The last decade witnessed a wild growth of the Internet traffic, promoted by bandwidth-hungry applications such as Youtube, P2P, and VoIP. This explosive increase is expected to proceed with an annual rate of 34% in the near future, which leads to a huge challenge to the Internet infrastructure. One foremost solution to this problem is advancing the optical networking and switching, by which abundant bandwidth can be provided in an energy-efficient manner. For instance, with Wavelength Division Multiplexing (WDM) technology, each fiber can carry a mass of wavelengths with bandwidth up to 100 Gbits/s or higher. To keep up with the traffic explosion, however, simply scaling the number of fibers and/or wavelengths per fiber results in the scalability issue in WDM networks. One major motivation of this dissertation is to address this issue in WDM networks with the idea of waveband switching (WBS). This work includes the author's study on multiple aspects of waveband switching: how to address dynamic user demand, how to accommodate static user demand, and how to achieve a survivable WBS network. When combined together, the proposed approaches form a framework that enables an efficient WBS-based Internet in the near future or the middle term. As a long-term solution for the Internet backbone, the Spectrum Sliced Elastic Optical Path (SLICE) Networks recently attract significant interests. SLICE aims to provide abundant bandwidth by managing the spectrum resources as orthogonal sub-carriers, a finer granular than wavelengths of WDM networks. Another important component of this dissertation is the author's timely study on this new frontier: particulary, how to efficiency accommodate the user demand in SLICE networks. We refer to the overall study as the resource management in multi-granular optical networks. In WBS networks, the multi-granularity includes the fiber, waveband, and wavelength. While in SLICE networks, the traffic granularity refers to the fiber, and the variety of the demand size (in terms of number of sub-carriers).

M.Ing. / The focus of the project is on the design of a variable output current source applied in the testing of circuit breakers. The possibility of the use of high-frequency, switch mode converters for the application, will be investigated. The expectation is the improvement of the system currently in use, with the help of a power electronic converter. For the application, a 1 to 200A adjustable current source must be developed, which will be powered from the 220V, 50Hz network. A number of possible solutions to the problem will be investigated. One of the challenges of the project is that the output of the converter must be a current and not a voltage, which is normally the case. Based on mathematical calculations and practical results, the best possible solution to the problem is obtained. An optimum system is presented, which meets the desired specifications.

Submitted in partial fulfillment for the Masters in Engineering Degree: Electrical Engineering, Department of Electrical Engineering, Durban University of Technology, Durban, South Africa, 2017. / All-optical networks (AON) based optical burst switching (OBS) promise to be the ultimate backbone network technology solution for next generation( NG) as well as fu­ ture generation (FG)networks because of their relatively higher resources utilization, great flexibility at lower cost and potential massive bandwidth capacities both at trans­ mission and switching levels. By design, buffering is not provisioned in interior nodes. End users exchange data with one another through end-to-end light channels, called lightpaths in which wavelength continuity is maintained. In practice, their establish­ ing, in a cost effective manner remains an inescapable challenge. The routing and wave­ length assignment (RWA) problem entices successful establishment of a physical route for each lightpath connection request, assigning a wavelength to each route and at the same time ensm·ing end to end continuity, subject to the limited number of wave­ lengths. The wavelengths must be assigned such that no lightpaths can share the same wavelength simultaneously on a given fibre, or else contentions may occur. Some data bursts may be discarded whenever contention occurs among multiple bursts that arrive simultaneously at any interior node using the same wavelength and are intended for the same output port. Because of the buffer-less nature of OBS networks, conten­ tion/congestion in the core network can quickly lead to degradation in overall network performance at moderate to high traffic levels due to heavy burst loses. In this disser­ tation we propose and evaluate a congestion management approach we refer to as '"en­ hanced congestion management" which gears towards rendering and guaranteeing a consistent QoS as well as rational and fair use of available network links. Simulation results show that the scheme can effectively minimize both contention and congestion and at the same time improving both throughput and effective utilization under mod­ erate to high network traffic loads / M

Electromagnetic interference (EMI) is a major concern in inverter motor drive systems. The sources of EMI have been commonly identified as high switching dv/dt and di/dt rates interacting with inverter parasitic components. The reduction of parasitic components relies on highly integrated circuit layout and packaging. This is the way to deal with noise path. On the other hand, switching dv/dt and di/dt can be potentially reduced by soft-switching techniques; thus the intensity of noise source is reduced. In this paper, the relation between the dv/dt di/dt and the EMI generation are discussed. The EMI sources of a hard-switching single-phase PWM inverter are identified and measured with separation of common-mode and differential-mode noises. The noise reduction in an auxiliary resonant snubber inverter (RSI) is presented. The observation of voltage ringing and current ringing and the methods to suppress these ringing in the implementation of RSI are also discussed. The test condition and circuit layout are described as the basis of the study. And the experimental EMI spectra of both hard- and soft-switching inverter are compared. The effectiveness and limitation of the EMI reduction of the ZVT-RSI are also discussed and concluded. The control interface circuit and gate driver design are described in the appendix. The implementation of variable charging time control of the resonant inductor current is also explained in the appendix. / Master of Science

