An acoustic echo cancellation (AEC) algorithm based on minimising the mutual information between the loudspeaker and system output signals over a sliding discrete Fourier transform (DFT) window, for single AEC parameter estimation, is introduced. Unlike the conventional least-mean-square (LMS) systems, the proposed algorithm requires no double-talk detection (DTD) and its AEC parameter can be continually updated. Although it has been shown that independent component analysis (ICA) allows continual adaptation of the AEC parameters under DTD, current ICA-based algorithms estimate a filter of the same length as that of the LMS techniques. The sliding DFT window is utilised to facilitate adaptation of only one AEC parameter for deflation of the far-end signal, thereby greatly reducing the computational load.

For high-resolution synthetic aperture radar polarimetry applications, a compact broadband dual-polarised antenna element with high port isolation and low cross-polarisation is investigated. The size of the element is 19 × 19 × 40 mm, which is much smaller than the traditional dual-polarised horn. The measured results show that the operating frequencies of both the ports are used from 8.2 to 11.0 GHz, about 30% centred about 9.6 GHz, the port isolation level is below −40 dB and the cross-polarisation is below −35 dB. The radiation patterns and the gain of the element are presented.

A novel wireless sensor for the detection of food quality is presented. The main idea is to transform radio frequency identification (RFID) tags into RFID sensors, owing to a specific design of the tag antenna. From knowledge of the variation of the permittivity of food over time through experimental characterisation, it is possible to detect the time from which the food becomes improper for consumption based on the read-range measurement of the designed sensing tags. This low-cost ultra-high frequency (UHF) RFID passive sensor was designed and experimentally tested on plastic-film-wrapped beef meat. The overall agreement between the experimental and simulation results shows the potential of this technique for real-world applications in food traceability.

A new antenna structure with broadband and high gain at the millimetre wave bands is presented. This antenna consists of a conventional patch fabricated on a thick low-temperature cofired ceramic (LTCC) substrate. By introducing an inner ground plane surrounding the main patch, the propagation of the surface wave is prevented, and antenna gain is increased. Furthermore, the bandwidth of this antenna becomes wider due to the coupling between the main patch resonator and the newly introduced dielectric resonator. The simulated results of the proposed antenna show a bandwidth and peak antenna gain at 60 GHz of 12 GHz and 9.4 dBi, respectively.

A novel dual-polarised ultra-wideband patch antenna is proposed. This antenna consists of a square ring patch and four inclined triangular-shaped patches. The vertices of the triangular patches are connected to the microstrip lines which are printed on the bottom layer of the grounded FR4 substrate. To realise the differential feed mechanism for reducing the coupling between the two polarisation ports and the cross-polarisation level, two identical tapered baluns are used to excite the antenna. The antenna prototype can provide two orthogonal polarisations with an impedance bandwidth (|S ₁₁| ≤ −10 dB) of 113.2% and isolation of over 28 dB across the whole operating frequency band. The measured antenna gain varies from 2.09 to 7.66 dBi for port 1 and from 0.02 to 6.77 dBi for port 2 with a 3 dB gain bandwidth of 96.3%.

A gain enhanced planar Yagi-Uda antenna with pattern reconfiguration is presented. The prototype provides an omnidirectional pattern as well as a high-gain directive pattern, which can be reconfigured on two orthogonal planes. For the first time, the principle of a Yagi-Uda antenna combined with pattern reconfiguration is implemented without complexity on a planar prototype. The pattern reconfiguration is achieved by alternately switching the driven and reflector elements of the Yagi-Uda antenna. The prototype concept was simulated using CST Microwave Studio, fabricated on a substrate of dielectric constant 4.4 and tested using an R&S ZVB8 network analyser. The gain variation with the reflector and director parameters is also analysed and presented.

The power combining efficiency of a dual-feed single-beam power combining reflectarray is shown to be substantially improved by reducing the spacing between the reflectarray elements. This improvement has been confirmed experimentally.

