A time–frequency audio feature extraction scheme is proposed, in which features are decomposed from a frequency–time–scale tensor. The tensor, derived from a weight vector and a Gabor dictionary in sparse coding, represents the frequency, time centre and scale of transient time–frequency components with different dimensions. The distinguishing Gabor atoms are represented by individual tensor elements, and their associated coding weights are represented by tensor element values. The experimental results of sound effects classification showed performance improvement against that of sparse coding features.

An architecture and its digital implementation using field programmable gate arrays of an electronic cochlea with an acoustic fovea and address event representation output are presented. The system is experimentally characterised with simple harmonic signals and employed to process signals from a micro-Doppler sonar.

A new structure single-stage dynamic comparator with a large input common-mode range is proposed. The proposed comparator is compared with previous dynamic comparators. With same size input transistors and load capacitance, it is more than 1.2 times faster with <80% power consumption. Also, the input-referred noise and offset are no more than the previous comparators.

A new robust harmonic lock detector (HLD) suitable for a wide-range delay-locked loop (DLL) is presented. This detector is composed of some delay indicators measuring the total delay of the DLL in real time, and it can detect the harmonic lock for a wide frequency range close to 1–20 times higher than the minimum frequency. Owing to its seamless operation without any initialising process using external reset control, it can ensure reliable lock detection despite some unusual conditions, such as frequency change, electrical surge input or power-ground fluctuation. The proposed HLD and the DLL were designed in 0.13 µm CMOS technology, and the simulation results showed the exact harmonic lock detection and reliable DLL locking properties over the whole frequency range from 25 to 500 MHz of the input clock.

A methodology to find the optimised switching sequence for gradient error compensation in the current cell array of current-steered digital-to-analogue converters (DACs) is presented. This new approach is simplified to be non-iterative and generalised to both linear and quadratic gradient errors. Simulation results show that the approach finds the optimised switching sequence to substantially reduce the nonlinearity of DACs due to the gradient errors.

New results and properties of recently proposed autonomous memristive circuits with mixed-mode oscillations (MMOs) comprising both large and small amplitude oscillations (LAOs and SAOs) are presented. Certain important properties of these circuits, such as period estimation, determination of variables’ values when SAOs occur and an extension to memristive MMOs circuits with zero initial conditions are examined.

A novel active contour model (ACM) for image segmentation, driven by both local and global image-intensity information encoded by a self-organising map (SOM), is proposed. Experimental results demonstrate the robustness of the proposed model to the contour initialisation and to the additive noise, when compared with the state-of-the-art local and global ACMs. They also demonstrate its robustness to scene changes.

A non-local selection scheme of regularisation parameters is exploited for non-local total variation (NLTV) regularisation, based on a content-aware function that can determine regularisation parameters through discriminating image contents, with the aid of a non-local structure tensor calculated using the weights from the original NLTV. Experimental results on simulated and real noisy images show the relatively good performance of the proposed scheme based on NLTV, in terms of visual improvement and peak signal-to-noise ratio value.

A new family of almost multiphase sequences (AMPSs) with perfect periodic autocorrelation function (PACF) is offered. The obtained sequences are based on Chu sequences and differ from them by the structure, the peak-factor which is nearly equal to one, and by fewer phases for the same periods.

An enhanced round-robin scheduling (RRS) scheme for multi-pair multi-input–multi-output relay systems is proposed, where the relay is equipped with a large-scale antenna array and performs minimum mean-square error (MMSE) data processing. Compared with the traditional RRS, significant sum rate performance improvement by setting the number of active user pairs to an appropriate value is achieved.