A graph in which every edge is labeled positive or negative is called a signed graph. We determine the number of ways to sign the edges of the McGee graph with exactly two negative edges up to switching isomorphism. We characterize signed graphs that are both sign-symmetric and have a frustration index of 1. We prove some results about which signed graphs on complete multipartite graphs have frustration indices 2 and 3. In the final part, we derive the relationship between the frustration index and the number of parts in a sign-symmetric signed graph on complete multipartite graphs.

Most current augmented reality (AR) displays present content at a fixed focal demand. At the same time, real-world stimuli can occur at a variety of focal distances. To integrate information, users need to switch eye focus between virtual and real-world information continuously. Previously, Gabbard, Mehra, and Swan (2018) examined these issues, using a text-based visual search task on a monocular AR display. This thesis replicated and extended the previous experiment by including a new experimental variable stereopsis (stereo, mono) and fully crossing the variables of context switching and focal distance switching, using AR haploscope. The results from the monocular condition indicate successful replication, which is consistent with the hypothesis that the findings are a general property of AR. The outcome of the stereo condition supports the same adverse effects of context switching and focal distance switching. Further, participants have better performance and less eye fatigue in the stereo condition compared to the monocular condition.

A single user deconcentrator is designed for a circuit switched digital switching system which emphasizes mass conferencing. The switching system consists of distributed concentrators which include a first level of conferencing and a broadcast bus which includes a second level of conferencing. The deconcentrator features volume adjustment of individual conferences as well as a third and last level of conferencing- a conference of the individual conferences obtained from the broadcast bus. The design identifies the MSI and SSI components of the deconcentrator and provides an estimate of timing and board size. An alternative volume adjustment algorithm and the extension of a single user deconcentrator to a multi-user deconcentrator is discussed.

This paper architecturally describes the switching portion of a digital communications system that is dedicated to conferencing. The basic ideas and methods of circuit switching and packet switching are introduced. The conferencing function is described, and some resulting design considerations are discussed. The architecture of the switch is then presented. Circuit switching techniques are used throughout the architecture of the switch, coupled with arithmetic processing to accomplish the conferencing function. The architecture is developed in such a way that it is expandable in all directions to meet a given set of requirements. The requirements include the number of users the system supports and the number of conference channels provided. The processing stages of the switch can be sized based on these requirements and the chosen component speeds. The basic timing of each stage is given to describe its operation and establish the critical delay paths. The resulting switching methods first introduced. The switch is also tested to see if it fits the criteria for being a distributed processing system. It is concluded that if the provision for dynamic reconfiguration is added, the switch fits the criteria. Finally, further topics of study are suggested.

This thesis is mainly a survey of optical devices used in optical logic and optical space, time and frequency division switching. These are mainly optical logic devices, optical switches, optical switching matrices, optical memories and optical wavelength multiplexers and demultiplexers and optical wavelength shifters. The devices are based on a variety of operating principles and are implemented with a variety of technologies. The basic principles and technologies are described together with typical performance parameters. The different devices and technologies are compared, and areas of possible improvements indicated. As an application of this survey, the possibility of implementing optical logic circuits and an optical time-division switch for Sonet based transmission systems is discussed. / Master of Science

Investigations of polarization switching over broad time and electric field domains, in various modified Pb-based perovskite ferroelectrics, were systematically performed by ferroelectric switching current transient and bipolar drive P-E responses. Studies were performed from E«Ec to E»Ec, where Ec is the coercive field These investigations have shown the presence of broad relaxation time distributions for the switching process, which can extend over several decades in order of magnitude in time, and where the distribution is strongly dependent on the applied electric field. By performing the study of domain dynamics and polarization switching over extremely broad time domains (10⁻⁸ t < 10² sec), more complete information has been obtained that allows for development of a better mechanistic understanding. Prior polarization kinetics studies have focused on relatively narrow time ranges, and were fit to the Avarami equation, which contains a single relaxation time. However, our broad band width polarization dynamics and frequency relaxation studies have been fit to multiple stretched exponential functions extending over decades of order of magnitude in the time domain. Stretched exponential functions for domain nuclei formation, and for domain variant growth have been found. For example, [001]c, [110]c, and [111]c oriented PZN-4.5%PT crystals, nucleation was found to be a volume process (n=3) rather than just a domain wall restricted process. Consequently, nucleation is heterogeneous. And, growth of a domain variant with reversed polarization was found to be a boundary process (n=2), involving diffuse or rough domain walls. We have extended these studies to various types of ferroelectrics including hard, soft and relaxor types. / Master of Science