A comparator circuit with a built-in programmable threshold is proposed. The threshold is embedded by providing different initial source voltages to each of the input transistors. This source-voltage shifting is done by means of a tunable capacitance, which can synthesise a very wide range of initial voltages, even voltages beyond the supply. This results in a power-efficient comparator structure with a very wide programmable tuning range and low decision delay. The inherently small integrating capacitance of the structure allows for high speed operation even when using small device sizes, minimising input capacitance and maximising efficiency. A practical transistor-level implementation is simulated in a standard 16 nm CMOS technology on a 0.8 V supply, with a tuning range of at least 568 mV.

A digital high-accuracy interface for sine–cosine shaft encoders (SCE) is proffered. The outputs of SCE vary as a sine and cosine functions of shaft angle. The proposed interface processes these outputs and renders a linear direct-digital output for 360° operating range. The interface has minimal dependence on many circuit non-idealities including the phase error of quadrature oscillator. The performance of the interface is verified using simulation and emulation-based studies. Furthermore, the interface is tested with a magneto-resistance-based SCE. Test results show that the output non-linearity remains around 0.27% for typical values of phase errors. The interface also possesses features such as low-component count, high linearity, direct-digital output etc.

High-speed comparators are extensively used in serial link receiver designs. Some comparator architectures can show significant hysteresis that degrade the sensitivity of the receiver, increasing the bit error rate. In this Letter, a comparator hysteresis compensation strategy that re-uses the first tap of a decision feedback equaliser to shift the comparator input voltage, increasing the decision margin is proposed. An updated equaliser coefficient adaptation scheme is also introduced. The proposed technique can be used for binary and multi-level modulations.

A novel energy-efficient radiation hardened by design 10T static RAM (SRAM) cell is proposed. The parasitic extracted simulations show that by employing the proposed 10T-SRAM cell, an average improvement of ∼ 29, 5/10%, and 108/129%, in layout area, write/read access time (WAT/RAT), and write/read static noise margin (WSNM/RSNM), respectively, is obtained over the recently reported 10T-SRAM cell at a supply voltage 0.4 V in STMicroelectronics 65 nm technology. The proposed SRAM cell at 32 nm technology node using technology computer-aided design mixed-mode simulations is also validated. In 32 nm technology, the proposed SRAM cell shows 42/125% and 54/8%, in WSNM/RSNM and WAT/RAT, respectively, better results as compared with 10T SRAM cell at a supply voltage 0.3 V. In 32 nm technology, the proposed SRAM cell can mitigate the impact of heavy-ion strike with a linear energy transfer of 30 MeV cm²/mg.

A set of design strategies for the tunable matching network are proposed, which are performed on the impedance domain rather than the reflection coefficient domain (Smith Chart). Moreover, the bandwidth and parasitic parameters of the network are discussed. The proposed methods guarantee the finding of optimal point and accelerate the convergence speed from O(N) to O(log N).

This Letter presents a method of predicting the phase noise limits of static frequency dividers. The proposed phase noise model for free-running and injection-locked modes is based on the circuit parameters only. The model provides an intuitive comprehension of the phase noise behaviour of static frequency dividers and is proved to be useful in primary hand calculations. The method was verified against measurement results of a static frequency divider-by-two, operated at 77 GHz and fabricated in a SiGe process.

A trapezoidal recursive convolution (TRC) technique is one of the simple ways to incorporate dispersive models into the finite-difference time-domain (FDTD) method. The authors newly develop the TRC–FDTD method for the analysis of arbitrary-shaped dispersive materials using a dispersive contour-path algorithm. The accuracy of the present method is found to be the same as that using the Z-transform technique with reduced computational requirements.

An illumination-invariant method for background subtraction is presented. Inspired by the observation that the degree of dispersion in the colour space, which is computed from the local region of each frame, is not severely changed even with varying illuminations, To exploit the colour coherence as the features for formulating the background model is proposed. Experimental results on various benchmark databases show the efficiency and robustness of the proposed method compared to the previous approaches introduced in the literature.