Given the significant concerns regarding carbon emissions from fossil fuels, global warming and energy crises, the renewable distributed energy resources (DERs) are having to be integrated in the smart grid (SG). The SG can spread intelligence of the energy distribution and control system from the central unit to long-distance remote areas, thus enabling accurate state estimation (SE) and wide-area real-time monitoring of these intermittent energy sources. In contrast to the traditional methods of SE, a novel approach for SG SE is proposed based on concatenated coding structures, where the grid state is treated as a dynamic outer code and the recursive systematic convolutional code is seen as a concatenated inner code for protecting and redundancy in the system state. Furthermore, Kalman filter (KF)-based online DERs SE and a discrete-time linear quadratic regulation method are proposed to control these state deviations for the SG and the Internet of Things (IoTs) communication network, which can exploit the far-reaching connectivity and privacy of DERs messages. Simulation results show that the proposed algorithm achieves 3 dB performance improvement compared with the existing KF.

The large dynamic range of multicarrier signals make them very sensitive to quantisation effects. Owing to the non-continuous nature of quantisation curves, conventional analytical methods used to characterise the distortion effects of quantisation signals are very complex, and can and even have convergence and/or accuracy problems. A simple and accurate analytical method for performance evaluation of quantisation effects on sampled multicarrier signals is presented.

A problem of multi-keyword search in a structured peer-to-peer (P2P) distributed computing system is considered. Methods have been developed to employ term-set indexing in a P2P system. Such an approach is an attempt to avoid excessive communication cost incurred by intersection operations in a single-term indexing method. In addition to limiting the maximum term-set size, index pruning was proposed to avoid exponential growth of the term-set index size. However, to obtain a global search result, query decomposition and query result intersection operations are still needed when the number of query terms exceeds the maximum term-set size. With index pruning employed, how a multi-term query is decomposed and affects the quality of retrieval results in terms of recall and precision is shown. A near-optimal query decomposition method is proposed to alleviate such a problem. Results from the simulation demonstrate the improvement gained by the proposed method.

A new approach to time delay estimation for spread spectrum signals which is based on a covariance-based fitting criterion is introduced. Since the conventional correlation method cannot resolve the spread spectrum signals that are separated less than a pseudo-noise chip apart, a further process may be necessary for correlation outputs to increase the resolution capability. To this end, a sparse covariance-based estimation algorithm that was originally proposed for direction-of-arrival estimation is adopted and its superiority is verified, especially in correlated signal environments.

A novel framework for automated design space exploration (DSE) of architecture and unrolling factor (UF) for perfectly nested loops in architectural synthesis (AS) is presented. The sub-contributions are that it: (a) proposes a novel model for determining execution time based on architecture and UFs for nested loop without tiresomely unrolling completely; (b) proposes a novel methodology for exploration of architecture and loop UFs for a perfectly nested loop, based on swarm intelligence (SI); and (c) maintains a trade-off between power and performance during exploration and also resolves orthogonal issues such as exploration speed and quality of result (QoR). Results of the proposed approach indicate an average improvement in QoR of >33% and reduction in runtime of >34% compared with recent approaches.

A novel pulse-induction (PI) metal detector (MD) is proposed to decrease the saturation time of the output voltage and increase the detection depth. The proposed MD utilises a receiver with a time-domain bucking circuit. The bucking circuit decreases the output voltage significantly without a target, so that the output voltage can be amplified more before becoming saturated. To demonstrate the performance of the proposed MD, it has been fabricated and measured. The measured output voltage shows that the saturation time is reduced and that the detection depth increases in comparison with a conventional PI MD.

A novel low-power injection-locked frequency divider (ILFD) is proposed, in which a complementary G_m -boosted Colpitts oscillator core is employed to improve negative transconductance (G_m ) and the current-reuse technique using a single transformer is applied to reduce the power consumption. The proposed ILFD exhibits a self-resonant frequency of 4.93 GHz with a maximum locking range of 870 MHz at 0 dBm of input power. The phase noises of the input and output signals are −128.54 and −134.53 dBc/Hz, respectively. The ILFD consumes 1.37 and 2.5 mW at 1.15 and 1.3 V, respectively. The chip area is 715.5 × 468 μm² in a 0.13 μm CMOS process. To the authors’ knowledge, this is the first demonstration of the current-reuse ILFD.

