This thesis is concerned with the problem of broadband angle of arrival (AoA) estimation for sensor arrays. There is a rich theory of narrowband solutions to the AoA problem, which typically involves the covariance matrix of the received data and matrix factorisations such as the eigenvalue decomposition (EVD) to reach optimality in various senses. For broadband arrays, such as found in sonar, acoustics or other applications where signals do not fulfil the narrowband assumption, working with phase shifts between different signals - as sufficient in the narrowband case - does not suffice and explicit lags need to be taken into account. The required space-time covariance matrix of the data now has a lag dimension, and classical solutions such as those based on the EVD are no longer directly applicable. There are a number of existing broadband AoA techniques, which are reviewed in this thesis. These include independent frequency bin processors, where the broadband problem is split into several narrowband ones, thus loosing coherence across bins. Coherent signal subspace methods effectively apply a pre-steering by focussing matrices in the assumed directions of existing sources, and sum the narrowband covariance matrices coherently. Subsequently, classical narrowband methods can be applied. A recent auto-focussing approach dispenses with the requirement of knowing the approximate direction of sources, and calculates the focussing matrices on a data-dependent fashion. A recent parametric covariance matrix approach for broadband AoA estimation is also reviewed, and it is shown that this can only detect a single - the strongest - source. Based on a polynomial EVD (PEVD) factorisation of polynomial matrices such as created by a space-time covariance matrix emerging from a broadband problem, this thesis proposes an extension of the powerful high-resolution but narrowband multiple signal classification (MUSIC) algorithm. While narrowband MUSIC is based on an EVD to identify signal and noise subspaces, the PEVD can extract polynomial subspaces. This also requires the definition of broadband steering vectors, which are used in the proposed polynomial MUSIC (P-MUSIC) method to scan the noise-only subspace. Two different P-MUSIC versions are proposed here: a spatio-spectral P-MUSIC (SSP-MUSIC) is capable to resolve sources with respect to the AoA and frequency range, and a spatial P-MUSIC (SP-MUSIC) extracts the AoA alone. Broadband steering vectors are proposed as polynomial vectors containing fractional delay filters. For the implementation, a number of methods are reviewed and compared, including windowed sinc functions and Farrow structures. All these techniques show degraded performance as the frequency approaches half of the sampling rate. Therefore, this dissertation also proposes a highly accurate fractional delay filter implementation based on undecimated filter banks, whereby the subband signals are modulated to lower frequency ranges, where individual fractional delay filters can operate with high accuracy. For the implementation of P-MUSIC, we demonstrate that the broadband steering vector accuracy is important. We also apply different iterative PEVD algorithms belonging to the families of second order sequential best rotation (SBR2) and sequential matrix diagonalisation (SMD) algorithms. We demonstrate the SMD familly, which offers a better diagonalisation of the space-time covariance matrix, is also capable of providing a more accurate subspace decomposition than SBR2. This is evidenced by a higher resolution that can be achieved if SP-MUSIC and SSP-MUSIC are based on SMD rather than SBR2. The thesis concludes with an extensive set of simulations for both toy problems and realistic scenarios. This is to explain and highlight the operation of the P-MUSIC algorithms, but also compares their performance to other state-of-the-art broadband AoA methods. For the closest competitor, the auto-focussing approach, an analysis in a polynomial matrix framework is provided, which highlights similarities and differences to P-MUSIC. The simulations indicate that PMUSIC is a powerful and robust extension of MUSIC to the broadband case.

This thesis investigates the possibility of deploying a fibre optic network on aircraft; the network should provide 1.25 Gbps links between various end nodes throughout the aircraft. Particular attention has been paid to Wavelength Division Multiplexing (WDM) in Passive Optical Networks (PONs) and how the network is affected by the harsh environment found on aircraft. The harsh environment presents a particular challenge with respect to the operational temperature range. Over this range it is desirable to minimise any additional weight and power consumption associated with cooling network components. One technique assessed to implement WDM with minimised cooling is to use spectrum slicing to seed reflective semiconductor optical amplifiers which will be positioned at each end node. In theory these 'colourless' nodes will be temperature insensitive and capable of uncooled operation. This has been thoroughly studied and temperature ranges for uncooled operation identified with different end node components. Different cooling techniques have also been studied; one technique using a phase change material for passive cooling with no direct power consumption and the other using a thermo-electric cooler. The efficiencies of these techniques have been compared by analysing their impact on fuel burn. The use of distributed feedback lasers in a WDM PON has also been investigated and a novel method for reducing power consumption has been proposed. It has been predicted that this could reduce the power consumption of the optical transmitters in a realistic avionic environment by up to 20 %. This network has been compared to the network using the spectrum slicing technique with particular attention paid to the relative power consumptions. The merits of both networks in the avionic context have been discussed and different scenarios where each network is suitable have been identified.

