Microwave filters and multiplexers are used in many application areas and have been studied for decades. However, with increasing demands on communications and radar systems more complex filters are required which not only have superior performance but also are required to be small and lightweight. This thesis looks at new techniques in microwave filter design to achieve these aims. Coupled resonator circuits are of importance for design of RF/microwave narrow-band filters with any type of resonator regardless its physical structure. The coupling matrix is used to represent the coupled resonator circuit. Each matrix entry value refers to a physical dimension of the circuit. The response of the circuit can also be calculated by using the coupling matrix. Different methods are developed to generate the coupling matrix. This thesis presents designs of the coupled resonator based diplexers and multiplexers by using the coupling matrix local optimisation technique. The design procedures and measurement performance of 3 X-band (8.2-12.4 GHz) rectangular waveguide circuits, including a 10th order diplexer, a 4th order diplexer with cross-couplings and a 4-channel multiplexer, are presented. A novel computer-aided physical structure tuning technique, called Step Tune method, is also presented in this thesis.

Novel designs of metamaterial structures as well as novel techniques and configurations for tuning metamaterials are presented in this PhD thesis. The proposed tuning techniques overcome the challenges that exist in other tuning techniques available thus far. Moreover, possible applications of tunable metamaterials in communication systems are proposed. Initially, tunable Electromagnetic Band-Gap (EBG) structures are proposed for low frequencies operation (3GHz to 6GHz) employing a novel biasing technique for varactor diodes. Subsequently, the proposed tunable EBG structures are applied to closely spaced antennas, achieving isolation enhancement for MIMO systems. Moreover, a new technique of tuning High Impedance Surface (HIS) structures is presented, with low-loss performance and no parasitic effects, based on employing two types of piezoelectric actuators, each type being suitable to a different frequency band. Particularly, bender piezoelectric actuators are used for configurations operating at low mm-wave frequencies (~15GHz) and stack multilayer actuators for operation at higher mm-wave frequencies (60GHz) where achieving a low loss performance is even more challenging. Two tunable antenna designs are also proposed incorporating both tunable HIS structures (at 15GHz and 60GHz). Finally, novel configurations of tunable Frequency Selective Surfaces (FSS) are proposed based on the concept of piezoelectric actuators to obtain a tunable response. Again, this is carried out for two operating bands, 15GHz and 60GHz. The particular designs of tunable HIS and FSS, are directly scalable to even higher frequencies (THz), offering a promising solution at this band.

This thesis summarises the research done on the feasibility of detecting and automatically classifying wind turbine faults using a short-range radar. Two main areas are included in the thesis: the theoretical and experimental analysis of wind turbine blade radar signatures in the near-field and the classification of wind turbine structural faults using machine learning algorithms. In the former, a new theoretical framework has been developed which extends the current far-field models and includes a mathematical and experimental analysis of simple flat blades as well as complex curved blades. The latter area comprises the analysis of the experimental results obtained using faulty wind turbine blades and methods of classifying these faults. This last task has been done in time and frequency domains using, respectively, the signals Statistical Parameters and the Principal Component Analysis algorithm for features extraction. The classification bas been performed employing the k-Nearest Neighbours algorithm. Finally, an Artificial Neural Network has been used as a more powerful classification tool in both domains.

Since the publication of research in the mid-1980s describing the formation of freeform graphene there has been an enormous growth in interest in the material. Graphene is of interest to the semiconductor industry because of the high electron mobility exhibited by the material and, as it is planar, it is compatible with silicon technology. When patterned into nanoribbons graphene can be made into regions that are semiconducting or conducting and even into entire circuits. Graphene nanoribbons can also be used to form the channel of a MOSFET. This thesis describes numerical simulations undertaken on devices formed from graphene. The energy band structure of graphene and graphene nanoribbons is obtained using nearest-neighbour and third nearest-neighbour interactions within a tight binding model. A comparison of the current-voltage characteristics of MOS structures formed on graphene nanoribbons and carbon nanotubes suggests that the nanoribbon devices may be better for switching applications. Conductivities of graphene nanoribbons and junctions formed from them were obtained using a nonequilibrium Green’s function formulation. The effects of defects and strain on these systems were also studied using this technique. Advancements were made when the self-energies used within the nonequilibrium Green’s function were obtained from an iterative scheme including third nearestneighbour interactions. An important result of this work is that accurate simulations of graphene based devices should include third nearest-neighbour interactions within the tight binding model of the energy band structure.