The design of a thin microwave absorber which exhibits a −10 dB reflectivity bandwidth of 108% at normal incidence and 16% for simultaneous suppression of TE and TM polarised waves over the angular range 0–45° is presented. The structure consists of a 3 mm-thick metal backed frequency selective surface (FSS) with four resistively loaded hexagonal loop elements in each unit cell. The surface resistivity and width of the loops are carefully chosen to maximise the bandwidth by merging the reflection nulls that are generated by the multi-resonant absorber. Measurement and simulation results are in good agreement over the broad frequency range 7.8–24 GHz.

Open-loop resonators (OLRs) are loaded with one or two stubs to design a compact bandpass filter (BPF) with a wide stopband by creating extra transmission zeros. By properly selecting the length of the stubs and their positions, extra transmission zeros are created in the stopband before the spurious passband, thus giving wide stopbands. Mathematical analyses are performed to show the effect of open stubs on the OLR. A two-stub-loaded BPF is designed and fabricated, which shows a harmonic suppression of up to 3.2f_o by creating five transmission zeros and 44% size reduction as compared with a simple OLR-based filter without stubs. Measured results are in good agreement with the theoretical analysis and simulation results.

A quad-band filter with compact size and high skirt selectivity utilising stub-loaded nested dual-open loop resonators is presented. By loading open-ended stubs at the centre of the two nested resonators, two odd modes as well as two even modes can be independently adjusted to realise a flexible quad-band filtering response. Moreover, both a pair of zero-degree feed lines and the frequency transformation method are adopted to introduce multiple transmission zeroes, which rightly enhance the out-of-band selectivity. According to the topology circuit design and coupling matrix synthesis, a quad-band filter operating at 1.58/1.81/2.50/2.66 GHz is finally designed and fabricated.

The use of a fluorescent fibre (15.5 mm in length) pumped in a forward direction with a 520 nm emitting LED for fast wavelength conversion to 560 nm is reported. In addition to the fact that this wavelength matches the attenuation minima of poly-methyl-methacrylate polymer optical fibres (POFs), the presented results show an external conversion efficiency of 4.7% and a bandwidth of 52.5 MHz. The device is a simple, inexpensive, compact, eye-safe and lightweight optical source at 560 nm. It can be suitable for single-channel ‘long-distance’ (>400 m) or WDM-based links using POFs, presenting a bandwidth–distance product of 50–100 MHz/100 m.

Optical microresonant mode switching in a hexagonal GaN microdisk, which acts as a whispering gallery mode (WGM) and quasi-WGM (QWGM), has been demonstrated by a change in the condition of the reflection surface under a room temperature photoluminescence (RT-PL) measurement. It has been confirmed that only the WGM in the microdisk, which uses all side walls of a microdisk as the reflection surface, was strongly affected by attaching another microdisk on a side wall. The result indicates that the WGM can be controlled independently by external factors.

The differential modal gain (DMG) in a few-mode erbium-doped fibre amplifier (FM-EDFA) is successfully improved by employing a ring-core FM erbium-doped fibre (RC-EDF). The DMGs between LP₀₁ and LP₁₁ mode signals in the FM-EDFA with the RC-EDF are 1.8 and 1.6 dB for LP₀₁ and LP₁₁ mode pumping, respectively, whereas those in a FM-EDFA with a conventional step-index FM EDF are 9.2 and 6.1 dB, indicating 7.4 and 4.5 dB DMG improvement for LP₀₁ and LP₁₁ mode pumping, respectively.

A novel planar wireless power transfer (WPT) system using strongly coupled magnetic resonance is proposed. Additional layers of printed spiral coils (PSCs) are applied to the transmitter/receiver resonator of the 4-coil planar WPT system to improve the quality factor and transmission efficiency. In addition, by connecting multilayer transmitter/receiver resonators together using conductive shorting walls, the transmission efficiency is further improved. Five planar WPT systems using coplanar PSCs and printed loops have been investigated. The measured results show that the proposed design with two additional layers of resonators and shorting walls could improve the transmission efficiency from 77.27 to 84.38%.

A new constant-envelope modulation multiplexing two quadrature phase shift keying (QPSK)-like global navigation satellite system (GNSS) signals and its generalised form with unbalanced power are presented. The modulations can be regarded as the extended form of generalised DualQPSK modulations and provide a solution to the Compass B1 band multiplexing problem.

