Filters
Availability
Publication Year
Resource Type
Faculty
School/Institute
Publication Search Results
Sort
Results per page
1 - 10 of 79
-
-
-
-
-
-
(2019)ThesisThe main aim of this thesis is to report work done to improve the imaging performance of Atomic Force Microscopes (AFMs). According to this purpose, a data-driven controller is designed using the frequency response data from the capacitive sensors of an AFM. Using an optimization process, the frequency response of the controller is obtained. An input-output error minimization method is considered while forming the optimization norm. The closed-loop stability of the AFM system is guaranteed by using negative-imaginary theory and small-gain theory based constraints. The proposed controller provides significant damping of resonant peaks on both X and Y-axes of the Piezo Tube Scanner (PTS). Conventional raster and spiral tracking performance of the controller are compared to the open-loop tracking. Moreover, the better performance of the designed data-driven controller is obtained compared to the AFM's built-in PI controller. Finally, high speed spiral images are generated by using this proposed controller which shows improved imaging performance of the AFM at high scanning frequencies.
-
(2017)ThesisMedium Access Control (MAC) layer protocols are responsible for the reliable transition of data by providing channel access control mechanism and act as an interface between the physical and network layer. However, in modern communication networks, the large relative propagation delay (α>>1) is problematic for the MAC layer of the protocol stack. Most contemporary MAC protocols have developed under the assumption of negligibly small relative propagation delay (α<<1) and these protocols perform poorly where α>>1 such as in space and Underwater Acoustic Sensor Networks (UWASNs). When $\alpha$ is large, carrier sensing protocols such as CSMA fail to gather information that is useful for collision avoidance since the channel state at the transmitter is not correlated with the state at the receiver when any transmission would arrive. Furthermore, the negotiation phase of handshaking based protocols substantially exceeds the time for data transmission making such protocol inefficient due to low channel utilization. In this work, we show by analysis and simulation that for networks with large relative propagation delay, ALOHA is competitive with other more complex specialized MAC protocols such as T-Lohi, MACA, CSMA and FS-MACA and is more attractive than commonly understood as a MAC. ALOHA is the basis of superior MAC protocols and is the simplest random access control technique. Furthermore, its performance is independent of relative propagation delay. The conventional analysis of ALOHA protocol estimates the effective throughput based on an infinite number of nodes with Poisson traffic distribution. Other works have extended the analysis to consider a finite number of nodes. However, these works lack a detailed representation of throughput against the offered and transmitted load. In this work, we develop a more accessible and easily extendable analytical expression to calculate the throughput of unbuffered or pure ALOHA with evenly distributed traffic loads and evaluate the relationship between transmitted and offered load. To find the maximum throughput of ALOHA, we isolate the causes of unsuccessful transmissions and characterize them as being dropped packets or two different types of collisions. The accepted analysis of ALOHA considers the protocol to operate without buffering at each node, but in modern use, buffering of transmissions is an obvious implementation feature, and is not addressed by the published works. We provide an expression to find the throughput for buffered ALOHA with a finite number of active nodes. This shows that the effect of buffering increases the throughput, which can be almost double that reported in existing theories. We show via analysis of the intra-packet interval for a finite number of active nodes, that the assumption of a Poisson traffic distribution concept is invalid at the MAC layer when buffering is applied. Finally, the more general case where the traffic load is distributed unevenly is examined and a mathematical model to calculate the throughput of buffered ALOHA with uneven traffic generation per node is presented. The consequence is that, ALOHA is a very attractive candidate for contemporary networks where the relative propagation delay is large, such as space and UWASNs.