Currently, the International Technology Roadmap for Semiconductors (ITRS) is targeting the 22nm technology node in accordance with Moore’s Law. The low mobility of silicon makes it inherently unsuitable as a channel material for devices at this scale, and therefore a significant amount of research is being focused at re-evaluating germanium as an alternative substrate. Germanium offers a higher mobility than that of silicon and is compatible with existing silicon device manufacturing techniques. P-type ultra shallow junction (USJ) implemented in germanium exhibit low leakage currents and low sheet resistivity, satisfying the ITRS demands. However, N-type USJ formed using phosphorus as the dopant species do not yet satisfy these requirements due to a high diffusivity and low levels of electrical activation. This is due to the fact that at high phosphorus concentrations, the difference between the equilibrium solid solubility limit and the effective solid solubility is related to the formation of phosphorus-vacancy complexes. These evolve into electrically inactive clusters, by capturing the additional phosphorus resulting in an overall reduction of the electrical activity of the phosphorus population. Another problem is phosphorus out-diffusion during annealing process. In order to overcome these problems, novel techniques are currently being research. This thesis investigates the phosphorus activation and diffusion characteristics as a function of implant temperature and co-implantation of low dose germanium. The samples were subsequently subjected to an isochronal annealing before Hall Effect and SIMS analyses were performed to characterize the electrical activation and diffusion respectively. The results from the studies indicate that it is a non trivial process for germanium to replace silicon in order to become the next dominant substrate.

The abrupt expansion of the Internet use over the last decade led to an uncontrollable amount of media stored in the Web. Image, video and news information has ooded the pool of data that is at our disposal and advanced data mining techniques need to be developed in order to take full advantage of them. The focus of this thesis is mainly on developing robust video analysis technologies concerned with detecting and recognizing activities in video. The work aims at developing a compact activity descriptor with low computational cost, which will be robust enough to discriminate easily among diverse activity classes. Additionally, we introduce a motion compensation algorithm which alleviates any issues introduced by moving camera and is used to create motion binary masks, referred to as compensated Activity Areas (cAA), where dense interest points are sampled. Motion and appearance descriptors invariant to scale and illumination changes are then computed around them and a thorough evaluation of their merit is carried out. The notion of Motion Boundaries Activity Areas (MBAA) is then introduced. The concept differs from cAA in terms of the area they focus on (ie human boundaries), reducing even more the computational cost of the activity descriptor. A novel algorithm that computes human trajectories, referred to as 'optimal trajectories', with variable temporal scale is introduced. It is based on the Statistical Sequential Change Detection (SSCD) algorithm, which allows dynamic segmentation of trajectories based on their motion pattern and facilitates their classification with better accuracy. Finally, we introduce an activity detection algorithm, which segments long duration videos in an accurate but computationally efficient manner. We advocate Statistical Sequential Boundary Detection (SSBD) method as a means of analysing motion patterns and report improvement over the State-of-the-Art.