A novel and compact tri-band microstrip-fed monopole antenna with dual-polarisation characteristics for wireless local area network (WLAN) and worldwide interoperability for microwave access (WiMAX) applications is proposed. The antenna is composed of a partial ground and a Y-shaped radiating patch that consists of two unequal monopole arms and a modified circle monopole. The antenna is able to generate three separate impedance bandwidths to cover the frequency bands of the WLAN (2.4–2.484 and 5.8 GHz) and the WiMAX (3.4–3.7 GHz). In addition, the antenna is circularly polarised in both the WLAN bands. Furthermore, the antenna structure is extremely simple, durable and occupies small space, making it suitable for practical applications.

A controlled accuracy approximation scheme of the sigmoid function for artificial neuron implementation based on Taylor's theorem and the Lagrange form of the error is proposed. The main advantages of the proposed solution are two: it provides a systematic way to guarantee the required accuracy and it reuses the circuitry of the linear part of the neuron to compute the sigmoid function. The sigmoid derivative is also available for artificial neural networks with online learning capabilities.

A new system for the quantification of super paramagnetic iron oxide nanoparticles (SPIONs) is presented. The proposed system relies on the particle electron paramagnetic resonance (pEPR) technique and utilises the linear response of the SPIONs to generate a pEPR signal under a static field. By extracting the pEPR signals, whose intensity is proportional to the SPIONs concentration, the quantitative information of the SPIONs is derived. To evaluate the system performance, sensitivity measurements have been conducted with two commercial SPIONs, FeraSpin M and Resovist, and reveal a measurement sensitivity of 300 ng.

This reported work demonstrates the use of convex optimisation to localise the transverse magnetic B ⁺ ₁ field in regions of interest for recently proposed multi-sectioned alternating impedance coils and the traditional transmission line coil. An approach based on different axial slices to identical radio frequency (RF) coils except upper stripline structure is investigated. Electromagnetic simulation results are compared for RF coils and discussed in detail at 7.0 T.

The quality of focusing in high-frequency ultrasound imaging is significantly affected by the length of the depth of field (DOF). A dual-concentric transducer combined with a phase-apodisation scheme is presented to achieve the extended DOF maintaining signal-to-noise ratio. As a preliminary study, computational simulation by using a Field-II program is conducted to demonstrate the feasibility. A dual-concentric transducer is composed of a disc- and a ring-type element with confocal apertures. When two input signals with 0° and 180° phases are simultaneously applied to the inner and outer elements, a bifocal-zone is generated in the axial direction. The overall − 6 dB DOF is 40% longer compared with a single element transducer. Thus, the proposed scheme can be a potential approach to increase the DOF for high-frequency ultrasound imaging.

A low-power bandwidth-improved derivative superposition (DS) mixer in a standard 90 nm CMOS technology is analysed in depth. The negative g _m″ in saturated NMOS pseudo-differential transistor is compensated by the positive g _m″ in the parasitic NPN bipolar junction transistor. Compared with the traditional DS mixer using the dual-NMOS method, the proposed DS mixer shows a wider bandwidth owing to the lower parasitic capacitance at the base. The maximum conversion gain of the DS mixer is 7.2 dB with the 1 dB frequency bandwidth from 0.5 to 6 GHz. Compared with the Gilbert-type mixer, the third-order input interception point (IIP₃) is improved by about 7.5 dB. The DS mixer consumes a DC power of 3.8 mW under 1 V supply.

A new last-level cache replacement policy for systems with a phase-change memory (PCM) main memory is presented. The proposed policy aims at reducing the write traffic to PCM by considering the fine-grained dirtiness of cache blocks when making a replacement decision. Experimental results show that the proposed policy reduces the write traffic to the PCM by 26 and 17% on average and up to 52 and 33% compared to not recently used and re-reference interval prediction, respectively.

