In this Letter, a high-gain sub-terahertz lens horn antenna with a metal guide is proposed. In this antenna, high gain is achieved by using a metal guide and dielectric lens. The experimental results show that the impedance bandwidth of the proposed antenna is above 200 GHz, and the measured peak gain of the antenna is 39 dBi at 240 GHz.

This Letter reports the design, fabrication and characterisation of a new differential linear driver, fabricated in the III-V Lab 0.7- emitter width indium phosphide (InP) double heterojunction bipolar transistor (DHBT) technology. Large-signal electrical characterisation shows 80-GBd symbol-rate four-level pulse amplitude (PAM-4) modulation conjugated with a driver output swing of 3-V_ppdiff and a 0.74-W power consumption. Thus resulting in a 1.22-GBd driving efficiency, the highest in over 70-GBd drivers' state-of-the-art, at that date. Accordingly, S-parameter measurements of the standalone linear driver exhibit the highest gain-bandwidth product of 556 GHz, in that current state-of-the-art.

A high-swing class-C/B VCO without tail current source has been proposed. Accurate current control and dynamic bias generation that are essential to robust start-up and class-C operation are achieved through mirroring VCO LDO current and comparing against bias current. The VCO oscillation amplitude is designed to be sufficiently large and the core transistors are compressed moderately, such that the oscillation amplitude is stable while the VCO operates in class-C mode. A prototype PMOS VCO in a 0.13 μm process demonstrated a measured VCO figure of merit of 192 dB at 4.7 GHz with 1.4 mA current and core VDD at 0.6 V.

Operational amplifiers (opamps) are often designed in stages. Distortion analysis of multistage opamps is challenging due to multiple non-linearity sources. Therefore, one effective analysis method is to consider a multistage opamp model as a cascade of condensed stages, with each stage modelled by one weakly non-linear element. Then, in order to obtain a satisfactory distortion model, accurate characterisation of the non-linearity of each stage is necessary. In this Letter, the authors propose an analytical calculation method for modelling the weak non-linearity of an amplifier stage consisting of several interconnected transistors. A stage circuit model created by this method can be used to produce analytical harmonic distortion characterisation of closed-loop circuits with peripheral elements. Numerical accuracy of the analytical method has been validated by transistor-level simulation.

This Letter presents a multi-touch detector architecture incorporating an efficient buffering scheme. Most of the power consumption in a detector is originated from the intensive use of buffers that necessitate dual-port memories. To alleviate the silicon area and the power consumption, the author scrutinises and manipulates the pattern of requests made to the buffers. More specifically, two consecutive requests are rescheduled to be processed at once, and the buffers and the surrounding circuits are modified accordingly. As a result, only one request is made to a buffer at a time, and the dual-port memories are substituted with the single-port counterparts. Implementation results demonstrate that the proposed architecture can diminish the silicon area by 36% and the power consumption by 17% compared with the state-of-the-art one.

Superparamagnetic clustering (SPC) is an unsupervised classification technique in which clusters are self-organised based on data density and mutual interaction energy. Traditional SPC algorithm uses the Swendsen–Wang Monte Carlo approximation technique to significantly reduce the search space for reasonable clustering. However, Swendsen–Wang approximation is a Markov process which limits the conventional superparamagnetic technique to process data clustering in a sequential manner. Here the authors propose a parallel approach to replace the conventional appropriation to allow the algorithm to perform clustering in parallel. One synthetic and one open-source dataset were used to validate the accuracy of this parallel approach in which comparable clustering results were obtained as compared to the conventional implementation. The parallel method has an increase of clustering speed at least 8.7 times over the conventional approach, and the larger the sample size, the more increase in speed was observed. This can be explained by the higher degree of parallelism utilised for the increased data points. In addition, a hardware architecture was proposed to implement the parallel superparamagnetic algorithm using digital electronic technologies suitable for rapid or real-time neural spike sorting.