-
(2017)ThesisMotion capture is the process of determining the kinematic and kinetic motions of objects or an object for analysis purposes. It can be performed by several means, such as mechanical, magnetic, inertial and optical. However, motion capture by computer vision based technique is gaining increasing interest. A significant part of it is human motion capture because of its number of potential applications and inherent complexities. The most significant complexities are capturing the pose and motion of a highly articulated human subject in outdoor environment, occlusions by self or other object and compensating the egomotion of camera for any jitter or vibration to name a few. Motion capture technology suffers mainly when the capturing is not performed in controlled surroundings which hinders motion capture in an outdoor environment requiring the traversal of large areas. Overcoming these challenges provide a great range of applications perspective as computer vision-based human motion capture is the best non-invasive solution for it. Applications are of broadly three types roughly, analysis, control, and surveillance. Although the most widely used optical motion capture system (MCS) to date has been Vicon, considering, it is evident that it will not be useful for many purposes of these application types as it only operates inside a laboratory with a robust structural setup. Therefore, as there is a need for more convenient MCSs in environments in which Vicon fails, vision-based MCSs have attracted the attention of many researchers, with their wide range of applications requiring different levels of accuracy. High-accuracy ones, such as in the clinical sector and gait analysis, usually use optical MCSs which are unaffordable for many users as they require complex arrangements of expensive equipment and, moreover, are sensitive to lighting conditions and shadows. Therefore, with the aim of overcoming these restrictions, lightweight stereo vision-based MCSs have been developed. Although there are many techniques available for them, very few have an ultra-wide baseline distance between cameras. In the first part of this study, a stereo camera rig with an ultra-wide baseline distance and conventional cameras with fish-eye lenses is designed. This system is tested in different situations to validate its applicability as a stereo camera rig as well as its performance for motion capture and, eventually, articulated human motion capture in indoor and outdoor environments. Its cameras provide a wide field of view (FOV) which increases the coverage area and also enables the baseline distance to be increased to cover the common area required for both cameras' views to perform as a stereo camera. In this study, a marker-based approach, which provides an additional facility to track articulated human kinematics and is computationally very cheap as it does not require all the image features to be computed, is developed, with an adaptive thresholding method applied to extract each marker. As the cameras have fish-eye lenses, they cannot be well estimated using a pinhole camera model which makes it difficult to estimate the depth information. In this study, to restore the 3D coordinates, a unique calibration method is used to develop a relationship among the pixels’ dimensions and distances in an image’s and real world’s coordinates respectively. The quality of this 3D coordinate reconstruction is compared with that of an existing commercially available `gold standard' MCS, the Vicon system for indoor data, and is shown to have the same order of accuracy while incurring less cost and complexity. In the subsequent parts of this study, occlusion detection is considered because, in the marker-based capturing of articulated human kinematics, the occlusion of a marker is one of the major challenges. The detection algorithm differentiates among types of occlusions and predicts any missing marker position where necessary. The designed approach is intended to capture motions where a traditional MCS fails to perform well, such as in an outdoor environment. Experiments are conducted in such situations and show similar results to those obtained indoors. Finally, as this design is intended to be mounted on a moving carrier, such as a drone or car, a method for compensating the camera's egomotion is proposed. Using it and the developed algorithms, data are taken from a cyclist in a situation in which both the camera and subject are moving outdoors under varying lighting conditions. While changing these conditions is another major challenge, the proposed system is not sensitive to different lighting levels and shadows.
-
(2017)ThesisThe technical, environmental, and economic incentives have led to the development of a new grid infrastructure known as microgrid (MG) where a number of renewable energy resources (RESs)-based distributed generators (DGs) and loads can be employed as a single controllable entity in a specified electrical boundary. The RESs-based DGs depend on environmental conditions which are intermittent and uncertain. Energy storage systems (ESSs) can mitigate the negative impact of RESs-based DGs through active and reactive power support and improve the power quality in MGs. The voltage source converter (VSC)-based power electronic interfaces are used in this dissertation to integrate these ESSs into the MG. The aim of this thesis is to design and implement nonlinear partial feedback linearization technique to regulate the switching schemes of the VSC. The performance of the proposed control approach is validated through electromagnetic transient simulations in MATLAB/SIMULINK platform. This thesis reports three main contributions. First, the dynamic model of a three-phase grid-connected proton exchange membrane fuel cell (PEMFC) system is developed and then based on the dynamic model, the partial feedback linearizing controller is designed. The active and reactive power control is achieved through the regulation of dq-axes components of the grid current. Additionally, a rate limiting action is used to prevent over modulation in the VSC which limits the rate of change of active power references sent to the controller. A set reference power tracking is achieved by satisfying the constraint of the modulation index. Secondly, the third harmonic injected modulation approach is incorporated with the partial feedback linearization technique for a grid-connected UC-based energy storage system (UCESS). The control signals generated from the partial feedback linearizing controller are modified for third harmonic injection purpose to increase efficiency and maximize the utilization of the DC-bus voltage in the VSC. Finally, an online energy management strategy is considered for an islanded MG to ensure the economic operation. The battery energy storage system (BESS)-based DGs with the partial feedback linearized inner controller is considered, for active and reactive power support to the MG. The optimal active power reference obtained from the MG central controller are sent to each individual DG which is then followed by the inner controller.