This thesis describes an experimental investigation into the performance of a radio frequency plasma thruster (RFPT). This type of thruster does not rely on the use of high voltage ion acceleration grids or beam neutralisers which are typically life limiting elements of an electric thruster. The RFPT excites an atomic or molecular gas into a plasma using an external RF antenna. This plasma is contained by a cylindrical dielectric source tube which is open at one end. An axial magnetic field is applied to the plasma using one or more solenoids or a permanent magnet source. The magnetic field is usually applied such that it is constant throughout the length of the source tube, after which the field diverges. The plasma couples with the imposed magnetic field, generating internal field structures, which can be used to regulate both the power coupling and the rate at which the plasma diffuses out of the source tube. In this study, the thruster performance is measured directly, thus providing missions designers with accurate performance data which can be used to assess applicability of the technology to future missions. A re-configurable lab prototype thruster was constructed and a pendulum type thrust balance was developed in order to make direct measurements of the thrust produced by the Surrey Space Centre (SSC) RFPT. The balance has been developed to allow mounting of the thruster together with the necessary RF and DC electrical feeds as well as the propellant feeds and incorporates sensors which allow measurements to be made in an RF plasma environment. A low magnetic field mode of RFPT operation was investigated in order to assess whether this mode of operation may be able to provide performance increases at reduced imposed magnetic fields (< 20 mT), which could enable the serious consideration of this technology for future flight opportunities. The lowered magnetic fields used here may reduce the risk of interference with spacecraft subsystems and perturbation to the spacecraft orbit, when compared to many of the proposed plasma thrusters which use fields in excess of 40 mT. Direct thrust measurements of an RFPT operating in a low field mode are presented as a function of the propellant flow, RF power and for two source tube lengths. The thrust is shown to peak at a field strength, B_0, which is demonstrated to vary with RF power and propellant flow. The peaks are also shown to correspond generally to peaks in the source plasma density, plasma potential and in some cases to exhaust ion beam current. Ion energy distribution function measurements show that strong beams are not present in thruster configurations which use a 170 mm length source tube but are shown to increase in strength when using a shorter 85 mm length tube. The low field mode is shown to generally provide performance increases over the non-magnetised case by a factor of 4. Low field mode thrust measurements using the shorter 85 mm source tube and matching double saddle antenna revealed enhanced peak performance gains over the 170 mm length case, resulting in an increase of the thrust efficiency by up to a factor of 15. The performance of the RFPT operating in this configuration is shown to produce enhanced or equivalent performance, when compared to similar state of the art configurations but at generally lower imposed magnetic fields. This particular optimisation may make possible a first flight opportunity for the RFPT as weak magnetic fields may be supplied easily by electromagnets. Electromagnets can be easily deactivated or field reversed to avoid interference with spacecraft subsystems and orbital parameters unlike permanent magnets which may be required to generate higher magnetic fields. Operation of the RFPT in the low field mode with an 85~mm source has also revealed new low field mode behaviour compared to the 170 mm case, producing sudden, large reductions in the performance of the thruster for imposed fields beyond B_0 which are shown to correspond to drops in source plasma density, plasma resistance and exhaust beam current.

Artificial intelligent and machine learning technologies have already achieved significant success in classification, regression and clustering. However, many machine learning methods work well only under a common assumption that training and test data are drawn from the same feature space and the same distribution. A real world applications is in sports footage, where an intelligent system has been designed and trained to detect score-changing events in a Tennis single match and we are interested to transfer this learning to either Tennis doubles game or even a more challenging domain such as Badminton. In such distribution changes, most statistical models need to be rebuilt, using newly collected training data. In many real world applications, it is expensive or even impossible to collect the required training data and rebuild the models. One of the ultimate goals of the open ended learning systems is to take advantage of previous experience/ knowledge in dealing with similar future problems. Two levels of learning can be identified in such scenarios. One draws on the data by capturing the pattern and regularities which enables reliable predictions on new samples. The other starts from an acquired source of knowledge and focuses on how to generalise it to a new target concept; this is also known as transfer learning which is going to be the main focus of this thesis. This work is devoted to a second level of learning by focusing on how to transfer information from previous learnings, exploiting it on a new learning problem with not supervisory information available for new target data. We propose several solutions to such tasks by leveraging over prior models or features. In the first part of the thesis we show how to estimate reliable transformations from the source domain to the target domain with the aim of reducing the dissimilarities between the source class-conditional distribution and a new unlabelled target distribution. We then later present a fully automated transfer learning framework which approaches the problem by combining four types of adaptation: a projection to lower dimensional space that is shared between the two domains, a set of local transformations to further increase the domain similarity, a classifier parameter adaptation method which modifies the learner for the new domain and a set of class-conditional transformations aiming to increase the similarity between the posterior probability of samples in the source and target sets. We conduct experiments on a wide range of image and video classification tasks. We test our proposed methods and show that, in all cases, leveraging knowledge from a related domain can improve performance when there are no labels available for direct training on the new target data.