A dynamic loss (DL) is proposed for the one-stage object detectors in computer vision, as an improved version of the focal loss in the literature. The proposed loss features a second-order item which can efficiently scale the conventional loss during training. A gradient update approach is then presented to employ the DL in the mainstream one-stage YOLO-V2 object detector. Experimental results shows that for the PASCAL VOC 2010 dataset, the mean average precision of the YOLO-v2 detector with the proposed DL is 88.51%, which is 2.6, 1.17 and 0.29% higher than that using the conventional mean square error loss, cross entropy loss and naive focal loss. Compared to the focal loss, the DL increases the training convergence speed of the YOLO-v2 detector by two times.

This Letter presents the quaternion conjugate symmetric-sequency ordered complex Hadamard transform (QCS-SCHT). The purpose of this research is to present the idea of the quaternion in sequency domain that is based on four complexified terms including real and imaginary values are discussed and developed with different types of quaternions planes for QCS-SCHT. QCS-SCHT is discussed for colour image processing.

An effective method for the design of the continuous-time all-pass network for the low-pass filter group delay response equalisation is presented in this Letter. Since the method is derived from the maximally flat conditions at the origin of the group delay response's rational function, the cascade connection of the designed all-pass network and low-pass filter exhibits a constant group delay response at the origin in the flat sense. Illustrative examples regarding group delay equalisation of the Butterworth filters are given.

A new common magnetic coupling structure has been proposed. A high selective microstrip bandpass filter and a diplexer with the common magnetic coupling and 0° feeding have been developed. A pair of transmission zeros can be generated by 0° feeding, while an additional transmission zero can be produced by the common magnetic coupling structure and can be controlled/adjusted by the electric/magnetic couplings. The proposed diplexer has been fabricated and measured, and the measurement shows a channel insertion loss of <1.3 dB, and an isolation of not <34 dB. The measured results agree with the prediction.

In this Letter, a flip-chip bonding based on ink-jet printing technology is proposed and verified by interconnecting a separated coplanar waveguide (CPW) samples. Before the flip-chip bonding process, diameter and height of ink-jet printed silver bumps for flip-chip bonding are measured to be about 300 and 13 µm, respectively. In a frequency range from 300 kHz to 3 GHz, return loss and insertion loss of the flip-chip bonded CPW samples are measured to be higher than 20 dB and lower than 1.8 dB, respectively, where the length of the signal line is 35 mm. There are negligible discrepancies between radio frequency (RF) characteristics of the flip-chip bonded CPW and those of a conventional CPW. These results show that the proposed flip-chip bonding can be used to integrate RF components into electronic devices with low-loss using simple procedures.

A semi-implicit finite-difference time-domain thin-wire (TW) scheme with a relaxed stability condition is proposed for modelling effective surface plasmons in waveguide metamaterials. Its formulation is based on introducing the Newmark-Beta method into the TW formulation based on the contour-path integral of Maxwell's equations around a wire. The resultant scheme is magnetically implicit. Its stability, accuracy and efficiency are assessed in the application to a wire-added parallel-plate waveguide perturbed with a small gap.

A W-band dual-band waveguide band-pass filter composed of two dual-mode (TE₁₀₁ and TE₂₀₁) cavities and four single-mode (TE₁₀₁) ones is presented. This dual-band filter with each passband of fourth-order Chebyshev response is synthesised and designed with the centre frequencies 85.35 and 101.45 GHz, as well as the 3 dB fractional bandwidths of 5.3 and 2.1% by adopting the dual-mode resonators and the controllable coupling technique. The prototype is fabricated by the conventional computer numerical control milling technology easily. The measured results are very close to the simulations in full W-band.

A completely new design for an all-silicon all-optical pass switch has been proposed based on the stimulated Raman scattering (SRS) in silicon nanocrystal embedded slotted photonic crystal waveguide (SPCW). Substantial miniaturisation in the device footprint and operating power has been attained by merging the benefits of the giant SRS gain and the ultra-high optical confinement of the SPCW. The proposed switch offers a tremendously high contrast ratio of between the logic levels and an extraordinary bit rate of 125 Gbps.