In the majority of microwave receiving and transmitting systems, a requirement is to have a filter immediately adjacent to the antenna or antenna array. This thesis presents a new methodology for antenna design where a filter is either fully or partially integrated with the antenna elements. The design of this antenna-filter follows the well-established coupled-resonator filter design theory, in which each resonator can not only be used as a filter element but also as a radiator. In order to verify the concept, a two-port bandpass filter designed using dipole antennas is the first work in this thesis to verify the use of dipole antennas as resonators. The coupling matrix has been used to obtain the filter response. One port antenna-filters made out of one, two and three dipoles. The method has also been utilised to implement X-band waveguide components which consist of an antenna-filter, antenna power divider and an antenna-diplexer. The calculation, simulation and measurement results are in good agreement. These proposed components has been designed, simulated, fabricated and measured. They have provided verification of the method, showing the antenna and filter theories and can be applied to miniaturise these components for use in the wireless communication and radar systems.

The paralinguistic information in a speech signal includes clues to the geographical and social background of the speaker. This thesis is concerned with automatic extraction of this information from a short segment of speech. A state-of-the-art Language Identication (ID) system, which is obtained by fusing variant of Gaussian mixture model and support vector machines, is developed and evaluated on the NIST 2003 and 2005 Language Recognition Evaluation (LRE) tasks. This system is applied to the problems of regional accent recognition for British English, and ethnic group recognition within a particular accent. We compare the results with human performance and, for accent recognition, the `text dependent' ACCDIST accent recognition measure. For the fourteen regional accents of British English in the ABI-1 corpus (good quality read speech), our language ID system achieves a recognition accuracy of 86.4%, compared with 95.18% for our best ACCDIST-based system and 58.24% for human listeners. The "Voices across Birmingham" corpus contains signicant amounts of telephone conversational speech for the two largest ethnic groups in the city of Birmingham (UK), namely the `Asian' and `White' communities. Our language ID system distinguishes between these two groups with an accuracy of 94.3% compared with 90.24% for human listeners. Although direct comparison is difficult, it seems that our language ID system performs much better on the standard twelve class NIST 2003 Language Recognition Evaluation task or the two class ethnic group recognition task than on the fourteen class regional accent recognition task. We conclude that automatic accent recognition is a challenging task for speech technology, and that the use of natural conversational speech may be advantageous for these types of paralinguistic task. One issue with conventional approaches to language ID that use high-order Gaussian Mixture Models (GMMs) and high-dimensional feature vectors is the amount of computing power that they require. Currently, multi-core Graphics Processing Units (GPUs)provide a possible solution at very little cost. In this thesis we also explore the application of GPUs to speech signal and pattern processing, using language ID as a vehicle to demonstrate their benefits. Realisation of the full potential of GPUs requires both effective coding of predetermined algorithms, and, in cases where there is a choice, selection of the algorithm or technique for a specific function that is most able to exploit the properties of the GPU. We demonstrate these principles using the NIST LRE 2003 task, which involves processing over 600 hours of speech. We focus on two parts of the system, namely the acoustic classifier, which is based on a 2048 component GMM, and the acoustic feature extraction process. In the case of the latter we compare a conventional FFT-based analysis with an FIR filter bank, both in terms of their ability to exploit the GPU architecture and language ID performance. With no increase in error rate our GPU based system, with an FIR-based front-end, completes the full NIST LRE 2003 task in 16 hours, compared with 180 hours for the more conventional FFT-based system on a standard CPU (a speed up factor of more than 11).