Speech separation by machines has been extensively studied for many decades and several algorithms and systems have been proposed. Since the speech separation task for machines is often likened to the speech separation task performed (remarkably well) by the human auditory system several analogies can be found in the proposed systems. This thesis takes a localised view on a few of the aspects of the speech separation task and explores some of the analogies from a machine audition perspective. The first part of the thesis presents algorithms for binaural localisation and separation of speech sources based solely on analysis of the Interaural Phase Difference (IPD) cue. The IPD cue encodes time delay information between two microphones which can be used to establish spatial locations of the sources in the mixture. One well known problem with processing the IPD cue is its periodic nature. This means that a single IPD value can represent several spatial locations of the corresponding source. The phase ambiguity problem has been studied for human auditory processing as well for machines, however, mostly from source localisation perspective. Relatively little attention has been given to phase ambiguity which relates to interaction of the IPDs between the sources present in the mixture. Investigations presented in the thesis explore the use of the IPDs by machines for robust source localisation and separation. Firstly, an algorithm for source localisation is introduced. The algorithm combines the Maximum Likelihood Sample Consensus (MLESAC) based search of line patterns which correspond to speech sources. The search is performed using Cross-phasogram representation of IPDs. Next, the study on the impact of phase ambiguity on separation performance is presented. A source separation algorithm called Localisation based Mask for Source Separation (LOCUS) is introduced. The LOCUS algorithm models the IPDs using Gaussian Mixture Model (GMM). The analysis of the IPDs interaction between different sources is shown to improve initialisation of the GMM and in consequence provided performance gains over the state-of-the art binaural separation methods. The second part of this thesis focuses on using the harmonicity cue for speech separation. The harmonicity is a feature of voiced speech therefore intuitively seems a powerful cue that could enhance separation of speech sources. However, in a multi-speaker scenario segregation of harmonic components is not trivial as it relies heavily on the underlying multi-source pitch determination algorithm. The proposed system uses an approach where speech sources are firstly reconstructed using the LOCUS algorithm and fed into single-source pitch determination algorithm. This gives the opportunity to use well-established single-source pitch determination algorithms which have been known for good robustness and accuracy of provided pitch trajectories. Based on this approach the Pitch based Harmonicity Mask for Source Separation (PRIMUS) algorithm is introduced. The approach is analogous to other separation systems that can be found in the literature however there has been little formal validation of some of the algorithmic choices that need to be considered for such approach. Therefore a detailed review followed by experimental studies of all the stages of the algorithm, from reconstruction of speech sources to calculation of corresponding separation masks, are presented. The final evaluation is done for the PRIMUS and the JANUS (Joint Localisation and Harmonicity Mask for Source Separation) algorithms where the JANUS algorithm computes a set of joint separation masks combining outputs of the LOCUS and the PRIMUS algorithms. The experimental results showed improvements in separation performance that were achieved over the state-of-the art binaural separation methods.

The Internet of things (IoT) refers to uniquely identifiable objects and their virtual representations in an Internet-like structure to be managed and inventoried by computers. Radio-frequency identification (RFID) - a prerequisite for the IoT - is an automatic way for data transaction in object identification and is used to improve automation, inventory control and checkout operations. An RFID system consists of a reader device and one or several tags. Smart reader systems are building blocks for cutting edge applications of RFID and as a subdivision of these systems, RFID smart shelf solutions are started to be implemented for large-scale item-level management where characteristics of reader antennas are critical issue. This work focuses on designing optimised reader antennas for high frequency (HF) RFID smart shelf systems which operate based on inductive coupling between the tag and the reader antennas and have good performance in crowded environments. Firstly, an approach is presented to increase band-width of HF RFID reader antennas to improve the reception of sub-carrier frequencies. A fabricated enhanced band-width antenna at 13.56 MHz is evaluated for its capability in being used for smart shelf applications. The obtained band-width supports sub-carrier frequencies for all the HF RFID standards to be detected easier and thus leads to increased identification range. It is shown the HF RFID technology is capable of identifying the distance of tagged books based on the received magnetic field intensity. Secondly, multi turn small self resonant coil (MT SSRC) antennas are introduced and analysed as a new model of inductively coupled reader antennas. Based on the analysis, two turn planar SSRC (TTP SSRC) antennas having similar dimension with the current HF RFID reader antennas are investigated. Fabricated TTP SSRC antenna operating at 13.56 MHz is resulted to optimised Q factor and more uniform near field pattern in comparison with the similar antennas. Also, a number of TTP SSRC antennas operating at a distinct frequency, 13.56MHz, are fabricated on different substrates and it is shown the desired Q factor and antenna dimension can be obtained based on the dielectric characteristics of the substrate.

