A single-cell tri-band composite right/left-handed (CRLH) resonant antenna is presented. The antenna is designed on a single-layer coplanar waveguide-fed based on the T-junction discontinuity equivalent circuit. The proposed antenna provides compact size, easy fabrication process, multi-band feature and higher efficiency in comparison with the previously reported CRLH resonant antennas. The single-cell CRLH resonant antenna is fabricated and the measurements are consistent with the simulation result.

A new balun used for feeding a planar log-periodic dipole array antenna (PLPDA) is proposed. This balun, which is based on mode conversion, converts gradually the transmission mode from quasi-TEM (coaxial cable) to full TEM (balanced line), with an accessible excitation current property (ECP), which is not achievable in classical feeding techniques (infinite or self balun). To validate this approach, an ultra-wideband (UWB) PLPDA antenna was designed with these balun structures for operating in the L and S bands (1–4 GHz) and the results were compared in terms of reflection coefficient and radiation pattern. A PLPDA antenna with a mode converter balun was fabricated. Results show that the proposed balun is a low length (10 mm) frequency-independent (4.5 GHz based on |S ₁₁| ≤ −10 dB) device useful for feeding the UWB PLPDA antenna. Moreover, antenna's ECP can be modified, which is necessary for phase centre variation compensation.

Microwave-based brain stroke detection demands low profile, compact, directive and wideband antennas for efficient imaging using portable systems. Using image theory of electromagnetics, the magnetic symmetry plane of a folded antenna is utilised for its miniaturisation to half of its original volume without sacrificing its penetration capability or radiation directivity. The antenna achieves 63% fractional bandwidth (1.25–2.4 GHz) with 3.5 dBi average gain along the direction of radiation. Both the near-field and far-field radiations are characterised to ensure the antenna's applicability in the detection system. Moreover, the radiation safety is also analysed since the antenna has to operate in close proximity to the head. The overall dimensions of the proposed antenna are 0.29λ ₀ × 0.12λ ₀ × 0.06λ ₀ (λ ₀ = lowest operating wavelength).

A method for wireless data transmission from an implantable image sensor through the biological tissue as a conductive medium is reported. A battery-powered implantable image sensor unit is dipped in phosphate-buffered saline (PBS) as the simulant body solution and received the data signal through the PBS. Its equivalent circuit model for simulation is also proposed. The transmitted signal waveforms are in good agreement with the simulation result. Image recovery from the signal obtained from the implantable image sensor received outside the PBS is also demonstrated.

Low-frequency (1–20 MHz) inductive coupling is a leading technique to power biomedical implants wirelessly. Even though these are non-radiating systems, the external and the implant coils can cause sufficient radiated electric field to exceed the federal standards for communication. A multi-coil-based wireless power transfer (WPT) technique is developed to selectively control the currents in the external and implant coils to reduce the radiated electric field. A four-coil WPT system, which can achieve more than 37% reduction in the radiated electric field as compared with a two-coil WPT system with the same dimensions and design constraints, is implemented and used to demonstrate the effectiveness of the proposed approach.

A 2 GHz sub-harmonically injection-locked phase-locked loop (SILPLL) with a self-aligned injection window is presented. The SILPLL adopts a mixer-based self-align technique to automatically adjust the injection timing, overcoming the speed limitation of the phase detection. Circuit techniques such as a symmetrical mixer and a V/I converter with mismatch cancellation are adopted to improve injection timing accuracy. Fabricated in a 180 nm CMOS technology, the SILPLL exhibits −127 dBc/Hz phase noise at 1 MHz offset and draws 6.9 mA current from a 1.8 V power supply. The measured root-mean-square jitter integrating from 1 kHz to 40 MHz is 214 fs and the reference spur is −61 dBc.

A new architecture for a multi-band voltage-controlled oscillator (VCO) phase-locked loop (PLL) is presented, which can continuously maintain the optimum VCO sub-band and the VCO control voltage even with severe supply voltage variation. A prototype 2 GHz PLL with the new architecture is successfully demonstrated in 0.13 µm CMOS technology.

A CMOS Blumlein generator including a two-stacked-metal–oxide semiconductor field effect transistor high-voltage switch, a Blumlein pulse-forming-line (PFL) network and an on-chip Klophenstein taper is presented. The Klophenstein taper is used to convert the high-impedance load to 50 Ω. The Blumlein pulse generator is fabricated in a 0.13 µm CMOS process. Gaussian-like pulses of 725 mV peak-to-peak amplitude, ∼126 ps duration and 3.18 GHz bandwidth are measured on a 50 Ω load (including the losses of the probes, cables and pads). After de-embedding the connection system and Klophenstein taper, the pulses of 1.88 V and 114 ps duration are obtained, which are higher than the rated operating power supply of the given process.

