mobile platform. A place recognition method called gridding place recognition (GPR) is proposed for ultra-fast loop closure detection with high accuracy maintained. The GPR splits the images into multiple grids for high parallelism and uses bags of word features to summarise each grid. The matching results for each grid are then retrieved and merged for further determination of loop closure under extra temporal and spatial constraints. The authors implement the method on a mobile platform, and the best accuracy and fastest execution time of 47 frames per second are achieved on several public benchmarks compared with other typical or fast loop closure algorithms.

A novel joints relation-reasoning, graph convolutional network (JRR-GCN) is proposed to solve the problem of skeleton-based action recognition (SAR). Different from the conventional spatial convolutional network-based methods, the adjacency matrices of JRR-GCN is reasoned by joints relation-reasoning network (JRR) automatically, which results in generating a more realistic representation of skeleton topology and yields better adjacency matrices for every sample. JRR is trained with the reinforcement learning with a novel state-action mapping scheme. Extensive experiments are conducted on two public SAR datasets, NTU-RGB+D and kinetics. Also the obtained results demonstrate the effectiveness of JRR-GCN comparing with the state-of-the-art SAR methods.

All test patterns are decoded in the conventional Chase-2 algorithm, but only a few candidate codewords are derived. In this work, a highly efficient Chase-2 decoding algorithm has been proposed, where only the test patterns that can be decoded into distinct candidate codewords are processed. It means that the total number of decoding attempts is reduced from that of the test patterns to that of the candidate codewords, which consequently reduces the complexity, especially in high signal-to-noise ratio (SNR) region. The criterion of the test pattern selection is based on the property of minimum Hamming distance, it can be expected that the proposed algorithm can be applied to all linear block codes without performance degradation compared to the conventional Chase-2 algorithm.

In this Letter, the authors propose a novel convolutional neural network (CNN)-based estimation of the distance between an ultra-wideband (UWB) transmitter and receiver for a localisation. By exploiting the UWB signal characteristics, such as high-resolution in the time domain, the CNN is designed. The proposed CNN-based method estimates the distance from only the received signals, without signal-to-noise ratio information which is used for the conventional time of arrival (TOA)-based methods. Furthermore, as verified by simulation, the proposed CNN-based method significantly outperforms the conventional TOA-based method with respect to the distance estimation accuracy.

In this Letter, a magnetic near-field probe aimed for improvement of the spatial resolution of radiated susceptibility mapping over the 80–3000 MHz frequency range is presented. The magnetic field simulation of the standard and the improved probe was performed and the measurement results for both probes are presented. The spatial resolution improvement of up to 47% was shown by modelling. The measured resolution improvement of 38.5% confirmed the enhanced localisation of susceptible RF PCB areas using the proposed probe.

Microacoustic resonators made on suspended continuous membranes of LiNbO₃ were recently shown to have very strong coupling and low losses at ≥5 GHz, suitable for high-performance filter design. Employing these simple resonator structures, the authors have designed, fabricated, and measured a 4.7 GHz bandpass ladder-type filter having 1 dB mid-band loss and 600 MHz bandwidth to address the 5G Band n79 requirements. The filter is fabricated on a monolithic substrate using standard i-line optical lithography and standard semiconductor processing methods for membrane release, starting with commercially available ion-sliced wafers having 400 nm thickness crystalline LiNbO₃ layers. The filter is well-matched to a 50 Ω network and does not require external matching elements. Through accurate resonator engineering using our finite element method software filter design environment, the passband is spurious-free, and the filter provides better-than 30 dB rejection to the adjacent WiFi frequencies. This filter demonstrates the performance and scalable technology required for high-volume manufacturing of microacoustic filters >3.5 GHz.

This Letter presents the closed-form expressions of the unloaded Q-factor in metallic cavities with equilateral triangular cross section, computed from its analytical resonant mode solutions. The obtained analytical formulas for the Q-factor extend the very restricted range of problems with closed-form solutions, complementing the classic cases of the rectangular and circular cavities found in the literature. The presented derivation provides extremely rapid reference solutions for diverse applications such as filter design by test cases for microwave characterisation systems or numerical full-wave solvers. A universal Q-chart, valid for any aspect ratio, is given and the achieved analytical results for the first resonant modes are tested with numerical commercial software, showing excellent agreement.

A new design of single-ended-to-balanced (SETB) filtering power divider (FPD) with wideband common-mode (CM) suppression is presented. To effectively realise high suppression of CM noise, a hybrid structure constructed by both the double-side parallel-strip line (DSPSL) and a half-mode defected ground structure (HMDGS) serving as the mid-inserted conductor plane are adopted. Besides, due to the effect of HMDGS, the rejection of second harmonics for the differential-mode (DM) transmission is also improved. Meanwhile, by employing two pairs of dual-mode resonators symmetrically coupled with a stepped-impedance input feed line and adding a DSPSL stub, wideband-filtering performance with high selectivity is achieved. For validation, a prototype of the proposed SETB-FPD centred at 5.9 GHz with a 3-dB fractional bandwidth (FBW) of 27% is designed, fabricated and measured. Results indicate that the proposed SETB FPD exhibits good performances of high selectivity, wide upper stop band, good isolation between two balanced output ports and 30-dB suppression of CM noise over the DM pass band.