The authors used 65-nm CMOS technology to develop a linear four-channel driver IC with low temperature dependence and ultra-low power dissipation for 64-Gbaud coherent optical transmitters. The driver showed more than a 48-GHz 3-dB electrical bandwidth and less than 1-W power consumption in four-channel operation. By employing a circuit that suppresses the temperature dependence, they achieved 3-dB electrical bandwidth variation of 3.0 GHz and the gain variation of 1.5 dB under the −5 to 75°C and ±5% supply voltage variation conditions. The CMOS driver has all the necessary functions for a high-bandwidth coherent driver modulator such as a gain control, peaking control, peak detection and temperature monitoring, all of which functions can be controlled by a serial peripheral interface. A fabricated sub-assembly consisting of the CMOS driver and an InP modulator showed a 48-GHz 3-dB electro-optic bandwidth.

The silicon carbide bipolar junction transistor needs large transient currents supplied into and out of its base terminal for rapid switching. To realise this, it is normally desirable to have a base driver circuit supply rail at a high voltage. However, the device also needs a steady base current to hold it in the on-state. Supplying this current from a high-voltage source is inefficient. A circuit is presented that applies high transient base–emitter voltages, but with low driver circuit power consumption.

For improved discrimination of a ballistic warhead from a decoy, a simple 3D feature vector (3DFV) is proposed that is composed of the fundamental frequency, the bandwidth, and the sinusoidal moment of the time–frequency image. Compared to two existing methods, the 3DFV was more efficient and gave results that were less sensitive to noise.

Target localisation methods using time difference of arrival measurements have to handle the correlated measurement noises caused by the common reference sensor. The noise correlation, as well as the reference sensor selection, impacts the Cramer–Rao lower bound and the target tracking result. This Letter proposes two theorems to analyse the effects of reference sensor selection theoretically. Simulation results validate the correctness of the proposed theorems.

The training overhead in pursuing pencil beams is commonly heavy for millimetre-wave (mmWave) communications, particularly in multi-user (MU) scenarios with conventional sequential-training methods. To reduce the training overhead, a parallel beam alignment algorithm is proposed for the MU mmWave system with a hybrid structure. The proposed algorithm jointly exploits parallel beam alignment among different user-subsets and hierarchical beam training for each user-subset. Consequently, the training overhead is significantly reduced to for the mmWave systems with transmit antennas. Numerical results verify that the proposed algorithm outperforms traditional alternatives in MU scenarios.

The authors formulate the closed-form expressions of the generalised moment generating function (G-MGF) for the extreme distribution, which enables one to calculate important metrics of wireless communications systems. The derived formula is utilised to evaluate the performance of communication systems under extreme fading channels, such as energy detection in terms of area under the receiver operating characteristic curve and outage probability in interference limited scenarios.

The impact of hybrid automatic repeat request (HARQ) retransmission on Voice over Long Term Evolution (VoLTE) cell capacity is investigated in a single LTE cell. An analytical model is proposed to estimate the cell capacity in terms of number of users. The effect of HARQ on cell capacity is investigated and it provides solution to derive transmission parameters in a given radio condition to estimate VoLTE cell capacity. This model may be further useful while estimating call blocking performances and to design efficient Call Admission Control for next generation cellular mobile networks.

The working frequency of wireless communication systems will inevitably be developed for the higher-frequency terahertz (THz) band. An effective way to improve data transmission rates in this band is to adopt higher-order modulation. However, existing THz wireless communication systems struggle to realise higher-order modulation and demodulation at high speeds. Baseband digital modulation mixes signals with the same modulation and symbol rate. Different bit streams are sent on the 0.34 THz band and the receiver uses a per-survivor processing algorithm to separate and demodulate the mixed signal. This doubles the transmission rate of THz data whilst avoiding the problem of modulation and demodulation at high speeds resulting from higher-order modulation. Both simulation and measured data is used to verify the feasibility of the method.

In massive multiple-input multiple-output (MIMO) systems with hundreds of antennas at the base station (BS) and few users with a single transmitter antenna, linear detectors, such as zero-forcing (ZF), achieve near-optimum performance due to the property of asymptotically orthogonal channel matrix. However, how far away is ZF from the optimum performance. To answer this question, in this Letter, closed-form bit error rate (BER) expressions are derived. These BER expressions are subsequently used to evaluate the performance difference between ZF and the optimum detector, which is a function of the number of antennas at the BS and the number of users. Numerical results verify the tightness of the performance difference.