The response of a neural circuit submitted to a bi-chromatic stimulus and corrupted by noise is investigated. In the presence of noise, when the spike firing of the circuit is analysed, a frequency not present at the circuit input appears. For a given range of noise intensities, it is shown that this ghost frequency is almost exclusively present in the interspike interval distribution. This phenomenon is for the first time shown experimentally in a FitzHugh–Nagumo circuit.

A novel resistorless bipolar junction transistor (BJT) bias and curvature compensation circuit for ultra-low-power CMOS bandgap voltage references (BGRs) is introduced. It works in the nanoampere current consumption range and under 1 V of power supply. The analytical behaviour of the circuit is described and simulation results for a 0.18 μm CMOS standard process are analysed. A junction voltage of 550 mV at room temperature is obtained (at an emitter current of 3.5 nA), presenting an almost linear temperature dependence, whereas the power consumption of the whole circuit is 3.4 nW under a 0.8 V power supply at 27°C. The estimated silicon area is 0.00135 mm².

Brushless motors are used in many applications owing to their advantages. In most of the applications, conventional control methods with a hysteresis band (HB) controller are used. An improved speed and current control scheme for brushless DC motors with a trapezoidal shape back EMF is presented. Instead of the conventional HB and proportional-integral (PI) controllers, a fuzzy logic controller that is independent of motor equations and based on expert knowledge has been employed for current and also for speed regulation; thereby, the disadvantages of the HB are eliminated and the overall performance of the controller is enhanced. Experimental studies have been carried out with a TMS320F2812 digital signal processor. The presented control scheme has been validated through comparative experiments with the fuzzy logic speed and HB current controller. The results show that the proposed control scheme operates satisfactorily and provides a constant switching frequency.

Pairwise classification is a task which predicts whether two samples belong to the same class or not. Boosting provides a way of combining many weak classifiers to produce a strong one and has been regarded as one of the most successful classification methodologies. The problem of pairwise classification is addressed by boosting decision stumps, the simplest weak classifier. Based on gentle AdaBoost, pairwise gentle AdaBoost of decision stumps is proposed to do pairwise classification. To make the classifier deal with a pair of inputs, sample-weighted linear discriminant analysis (LDA) is proposed, which is tailored to boosting the framework. For pairwise classification, the proposed algorithm shows better performance than traditional boosting of decision stumps on two UCI data sets.

A method combing simple linear and nonlinear filters is proposed for real-time ‘non-uniformity correction’ of infrared video sequences, which suppresses ‘ghosting’ artefacts due to both lack of motion and strong edges. In this ‘least mean square’ (LMS)-based method, a mean filter is used first, when the ‘fixed pattern noise’ (FPN) level is high, taking advantage of its noise smoothing capability. When the FPN level drops to a low level, a sigma filter is used instead to reduce edge smearing. The sigma filter is also used to detect abnormal pixels like dead pixels and pixels contaminated by impulse noise, in addition to adaptive adjustment of the learning rate, with no extra cost. Experiments with simulated data and real infrared sequences show that the proposed method outperforms several other LMS methods. It is of the same computational complexity as Scribner's method, which makes it a good candidate for real-time hardware implementation.

Proposed is a max-margin discriminative random fields model for multimodal human action recognition. To incorporate multiple modalities for joint modelling, a specific graphical structure with parallel sequential observations and related hidden-state layers is designed. Moreover, the corresponding potential functions for model formulation are designed. For model learning, the max-margin learning method is proposed to discover both latent correlation among multimodal data and temporal context within individual modality. A comparison experiment shows that the proposed model can boost the performance of human action recognition by taking advantage of complementary characteristics from multiple modalities.

Recently developed two-dimensional (2D) histogram specification (HS) methods can improve the contrast in an image by increasing the pixel-value differences among the neighbouring pixels. However, the conventional 2D HS methods use the 1D cumulative distribution function (CDF) for pixel-value mapping, and thus the 2D histogram of an output image largely deviates from a 2D target histogram. A 2D CDF-based pairwise pixel-value mapping method is presented. Experimental results show that the 2D histogram of the output image obtained by the proposed method well approximates the 2D target histogram.

Dynamic spectrum leasing provides a room for better spectrum utilisation. Primary user (PU) temporary leases its unused spectrum in exchange for monetary profits and secondary users (SUs) opportunistically use the available spectrum. Recent works proposed a non-cooperative game to model PU–SU interaction, but the outcome is not Pareto efficient. The SUs’ utility is modified to indicate their opportunistic behaviour. In addition, the outcome of the non-cooperative game would be Pareto efficient. Simulation results verify that the proposed method outperforms the conventional one.