A solution for the computation of modular inversion based on speculative addition is presented. The proposed circuit effectively addresses the problem of long carry chains caused by signed operations and allows fast low-overhead implementations of modular inversion. The solution is particularly suitable for devices providing optimised carry-propagation logic such as field-programmable gate arrays.

In the framework of the behavioural approach, a new definition of the passive circuit component is proposed. It is based on the concept of a forwards concatenation set of an admissible trajectory. The so far elusive notion of the non-dynamic circuit component is also clarified. As an example of application, the passivity and available energy of the N-dimensional resistive, the Thévenin source, is thoroughly investigated.

A new discontinuity adaptive depth upsampling method is proposed in order to obtain a high-resolution depth map. The proposed method adaptively employs both the local minimum filter and the joint bilateral filter according to the variance of the depth map. Thus, it can reduce the over-blurred depth regions along the depth discontinuities. The experimental results demonstrate the feasibility of the proposed method.

A vein recognition method using the local block pattern is presented. After the phase information of a vein image is extracted by a group of Gabor filters and local XOR pattern operators, the phase coding image is divided into some even non-overlap blocks. These local block patterns can alleviate the effect of noises and enhance the discriminative power. Each block is divided further into some sub-blocks, and Fisher's linear discriminant is applied here to reduce the dimensionality of the proposed descriptor and improve the performance at the same time. Finally, these histograms of blocks are concatenated to form the descriptor of the whole vein image. The proposed method is evaluated on the hand vein image database and the HK PolyU's database. The results show that the local block pattern strategy plays a crucial role in the vein recognition method. When the sizes of the block and the sub-block are right, the method can achieve an error equation rate of 0.88 and 0.13% on the two databases, respectively.

The statistical assessment of system performance is addressed via the efficient estimation of eye diagram parameters under process variability. Statistical information pertaining to eye parameters is reconstructed by interpolation of a reduced set of simulations. The proposed strategy shows a remarkable efficiency improvement with respect to classical blind and brute-force sampling-based methods.

Error correction codes are commonly used in memories to ensure that data are not corrupted. Single error correction double error detection (SEC-DED) codes are among the most widely used codes to protect memories. One common technique to implement SEC-DED codes is to construct a parity check matrix with odd-weight columns. This ensures that double errors have an even weight syndrome and therefore are not confused with single errors thus providing the DED feature. Recently, a technique that reduces the decoding complexity for odd-weight SEC-DED codes has been proposed. This technique can be used only for small data block sizes being the practical limit 32 bits. However, memories with 64 bits are commonly found in modern computing systems. Therefore, it would be advantageous to also reduce the decoding complexity for larger block sizes. A scheme to optimise the decoding of odd-weight SEC-DED codes with block sizes of 64 bits is presented and evaluated. The results show that the new scheme can provide significant reductions in the decoder circuitry area and delay.

A space-alignment (SA) scheme with K primary users in different cells working in the same frequency band based on K-symbol extension for cognitive transmission in a multiple-input multiple-output system is proposed. The main idea is that the secondary transmitter sends a supersymbol over the K timeslot; meanwhile, the secondary transmitter aligns its transmit directions by precoding to the kth primary channel eigenmodes unused at the kth timeslot, and at the same time adapts its transmit power to satisfy the interference-power constraints at the other K −1 primary receivers. The simulation results show that the proposed scheme achieves a higher performance compared with both the SA scheme with the single primary user and the scheme of the direct singular value decomposition of the secondary channel with interference-power constraint.

Powered two wheelers (PTWs) represent only 2% of the traffic, but 30% of all deaths on the road. European governments have made this particular point a priority for road safety. This presented study overcomes the lack of PTW traffic-analyser systems in real traffic conditions. A detection technique based on a single two-dimensional LIDAR sensor is proposed. This system is able to accurately detect and count PTWs with a single sensor in dense traffic on a highway with multiple carriageways. The method consists, firstly, in generating spatio-temporal observations of vehicles crossing under the LIDAR, then, in classifying vehicles using machine-learning techniques, and finally, in evaluating PTW traffic. Such a system has direct industrial applications since it is a non-intrusive traffic-analyser system that is reliable, easily deployable and inexpensive compared with existing industrial solutions.