-
(2017)ThesisWhiplash is a very common ailment encountered in clinical practice that is usually a result of vehicle accidents but also domestic activities and sports injuries. It is normally caused when neck organs (specifically muscles) are impaired. Whiplash-associated disorders include acute headaches, neck pain, stiffness, arm dislocation, abnormal sensations, and auditory and optic problems, the persistence of which may be chronic or acute. Insurance companies compensate almost fifty percent of claims lodged due to whiplash injury through compulsory third party motor insurance. The morphological structures of neck muscles undergo hypertrophy or atrophy following damage caused to them by accidents. Before any medical treatment is applied , any such change needs to be known which requires 3D visualization of the neck muscles through a proper segmentation of them because the neck contains many other sensitive organs such as nerves, blood vessels, the spinal cord and trachea. The segmentation of neck muscles in medical images is a more challenging task than those of other muscles and organs due to their similar densities and compactness, low resolutions and contrast in medical images, anatomical variabilities among individuals, noise, inhomogeneity of medical images and false boundaries created by intra-muscular fat. Traditional segmentation algorithms, such as those used in thresholding and clustering-based methods, are not applicable in this project and also not suitable for medical images. Although there are some techniques available in clinical research for segmenting muscles such as thigh, tongue, leg, hip and pectoral ones, to the best of author's knowledge, there are no methods available for segmenting neck muscles due to the challenges described above. In the first part of this dissertation, an atlas-based method for segmenting MR images, which uses linear and non-linear registration frameworks, is proposed, with output from the registration process further refined by a novel parametric deformable model. The proposed method is tested on real clinical data of both healthy and non-healthy individuals. During the last few decades, registration- and deformable model-based segmentation methods have been very popular for medical image segmentation due to their incorporation of prior information. While registration-based segmentation techniques can preserve topologies of objects in an image, accuracy of atlas-based segmentation depends mainly on an effective registration process. In this study, the registration framework is designed in a novel way in which images are initially registered by a distinct 3D affine transformation and then aligned by a local elastic geometrical transformation based on discrete cosines and registered firstly slice-wise and then block-wise. The numbers of motion parameters are changed in three different steps per frame. This proposed registration framework can handle anatomical variabilities and pathologies by confining its parameters in local regions. Also, as warping of the framework relies on number of motion parameters, similarities between two images, gradients of floating image and coordinate mesh grid values, it can easily manage pathological and anatomical variabilities using a hierarchical parameter scheme. The labels transferred from atlas can be improved by deformable model-based segmentation. Although geometric deformable models have been widely used in many biomedical applications over recent years, they cannot work in the context of neck muscles segmentation due to noise, background clutter and similar objects touching each other. Another important drawback of geometric deformable models is that they are many times slower than parametric deformable ones. Therefore, the segmentation results produced by the registration process are ameliorated using a multiple-object parametric deformable model which is discussed in detail in the second part of this thesis. This algorithm uses a novel Gaussian potential energy distribution which can adapt to topological changes and does not require re-parameterization. Also, it incorporates a new overlap removal technique which ensures that there are no overlaps or gaps inside an object. Furthermore, stopping criteria of vertices are designed so that difference between boundaries of the deformable model and actual object is minimal. The multiple-object parametric deformable model is also applied in a template contours propagation-based segmentation technique, as discussed in the third part of this dissertation. This method is semi-automatic, whereby a manual delineation of middle image in a MRI data set is required. It can handle anatomical variabilities more easily than atlas-based segmentation because it can segment any individual's data irrespective of his/her age, weight and height with low computational complexity and it does not depend on other data as it operates semi-automatically. In it, initial model contour resides close to the object's boundary, with degree of closeness dependent on slice thicknesses and gaps between the slices.