An integrated circuit for a physical unclonable function (PUF) to generate an identifier for each device is proposed based on the via formation probability. The via hole size is determined to be smaller than that specified by the design rule which guarantees successful via formation. As a result, a via is formed with a certain probability. A proper via hole size and a post-processing method are found to obtain very high randomness in the bit sequences, and it is confirmed that the bit error rate is zero through repeated measurements over one year under the supply voltage variations with noises and in a wide range of temperature. This time invariance of bits can be attributed to the fact that the via formation does not change over time, once they are formed.

An augmented radial basis function neural network (ARBFNN) is proposed for modelling and linearising a wideband Doherty power amplifier (DPA) with strong memory effects and static nonlinearity. To evaluate the performance of the ARBFNN, a 51 dBm DPA and a 25 MHz mixed test signal were used in modelling and linearisation measurement. Compared with the memory polynomial (MP) model and the real-valued time-delay neural network (RVTDNN), the ARBFNN is highly effective, leading to 3 and 5 dB improvements in the normalised mean square error. More importantly, the ARBFNN predistorter represents a significant improvement over the RVTDNN and MP in the suppression of the out-of-band spectral regrowth. In addition, the ARBFNN has a similar linearisation capability as the generalised MP model, but has much better numerical stability.

A compact wideband bandpass filter design that provides a sharp roll-off and low loss performance is presented. The design is based on employing two separate paths between the adjacent resonators constituting the filter, where one path enables electric coupling and the other one magnetic coupling. This structure generates zeros in the filter's transmission response due to the dominance of electric coupling. A fourth-order filter was fabricated and tested. Correlation between the simulated and measured results is remarkable. The measured results confirm the fractional bandwidth of 70% at the centre frequency of 4.58 GHz, insertion-loss is <0.55 dB with a return-loss better than 15 dB in the passband. The filter occupies an area of 4.1 × 17.8 mm, which corresponds to 0.1λ _o × 0.44λ _o, where λ _o is the wavelength of the centre frequency.

A 245 GHz sensor system for gas spectroscopy is presented, which includes an integrated SiGe transmitter (TX) and receiver (RX), and a 0.6 m-long gas absorption cell between the TX and RX modules. The integrated local oscillators (LOs) of TX and RX chips are controlled by two external phase-locked loops (PLLs), whose reference frequencies are swept with constant frequency offset for a low intermediate frequency of the RX. The RX consists of a differential low-noise amplifier, an integrated 122 GHz LO with a 1/64 divider, a 90° differential hybrid and an active subharmonic mixer. The TX consists of an integrated 122 GHz LO with a 1/64 divider, and a frequency doubler. The TX and RX chips are fabricated in 0.13 µm SiGe BiCMOS technology with f _T/f _max of 300 GHz/500 GHz. Using external dielectric lenses for the TX and RX modules, the gas absorption spectra were measured for acetonitrile and methanol.

A technique that can increase the bandwidth of a regulated cascode (RGC) transimpedance amplifier (TIA) is presented. The proposed new RGC input topology with a positive amplifier and an inductor enables the control of the input impedance of the circuit and isolates the large input parasitic capacitance. Using this mode the dominant pole of the input node is shifted up to a high frequency, and then the bandwidth of the CMOS RGC TIA is improved. To verify the feasibility of the proposed technique, the CMOS RGC TIA is implemented using a 1P6M 0.18 µm RF CMOS technology. The 3 dB bandwidth of 4.98 GHz is measured in the presence of a 0.5 pF photodiode capacitance with a 1.8 V supply voltage. The measured result shows that the bandwidth of the TIA can increase by about 2 GHz larger than that with the conventional RGC configuration.

It has been reported that subharmonic instability exists in the V ² controlled buck converter in continuous conduction mode (CCM) when the duty ratio is above 0.5 (D > 0.5) and no stability issue exists in discontinuous conduction mode (DCM) when 0 < D < 1. However, the CCM and DCM instabilities will appear in the safe duty ratio range when the equivalent series resistance (ESR) of the output capacitor is below a critical value. The critical output-capacitor ESR for the stabiiity of a V ² controlled buck converter in CCM and DCM is investigated by using the discrete-time model and verified by circuit simulations.

The potential use of sheet-like waveguide-based wireless energy transfer systems for a two-way power and/or high-data rate transmission–reception through thick metal walls is investigated. The proposed system utilises the near-field evanescent wave concept to transmit energy and/or data. Compared with wireless energy transfer devices based on magnetic resonance coupling, the transmission efficiency of the investigated method is not affected by the shielding offered by metal sheets. For the investigation, a 150 × 150 mm² sheet-like waveguide transmitter and a cylindrical wave-cavity receiver were designed. The receiver achieved a transfer efficiency of −3.76 dB, while transmitting power through galvanised iron sheets. The prime objective of this investigation was to sustain the power and/or data transfer efficiency despite the variation of the obstructing metal wall thicknesses from 1.5 to 22.5 mm. The proposed system was designed to operate at 25 MHz, enabling 50 Mbit/s data transfer using a function generator.