A 30–50 GHz CMOS ultra-wideband (UWB) low-noise amplifier (LNA) with a flat high power gain (S ₂₁), along with a flat low-noise figure (NF) is demonstrated for the Atacama large millimetre array (ALMA) band-1 (31.3–45 GHz) system applications. The high S ₂₁ and low NF are achieved because the triple-well transistors are utilised with their respective source and body terminals connected together. Furthermore, the bandwidth extension and gain flatness is achieved due to the careful design of the inductive-peaking networks. The LNA has a measured S ₂₁ of 21.5 ± 1.5 dB, a minimum NF (NF_min) of 3.8 dB at 32.5 GHz, an average NF (NF_avg) of 4.67 dB over the range of 30–50 GHz and an input third-order intercept point (IIP3) of 0 dBm, with a DC power consumption of 20.4 mW at 1.2 V supply. The proposed LNA outperforms all the reported commercial standard CMOS Q-band LNAs, with the highest gain bandwidth product and highest IIP3 suitable for the ALMA band-1 system applications.

By using a non-isolated power combiner, the overall power efficiency can be improved by reducing the wasted power at the combined out-phased signal. To further improve the efficiency of the low-power mode, a multi-power mode out-phasing power amplifier (PA) is proposed with load modulation by switches. Compared with the conventional linear amplification using nonlinear component (LINC) transmitter, the proposed multi-level LINC improves the power-added-efficiency more than 40% for a 7 dB peak-to-average power ratio orthogonal frequency division multiplexing signal. To the authors' knowledge, the proposed LINC PA is the first multi-level LINC PA without multiple power supplies.

A three-dimensional integrated substrate integrated waveguide bandpass filter has been demonstrated by using low-temperature co-fired ceramic technology. By adopting the extracted-pole and the cross-coupling mechanisms simultaneously, four controllable finite transmission zeros outside the passband are achieved to realise high selectivity. A U-band prototype has been designed, fabricated and tested. The measured results match well with the predicted theoretical counterparts. The proposed filter has the merits of desirable insertion loss, high design freedom and good out-of-band rejection.

A novel system mitigating self-phase and cross-phase modulation effects simultaneously using optical dispersion compensation at optical nodes and multichannel single-stage digital backward propagation (DBP) is proposed. Experiments demonstrate 2.4 dB Q-factor improvements with both self-phase modulation and cross-phase modulation compensation.

4H-SiC p-n ⁻-metal photodetectors are demonstrated having high external quantum efficiency >∼40% in the deep ultraviolet spectrum between 200 and 235 nm. This improvement is attributed to the improved collection of carriers generated by deep ultraviolet photons through absorption in the depletion region of the detector.

A simple technique is developed for etching the overcladding layer of a perfluorinated graded-index polymer optical fibre by the combined use of chloroform and water. The boundary between the two liquids can mitigate the evaporation of chloroform, leading to a much lower loss than that of the conventional technique which does not employ water. The effectiveness of this method is verified by Brillouin signal measurement and infrared thermometry.

By integrating triple terahertz quantum cascade lasers (at ∼3.25 THz) monolithically into an array, a peak power of 213 mW is obtained at 10 K with a threshold current density of only 257 A/cm². The device shows distinct single-lobe far-field behaviour in the temperature range from 10 to 90 K, with a full-width at half-maximum of 36°. The highest operating temperature of the array laser is identical to that of the single ridge laser, indicating good heat dissipation design in the array. These results are promising for realising high power THz quantum cascade lasers.

Uniquely and arbitrarily shaped photonic crystal laser cavities were designed, fabricated and characterised. Room temperature lasing emitting at about 1550 nm was observed for all devices when photopumped by an 830 nm wavelength pulse laser with a pulse width of 100 ns. Continuous-wave lasing was even observed for select devices.