Photovoltaic technologies are likely to become one of the world’s major renewable energy generators in the future provided they are able to meet the increasing world energy demands at a significantly lower generation cost compared to conventional non-renewable energy sources. Photovoltaic systems based on 1st generation mono or poly crystalline silicon wafers have already been commercially successful over the past two decades. As the technology further develops however, it faces fundamental limits to further reduce cost which are primarily due to processing of silicon wafers. Hence, a 2nd generation of “thin film” photovoltaic systems, such as amorphous and poly silicon, CdTe and CIGS, which use cheap materials and inexpensive manufacturing processes with relatively high power conversion efficiency, have been developed. In order to commercialise the 2nd generation technology successfully, the efficiency of the thin film photovoltaic panels needs to increase to compete with the 1st generation silicon photovoltaics. Plasmonic structures provide a route to increase the efficiency of 2nd generation thin film photovoltaic devices. With the unique properties of plasmonic structures, such as ability to guide and trap light at nanometre dimensions, light absorption in the photoactive layer of thin film photovoltaic device can be increased resulting in improved device performance. In this research, plasmonic nanoparticles are utilised as an anti-reflection coating on the front side of the PV, coupling light into the active PV layer, and as scattering centres at the back reflector, increasing the path length of the light through the photoactive layer. The optical and electrical effects of the plasmonic structures are modelled simultaneously using a commercial technology computer aided design (TCAD) simulation package to understand and optimise the plasmonic effects on the performance of the 2nd generation thin film amorphous silicon, and 3rd generation organic, photovoltaic devices. The thesis describes the first ever dedicated optoelectronic model to simultaneously simulate optical and electrical properties of plasmonic thin film photovoltaics devices in collaboration with the TCAD software developer Silvaco Inc. The model demonstrates a maximum 12% relative increase in the power conversion efficiency of plasmon enhanced n-i-p configured amorphous silicon thin film photovoltaic devices. This remarkable increase in the performance is due to the light trapping in the photoactive layer of the thin film amorphous silicon photovoltaic devices, which results in improvements in the both the optical and electrical properties. Experimental work was also carried out to observe the plasmonic effects of the metal nanoparticles on the performance of 3rd generation organic photovoltaic devices which were subsequently modelled using the simulation package. A 4% relative increase in the efficiency was achieved using gold nanoparticles. A plasmonic organic photovoltaic device model and material library for the commercial organic semiconductor P3HT:PCBM, has also been developed and benchmarked experimentally. The model has assisted in the understanding of the effect of the plasmonic gold nanoparticles on the increased performance, as well as degradation effects due to the incorporation of silver nanoparticles.

Self-assembled III-V quantum dots (QDs) have been intensely studied for potential applications in solar cell (SC) devices in order to increase power conversion efficiency. Due to their quantum confinement of carriers, QDs have been proposed as a means of implementing the intermediate band solar cell (IBSC). The IBSC concept is characterised by in an increase in photocurrent and a preservation of output voltage, resulting from an enhanced sensitivity to the solar spectrum. The work reported in this thesis is concerned with the development of InAs QDs in GaAs p-i-n solar cell structures, with the aim of realising of an IBSC. The work involves the design, epitaxial growth by molecular beam epitaxy (MBE), device processing and characterisation of the QDSCs. This thesis first investigates InAs/InGaAs dot-in-a-well (DWELL) solar cell structures grown under different conditions. The use of a high-growth-temperature GaAs spacer layers is demonstrated to significantly enhance the performance of the multilayer DWELL solar cells. Threading dislocations were observed for a 30-layer QD structure with GaAs spacer layers grown at a low temperature (510 oC). By growing the GaAs spacer layer at a higher temperature (580 oC), the formation of threading dislocations were suppressed, resulting in enhanced optical properties. The thesis then goes on to address the main challenges facing QD IBSCs, that is, the reduction in open-circuit voltage and the lack of significant increase in short-circuit current. To eliminate the wetting layer and enhance the open-circuit voltage of the QD solar cell, an AlAs cap layer technique was used. This resulted in an enhancement of the open-circuit voltage of a 20-layer InAs/GaAs QDSC from 0.69 V to 0.79 V. Despite a slight reduction in short-circuit current, for the QDSC with AlAs cap layer, the enhancement in the open-circuit voltage was enough to ensure that its efficiency is higher than the QDSC without AlAs cap layers. In an attempt to enhance the short-circuit current, an antimony-mediated growth approach was used to grow high-density QDs. After optimisation of the growth temperature and InAs coverage, a very high in-plane QD density of 1  1011 cm-2 was achieved by applying a few monolayers of antimony prior to QD growth. Compared with a reference QDSC without the incorporation of antimony, the high-density QDSC demonstrates a distinct improvement in short-circuit current from 7.4 mA/cm2 to 8.3 mA/cm2. This result shows that a significant increase in short-circuit current could potentially compensate for the drop in open-circuit voltage observed in InAs/GaAs QD solar cells. Ongoing work on the development of QDSCs with both AlAs capping and antimony-mediated growth have resulted in the simultaneous elimination of the wetting layer and increase in QD absorption in a single device. Overall, the studies in this thesis present important implications for the design and growth of InAs/GaAs QD solar cell structures for the implementation of IBSCs.