A unique approach is proposed to realise dopingless impact-ionisation MOS (dopingless-IMOS) using work-function engineering of electrodes to form charge plasma as a substitute for doping. Using appropriate work-function difference between metal electrodes and intrinsic silicon film with thickness less than the intrinsic Debye length (L _D) leads to charge plasma formation, in turn this charge plasma induces uniform p-type source and n-type drain. The proposed dopingless-IMOS is simulated using two-dimensional technology computer-aided design (TCAD) Sentaurus to compare its electrical characteristics with conventional IMOS having identical device dimensions and bias conditions. Dopingless-IMOS (DL-IMOS) offers a simple fabrication process flow as it avoids the need for ion implantation, photo masking and complicated thermal budget. Hence, it is highly immune to process variations, doping control issues and random dopant fluctuations and it retains the inherent distinctive advantages of IMOS.

A speech enhancement method is presented, which applies a soft mask to a target speech output of spatial filtering, such as conventional beamforming or independent component analysis (ICA). In contrast to conventional methods using either outputs or filters estimated by spatial filtering, the mask is constructed by exploiting both local output signal-to-noise ratio (SNR) and spatial selectivity obtained from the directivity pattern of the estimated filters. Experiments were conducted for both ICA and minimum power distortionless response beamforming as spatial filtering in order to demonstrate that the described mask estimation is not a tuned method for particular preprocessing. The results in terms of both SNR with a retained speech ratio and word accuracy in speech recognition show that the described method can effectively suppress residual noise in the target speech output of spatial filtering.

A compact reconfigurable CMOS low-noise amplifier (LNA) is presented for applications in DCS1800, UMTS, WLAN-b/g and Bluetooth standards. The proposed LNA features first a current reuse shunt-feedback amplifier for wideband input matching, low-noise figure and small area. Secondly, a cascode amplifier with a tunable active LC resonator is added for high gain and continuous tuning of bands. Fabricated in a 0.13 μm CMOS process, the measured results show >20 dB power gain, <3.5 dB noise figure in the frequency range of 1.8–2.4 GHz, return losses S ₁₁ and S ₂₂ lower than −12 and −14 dB, respectively, with a moderate IIP3 of −11.8 dBm at 2.4 GHz. It consumes 9.6 mW from a 1.2 V supply voltage, while occupying an active silicon area of only 0.052 mm².

A novel distributed range-free localisation algorithm to accurately locate regular nodes in wireless sensor networks is presented. Existing methods reduce the possible area where a regular node would be located by using anchors (i.e. nodes with known locations) only. These methods are effective only when anchor density is high. In contrast, the work presented here reduces the possible area of a regular node iteratively, using both anchor node locations and the possible areas of its neighbouring and non-neighbouring ‘regular’ nodes from early iteration. This allows a significant improvement in the localisation accuracy. Simulations verify the advantage of the proposed algorithm and show that the localisation error can be controlled within 15% with off-the-shelf devices.

Large-scale multiple-input multiple-output (MIMO) with high spectrum and energy efficiency is a very promising key technology for future 5G wireless communications. For large-scale MIMO systems, accurate channel state information (CSI) acquisition is a challenging problem, especially when each user has to distinguish and estimate numerous channels coming from a large number of transmit antennas in the downlink. Unlike the conventional orthogonal pilots whose pilot overhead prohibitively increases with the number of transmit antennas, a spectrum-efficient superimposed pilot design for downlink large-scale MIMO scenarios is proposed, where frequency-domain pilots of different transmit antennas occupy completely the same subcarriers in the frequency domain. Meanwhile, spatial–temporal common sparsity of large-scale MIMO channels motivates us to exploit the emerging theory of structured compressive sensing (CS) for reliable MIMO channel estimation, which is realised by the proposed structured subspace pursuit (SSP) algorithm to simultaneously recover multiple channels with low pilot overhead. Simulation results demonstrate that the proposed scheme performs well and can approach the performance bound.

By exploiting the inherent nature of the H.264 encoded video streams, a practically optimal scheme that is devised to address the video-cognizant link adaptation problem in the IEEE 802.11ac downlink multi-user (DL-MU) is proposed. To maximise the received video quality of the downlink users, it selects and allocates the modulation and coding scheme and power in a manner of minimising the expected video distortion caused by transmission failures. The objectives and constraints are set such that the users gain max–min fair opportunities to perceive, at least, the required video quality. The proposed scheme is analysed, and its efficiency is validated by experiments with respect to fairness and video quality compared with the video-oblivious methods.