Power systems are under greater stress today due to the rapid growing demand and market-oriented activities. Operation of the existing system networks is gradually approaching their transmission limits and this raises a lot of stability problems which could potentially result in series consequences. The advent of FACTS provides new solutions to the reinforcement of the existing networks. Furthermore, the integration of FACTS also creates additional opportunities for the enhancement of system dynamic stability. This thesis presents the robust damping control of power systems with FACTS for the purpose of improving system small-signal dynamic stability. A Novel BMI-based methodology is proposed for the design of robust FACTS damping controllers. Different from most of the existing method, the proposed method is capable of managing multiple control objectives under several preselected operating points which could guarantee controller robustness in a broader range. The generality and feasibility of the proposed method is validated by controller designs on a two-area four-generator system and a five-area 16-generator 68-bus system with different FACTS devices. As an extension of the proposed BMI-based method, a coordinated design approach for multiple FACTS damping controllers is developed to address the damping problem with respect to multiple dominant oscillatory modes in large interconnected power systems. To reduce the adverse interactions between different FACTS devices, multiple SISO controllers are designed in a sequential manner with cautiously selected feedback signals. The coordinated design approach is then applied on a five-area 16-generator 68-bus system with an SVC and a TCSC to evaluate its effectiveness.

The thesis describes novel processing routes that have been developed to fabricate neural interfaces. A process has been investigated that uses microfabrication techniques to fabricate a multi-channel regenerative implant that can record nerve impulses in the peripheral nervous system (PNS), called the Spiral Peripheral Nerve Interface (SPNI). It is shown both theoretically and experimentally that the implant improves the ability to record signals in the PNS via micro-channels that act as axonal amplifiers. New processing routes are introduced to create robust interconnections from the SPNI to external electronics via ‘Microflex’ technology. To incorporate the new interconnection technology the SPNI had to be modified. During this modification the strain in the device was given specific consideration, for which a new bending model is presented. Modelling is used to show that electrochemical impedance spectroscopy can be used to assess the quality of the fabrication process. Electrochemical and mechanical tests show that the interconnection technology is suitable for a neural interfaces but the fabrication of perfectly sealed micro-channels was not evident. Thus, the SPNI was further improved by the introduction of a silicone sealing layer in the construction of the micro-channel array that was implemented using a novel adhesive bonding technique.

This thesis presents work on antennas and propagation for WBANs at 60 GHz. First, a compact, wearable Vivaldi antenna and Vivaldi antenna array covering the whole unlicensed band from 57 to 64 GHz are proposed to overcome the shadowing due to human movements. Second, multi-hop channels for on-body communication at 60 GHz are investigated through applying the proposed antennas, and it is found that multi-hop wireless network adaption can increase reliability when the separation between sensors exceeds 40 cm compared to single-hop. Radiation pattern diversity is selected among different diversity techniques for providing stable links for 60 GHz WBANs. Results show that radiation pattern diversity enlarges the signal coverage area on the human body, which compensates for the narrow beamwidth antenna, thus more stable links can be established. Finally, a representative channel model for 60 GHz on-body network with an appropriate power control method is presented. With this model and power control method, it has been proved in this thesis that a multi-hop method for on-body communication at 60 GHz is feasible to establish a stable network for different applications.

The problem of fault diagnosis under partial observation is a complex problem; and the challenge to solve this problem is to find a compromise between the space complexity and time complexity. The classic method to solve the problem is by constructing an automaton called a diagnoser. This method suffers from the state explosion problem which limits its application to large systems. In this thesis, the problem of fault diagnosis in partially observed discrete event systems is addressed. We assume that the system is modelled by Petri nets having no cycle of unobservable transitions. The class of labelled Petri nets is also considered with both bounded and unbounded cases. We propose a novel approach for fault diagnosis using the Integer Fourier-Motzkin Elimination method. The main idea is to reduce the problem of constructing the diagnoser to a problem of projecting between two spaces. In other words, we first obtain a set of inequalities derived from the state equation of Petri nets. Then, the elimination method is used to drop the variables corresponding to the unobservable transitions and we design two sets of inequalities in variables representing the observable transitions. One set ensures that the fault has occurred, whereas the other ensures that fault has not occurred. Given these two sets, we have proved that the occurrences of faults can be decided as any other diagnoser can do. The obtained result are extended to diagnose violations of constraints such as service level agreement and Quality of Service, which is of particular interested in telecommunication companies. We implement our approach and demonstrate gains in performance with respect to existing approaches on a benchmark example.