This Letter presents an artificial magnetic conductor (AMC) with self-complementary unit cells (SCUCs). The use of SCUCs results in very high angular stability of the structure; up to an incident angle of 88° for both transverse electric and transverse magnetic polarised incident waves with no shift in resonant frequency f ₀. No other structure in the literature has zero frequency shifts throughout the whole range of incident angle variations. Another interesting feature of this structure is the gradual increase of the ±90° reflection phase bandwidth with an increase in E-field polarisation angle from φ = 0° to φ = 90° while maintaining the same f ₀. These two features make the structure unique and useful in applications where angular stability and polarisation independent f ₀ are the primary concerns.

Though detectors based on deep networks have witnessed a significant boost in pedestrian detection performance, small-scale pedestrian detection remains to be a challenging task. To this end, the authors propose a scale prediction method to eliminate the dependence of most existing detectors on pre-defined anchor boxes. Due to projective transformation, a pedestrian standing afar will appear smaller than one standing closeby. For a stationary surveillance camera, different blocks on the image correspond to the views of various depths leadings to diverse sizes of object proposals. A neural network structure is developed to empirically estimate the size of a pedestrian candidate in the pixel coordinates given its central location. Comprehensive sets of experiments on two real-world datasets demonstrate that the proposed method achieves superior performance with state-of-the-art methods. Besides, significant performance advantages on small pedestrian detection are observed.

In this Letter, the authors present a novel graph clustering-based method for crowd counting only using very limited labelled samples. Based on an intuitional observation that the distribution of low-level features of a specific scene containing the same or similar number of pedestrians are close to each other in the feature space, the authors adopt a first neighbour propagation (FNP) based clustering method to divide all unlabelled data into different groups. Next, an active sampling learning strategy that measures representativeness and diversity of the training data is used to obtain a few informative samples for annotation. Finally, the counts of those labelled informative samples are effectively propagated to predict the unlabelled samples in the constructed clusters by FNP-based clustering. The compelling results on two benchmark datasets demonstrate that the proposed method is not only effective to estimate crowd counts with very few labelled samples but also applicable to annotate a large number of unknown video frames for scene-specific crowd counting models.

Lacking enough feature expression on the shallow part of the network always hinders the object detection results by missing the position of small instances. To address this, the authors propose a selected and refined local attention module (SRLAM) for object detection. SRLAM tries to improve the feature expression of local areas by establishing a relationship graph between different channels. Inspired by the classical non-local neural networks for video classification, they present the local attention module (LAM) to more effectively use remote information. The LAM can suppress the influence of irrelevant areas for improving detection accuracy. Moreover, for further information utilisation, the weight map generated by LAM is integrated together with the original feature map and the information from the higher layer by a feature fused module. Experiments on the PASCAL VOC2007 show that the authors’ model has good detection performance, the proposed model can achieve 81.6 mAP when the size of the input image is 300 × 300, that outperforms many other mainstream detectors.

Special QPSK has been demonstrated to have superiority of spectral efficiency. However, due to its scatter of constellations and no pilots inserted, the channel estimation is a challenge. In this Letter, clustering based two-stage blind channel estimation is proposed for special QPSK transmission. For the scheme, k-means clustering is applied to reduce the complexity and improve the accuracy of channel estimation in the first stage, while frame synchronisation helps to confirm the estimated result. Simulation results show that this scheme presents small mean square error and good BER performance. Furthermore, this scheme can implement frame synchronisation and channel estimation simultaneously.

The authors analyse the effect of correlation between the forward and backward links on the capacity of backscatter communication systems. With that aim, we obtain an analytical expression for the average capacity under a correlated Rayleigh product fading channel, as well as closed-form asymptotic expressions for the high and low signal-to-noise ratio (SNR) regimes. The proposed works' results show that correlation is indeed detrimental for a fixed target SNR; contrarily to the common belief, the authors also see that correlation can be actually beneficial in some instances when a fixed power budget is considered.