An optimal order selecting scheme for 4-weighted fractional Fourier transform (4-WFRFT) over doubly selective channels is developed. First, the expression for carrier-to-interference ratio (CIR) is deduced and then through maximising the CIR the optimal order factor is obtained. Through selecting the optimal order factor of 4-WFRFT, this system can match the doubly selective channel characteristics through switching between orthogonal frequency division multiplexing (OFDM) and single carrier (SC) systems according to obtained channel state information. The simulation results show that the optimal order 4-WFRFT scheme can improve the performance over doubly selective channels with respect to the traditional SC or OFDM.

Time synchronisation plays an important role in time division multiple access (TDMA) wireless sensor networks (WSNs). Existing time synchronisation methods suffer from cumulative time error and the imbalance of time synchronisation precision, and those factors bring fluctuations to nodes’ time and will become worse with the increase of hops, inducing more unstable communication links for sensor nodes. A dual time synchronisation method (DTSM) for WSNs is introduced, which is endowed with the maintenance of two running time sources per node, one for its parent and the other for its sons. This method eliminates the time fluctuations among nodes especially for the nodes with greater hops, so as to obtain more stable links for large-scale networks. This method is implemented on the SCSC-RFA1 platform and the results show its validity.

A fully integrated power amplifier using a power cell switching technique, implemented in 65 nm CMOS technology is presented. The main objective of the proposed architecture is to significantly improve the efficiency at high power back-off. To do so, distributed active transformers are used as the splitter, the combiner and the DC bias feed to partition the power requirements among the parallelised power cells. An individual cell can be dynamically turned ON/OFF according to the desired output power. At 2.5 GHz, the measured maximum output power is 28.2 dBm and the power-added efficiency is improved for low level, +3.2 and +4.9% for 18 and 23.7 dBm, respectively.

A novel highly flexible mixer-based ultra-wideband (UWB) pulse generator is designed. The baseband pulse generator has a controllable pulse width between 200 ps and 8 ns, while the output amplitude only varies over 80 mV in this entire range. The mixer is adjustable so that the voltage controlled oscillator leakage can be controlled and minimised. The proposed architecture gives a highly flexible UWB pulse generator for which the characteristics can be adjusted to fit the specifications of different applications and the spectral mask of the country/region of usage.

IEEE 802.11-based wireless ad hoc networks are well suited for critical scenarios such as rescue operations during natural calamities, military applications on battle fields and so on where the existing infrastructure fails. In these networks, since nodes have limited battery power, the reduction in packet collision is very important, otherwise the nodes have to spend more energy for packet transmission. In these wireless networks, the packet transmission procedure plays a pivotal role in performance. Several packet transmission procedures represented with Markov chain models have been suggested by the authors to improve the network performance. However, these procedures have not been suitable for dense networks due to the high collision rate. A new packet transmission procedure for the IEEE 802.11 distributed coordination function (DCF) protocol is proposed to reduce the number of collisions and enhance the throughput of wireless networks. A mathematical framework is carried out for transmission probability and performance analysis is presented showing that the new procedure greatly enhances the network performance and as a result can be applied in the design and performance optimisation of wireless network protocols.

The effect of packet size differences, that has not been taken into account in existing papers on the IEEE 802.11 distributed coordinated function (DCF) and 802.11e enhanced distributed coordinated function (EDCF), has been investigated, and improved throughput equations that lead to accurate performance representation in erroneous wireless channels are provided. Wi-Fi devices that use the DCF and the EDCF reduce the risk of data packet transmission failure using request to send (RTS) and clear to send (CTS) packets by removing the hidden node problem. However, as the bit error rate (BER) increases (especially near the signal coverage edge of the other communicating device), even though RTS/CTS packets may succeed in communication, data packets may fail due to packet errors based on their larger packet sizes. This phenomenon is confirmed through actual measurements and mathematical performance analysis is carried out to accurately express the performance of 802.11-based communications in erroneous conditions.