The authors examine a resonant tunnelling diode (RTD)-based wireless transceiver system in conjunction with radio-over-fibre (RoF) and identify the performance limiting factors. Ultra high-definition television 4K-video over 1 km fibre and 15 cm wireless is demonstrated. In addition error-free data transmission up to 11 Gbit/s over 1 km fibre and 10 cm wireless is also presented. Eye-diagram and jitter results for laboratory RoF equipment show the performance is not degraded with respect to an RTD-only system, and that RoF-RTD is considered promising and complementary technologies.

The authors obtained very good back-to-back bit error rate performance in up to 80 Gbaud DP-16QAM operation by using their coherent driver modulator of a 3-dB electro-optic bandwidth of over 50 GHz. To achieve high bandwidth and good signal integrity, they used double ultra-low loop wires between the driver IC and the modulator and a differential capacitively loaded travelling-wave RF electrode based on straight coupled coplanar waveguide with a ground-signal–signal-ground configuration for their InP modulator. As far as they know, this is the highest bandwidth and baud rate operation so far for the high-bandwidth coherent driver modulator configuration.

When multiple ground-based differential interferometric radar (GB-DInRad) systems are utilised together to resolve three-dimensional (3D) deformation, the resolving accuracy is related with the measurement geometry of these radar systems. This Letter focuses on how to determine an optimal geometry. Based on the resolving equation of the 3D deformation, the Geometric Dilution of Precision (GDOP) is defined. Then the minimum value of the GDOP is calculated and a detailed geometrical analysis is made to determine the geometrical conditions of the minimum GDOP. The results prove that the optimal geometry could be obtained when the unit measurement vectors of these radar systems construct a regular pyramid with a specific structure. A numerical simulation is utilised to validate the analysis conclusion.

Non-filament 3D vertical RRAM (VRRAM) is a promising technology for emerging high-density memory applications. In this Letter, the layer-dependent resistance variability at both low resistance state (LRS) and high resistance state (HRS) is assessed on state-of-the-art 8-layer 3D VRRAMs. 2048 devices are measured with the aid of an FPGA-controlled relay matrix which enables automated switching of devices without changing the cabling. The results show LRS exhibits little layer dependence while both the average and the standard variation of HRS decrease from the top layer to the bottom layer. A qualitative speculation is proposed to explain the observation based on the high-resolution transmission electronic microscope image. This work is beneficial for future 3D VRRAM circuit design and performance improvement.

In this letter, the authors propose block normalised iterative hard thresholding (BNIHT) algorithm for the recovery of block sparse signal, in which the non-zero elements are presented in clusters. Based on block restricted isometry property, the sufficient conditions to guarantee the convergence of BNIHT are derived. In addition, the number of required iterations is obtained. The simulation experiment shows that BNIHT algorithm is superior to the block IHT (BIHT) algorithm when the step size satisfies .

Block orthogonal matching pursuit is an efficient reconstruction algorithm in compressed sensing, which exploits block sparsity during support index selection. In this letter, to further improve the performance, the authors propose two block sparse reconstruction algorithms by incorporating the prior information of block support probability. Based on Gamma distribution approximation, such information is formulated as an additive term during index selection. Moreover, the second algorithm extends the first one to the scenario with inaccurate prior information by introducing an additional judging mechanism with block correlation and prior factor simultaneously. Numerical results show that the proposed algorithms outperform existing algorithms.

In array signal processing, most existing direction-of-arrival (DOA) estimation methods fail to work given deficient snapshots. To solve this problem, this Letter proposes a novel wideband DOA estimation technique by utilising the sparse representation in terms of the low rank Toeplitz structure presented in uniform array geometries, followed by an enhanced MUSIC method. Specifically, each sub-band segment is first focused into the reference frequency band to collect the coherent covariance matrix even in the lack of sufficient samples. Then, the covariance matrix is denoised through a structure-based sparse reconstruction, which exploits the low rank Toeplitz structure. Finally, the DOAs is efficiently estimated via an enhanced MUSIC method. Simulation result confirms the efficacy of the proposed method that works well in MIMO and radar applications.

In frequency-division-duplex (FDD) massive multiple-input multiple-output (MIMO) systems, noisy feedback is a constant challenge for the base station (BS) to acquire accurate downlink channel state information (CSI). In this Letter, the authors propose a convolutional neural network (CNN)-based approach to overcome this problem, which they refer to it as an anti-noise CSI acquisition network (ANCAN). Results demonstrate that ANCAN can reconstruct CSI more accurately than other emerging CSI acquisition methods in the presence of noisy feedback links.

With the huge growth of the Internet of Things (IoT) in manufacturing, agricultural and numerous other applications, connectivity solutions have become increasingly important especially for those covering wide remote area in the scale of kilometre squares. Although many low-power wide-area network (LPWAN) technologies such as Long Range are supposed to support long-range low-power wireless communication, the underneath star topology limits the scalability of the networks due to the need of a central hub. To provide connectivity to a wider area, the authors propose to build the mesh topology upon these LPWAN technologies. One of the challenges of meshing these networks is the routing mechanism originally designed for star networks is not energy sensitive. In this Letter, the authors address this issue by proposing a distributed as well as energy-efficient reinforcement learning based routing algorithm for the wide area wireless mesh IoT networks. They evaluate the failure rate, spectrum and power efficiencies of the proposed algorithm by simulations, which resemble the long-range IoT networks, by comparing it to that of a random routing with loop-detection algorithm and a centralised pre-programmed routing algorithm which represents the ideal scenario. They also present a progressive study to demonstrate how the learning in the algorithm reduces the power consumption of the entire network.