This Letter presents a single-pole single-throw (SPST) switch for high-power applications in the mm-wave frequency band. A stacked-FET structure is adopted in the shunt switching cell for reducing the insertion loss and improving the linearity of the switch. To enhance the isolation, the parasitic inductance of the stacked-FET structure is resonated out by connecting a series capacitor. The number of stacked FETs in each shunt cell and the number of the shunt cells are optimised considering the trade-off among the insertion loss, isolation, and linearity. The SPST switch is fabricated in a 70-nm GaAs m-HEMT process. The measurement shows that the insertion loss is <1.1 dB, and the isolation is >25 dB over the frequency from 59 to 77 GHz. Both the insertion loss and isolation are not compressed at all until the input power reaches 19.5 dBm at 75 GHz.

A compact dual-band bandpass filter (BPF) based on parallel coupled lines and shorted stubs has been presented in this Letter. The transmission and reflection coefficients of the proposed dual-band BPF are straightforwardly derived via the input impedances using odd- and even-mode analysis methods. Four transmission poles in the two passbands and nine deep transmission zeros (TZs) in the stopband can be realised. The centre frequencies and bandwidths of the two passbands can be tuned by controlling the impedance parameters of the coupled lines and shorted stubs. For demonstration, a prototype example of this dual-band BPF with a small size is experimentally characterised. The measured results show that it has high-frequency selectivity with multiple TZs at the stopband.

Moving towards commercialisation of chipless RFID systems, there are still some pressing issues. In the tag design level, the issues are high data capacity in the order of tens of bits, compact design and the immunity to the interference which can be achieved by a cross-polar response. It is demanded to yield the highest numbers of bits in a tag ID while the occupied physical space and allocated bandwidth are minimised. Increasing the number of resonators in a conventional multi-resonator chipless tag causes a significant tag size growth. The destructive coupling effect between resonators in multi-resonator tags is another critical issue. In this Letter, to combat those issues, a multi-band resonator is proposed. The comb-shaped tags contain both co-polar and cross-polar Radar cross section (RCS) responses, hence they can be used in high reflective environments. Two samples of 14-bit and 40-bit comb-shaped tags are designed, which have physical sizes of 17 mm × 13 and 30 mm × 45 mm, respectively. The promising performance of the proposed structure, which is beyond the state of art hybrid and frequency-domain tags, shows the potential of the proposed designs for identification and authentication applications.

An optical frequency divider for dual-wavelength optical signals based on an optoelectronic oscillator (OEO) is proposed and experimentally demonstrated. The Mach–Zehnder modulator in the OEO is biased at the minimum transmission point, the OEO-based frequency divided signal originates from system noise. When the oscillation phase and amplitude conditions met, it gets stabilised at an oscillation frequency which is one-third of the frequency interval of the injected dual-wavelength optical signal. A proof-of-concept experiment is verified. With the frequency interval of the dual-wavelength optical signal ranging from 101.4 to 120.6 GHz, the RF signals ranging from 33.8 to 40.2 GHz are successfully generated. The power ripple is lower than 2.92 dB and the rejection ratio of the harmonic components is larger than 39 dB for all the conditions. The phase noise values of the generated RF signals are kept to be around −105 dBc/Hz@10 kHz.

Fractional-N phase locked loops (PLLs) offer a low-cost solution to the precision frequency ramp requirements of frequency-modulated continuous-wave (FMCW) radar. In this application, however, the infamous spurious signals generated by these PLLs have not received adequate analysis. In this Letter, the authors investigate the impact of integer boundary spurs on FMCW radar performance. They show that the offset of these spurs is swept during the course of waveform generation, and that undesirable spur chirps manifest in the beat signal as a result. These spur chirps span the radar's intermediate frequency bandwidth and produce Fresnel ripples in the range profile which reduce spurious-free dynamic range. They present simulations and measurements that demonstrate how reduction of PLL bandwidth can be used to suppress these spurii.