By fusing the radar images of a certain target obtained from multiple aspects, complementary information can be made full use of for better target descriptions and higher image quality. A fusion scheme based on non-negative matrix factorisation (NMF) is proposed. A sparsity-enhancing regularisation term is introduced into the original NMF, and the corresponding modified multiplicative update rule is derived to iteratively fuse the images. The composite image generally demonstrates enhanced feature characteristics and improved signal-to-noise ratio. The experimental results prove the validity and superiority of the proposal.

Several high-quality methods for compressing LiDAR data stored in the LAS format have evolved in recent years. They offer good compression for large datasets, but are less efficient on small data packets, which are needed in web applications. This problem is focused on by analysing two state-of-the-art implementations for large LAS datasets and then proposing improvements suitable for small data packets.

A novel reduced surface field (RESURF) AlGaN/GaN high electron mobility transistor (HEMT) with back electrode is proposed. The back electrode is electrically grounded and attached to the AlN nucleation layer after the substrate is removed. The back electrode, which attracts the electric field lines at the AlGaN/GaN interface by inducing negative charges, leads to a more uniform horizontal electric field along the channel and, hence, a significant improvement in breakdown voltage. Meanwhile, there is negligible negative impact on the ON-state resistance. Numerical simulation demonstrates a breakdown voltage of 1701 V and an ON-state resistance of 10.45 Ω mm with a gate–drain spacing of 6 μm for the proposed device, compared to a breakdown voltage of 1118 V and an ON-state resistance of 10.34 Ω mm for conventional RESURF AlGaN/GaN HEMT.

Cross-spectral analysis is a mathematical tool for extracting the power spectral density of a correlated signal from two time series in the presence of uncorrelated interfering signals. A set of conditions is demonstrated and explained where the detection of the desired signal using cross-spectral analysis fails partially or entirely in the presence of a second uncorrelated signal. Not understanding when and how this effect occurs can lead to dramatic under-reporting of the desired signal. Theoretical and simulated demonstrations of this effect are presented.

The application of probabilistic linear discriminant analysis (PLDA) to speaker adaptation for automatic speech recognition based on hidden Markov models is proposed. By expressing the set of acoustic models of each of the training speakers in a matrix and treating each column as a sample, the small sample problem that can be encountered in PLDA if only one sample is available for each training speaker is overcome. In the continuous speech recognition experiments, the performance of the PLDA based approach improves over the principal component analysis (PCA) based approach and the two-dimensional PCA based approach for adaptation data longer than 12 s.

Theoretically, constant output power and high DC-to-RF efficiency cannot be simultaneously achieved by a broadband Class E power amplifier (PA) even if the load network could provide the accurate impedance desired at each frequency point. Such a problem is solved with the presented concept of continuous Class E modes. New demands on the load and corresponding load network topology are discussed to provide a theoretical basis. To verify the feasibility, a prototype PA in TSMC 0.18 μm CMOS process is also presented. In the simulation, it exhibits a maximum output power of 18.7 dBm, a power added efficiency of 64–80% and an output power flattening of 0.5 dB in the frequency range from 360 to 610 M (52% of the central frequency). A comparison between it and a broadband parallel Class E PA implemented under the same conditions is also given.

An effective outage evaluation approach of asymmetric two-way amplify-and-forward relaying is presented. The proposed approach, characterised by generality, can be applicable to various realistic fading channels. The analytical and simulation results in Nakagami-m fading demonstrate the accuracy of the approach.

Selection-and-stay combining for multi-branch (SSTC-MB) systems is investigated. The mathematical model is first established. Based on the Markovian property of the received signal-to-noise ratio, the outage performance is analysed and the closed-form expression is obtained. Numerical results validate the theoretical analysis and show that the SSTC-MB system not only enhances outage performance but also expands the applied scope of a diversity system.