In this Letter, the authors propose a low-complexity estimator for the correlation coefficient based on the signed process. The introduced approximation is suitable for implementation in low-power hardware architectures. Monte Carlo simulations reveal that the proposed estimator performs comparably to the competing methods in the literature with maximum error in order of . However, the hardware implementation of the introduced method presents considerable advantages in several relevant metrics, offering more than 95% reduction in dynamic power and doubling the maximum operating frequency when compared to the reference method.

Increasingly high bandwidths are desired in modern navigation and autonomous RF ranging signal design to achieve better pseudo-noise (PN) ranging accuracy. However, fulfilling the Nyquist criterion for such high-bandwidth signals would be a challenging issue for system design. Considering that the Nyquist criterion is not a necessary condition for PN code tracking, sub-Nyquist sampling can be adopted as a feasible solution to this issue. Nevertheless, it is at the expense of an increase in thermal noise caused by tracking error. This problem is particularly prominent for tracking weak signals such as navigation signals, which limits the applicability of this sampling scenario. As far as high-precision autonomous PN ranging is concerned, however, the tracking error due to digital implementation is not negligible and even becomes the dominating factor instead of thermal noise. This means that the performance loss incurred by sub-Nyquist sampling can be eliminated under high-signal-to-noise ratio conditions. Exploiting this fact, this Letter proposes a novel PN code tracking approach based on sub-Nyquist and non-commensurate sampling, providing a solution to achieving high-ranging accuracy while breaking the bounds of the Nyquist sampling theorem without performance degradation. The approach is very attractive for precise PN ranging applications demanding miniaturised implementation.

With the widespread use of automatic speaker recognition in realistic world, it suffers a lot when there is a domain mismatch, including channel, language, distance etc. Recent research studies have introduced the adversarial-learning mechanism into deep neural networks to reduce the distribution mismatch between different domains. However, they only aligned the domain distributions between the background training and evaluation data. Few focused on the diverse distribution underlying the enrol and test data. In this Letter, the authors propose a domain adversarial siamese (DAS) network trying to eliminate the domain influence on speech representation. Specifically, they feed a network with speech pairs from the same speaker. Then a domain discriminator is introduced to capture the domain consistence or difference between pairs. Final embeddings become domain-invariant and more speaker-discriminative. A cross-channel data set is sort out from NIST speaker recognition evaluation and more experiments are conducted on AISHELL-Wake-Up-1 data set. Results show that DAS performs equally effective with typical domain adversarial methods, improving results at least on energy efficiency rating. Furthermore, it is proved to be more valid for scenarios such as unbalanced data amount and unknown domain, achieving relatively improvements.

The current transformation scenario in power distribution networks forecasts the need for alternative solutions in communication systems employed in the process of making them intelligent. As a premise, these alternatives should be easy to implement and affordable to allow their adoption in rural distribution networks, where, often, the use of technologies, such as mobile, is not available. This Letter proposes a new communication architecture with mesh topology, based on LoRa technology. The architecture adapts packet prioritisation and MIMO mechanisms to ensure the communication requirements necessary for the services. In this sense, several applications and protocols of a smart grid can coexist in the same communication system. The deployment of the architecture gave rise to a routing protocol adapted to the limitations of the radio technology used. The benefits of the proposed architecture can be observed when compared to the use of traditional low power wide area network networks and cellular technologies. The robustness of the system has been proven with average latency measurement within the acceptable range for DNP3 Modbus and NBR14522 applications. The results have shown its potential employment in the management of intelligent rural distribution systems.

A closed-form expression of the achievable secrecy rate is derived in this study, which is used as the objective function to find the optimal time allocation of channel estimation and transmission. Theory analysis and numerical results demonstrate that the maximum secrecy capacity depends on the total time budget as well as the power at the transmitter.