Internet of Things (IoT) is an upcoming prominent technology which is quickly growing. Not all IoTdemands of computing resources can be satisfied by cloud, and obstacles are firmer when it comes to mobility and agility. Thus, edge computing as a suitable middleware can fill the gap between cloud and IoT devices. Refer to the latest researches, edge security is still evolving, and forensics is almost untouched. In this work, we attempt to study available technologies and materials then design and implement an edge computing application which addresses the challenge of log collection from different edge devices. The interaction between edge and cloud is in a fashion that cloud entity can perform log collection from heterogeneous edge devices belong to different owners. On the other hand, due to local computing on the logs, the edge devicecan trust cloud party. Results show that thanks to the crucial topological position of the edge devices, the concept of edge computing can easily solve similar cloud challenges.

Waveguides are used to transmit electromagnetic signals.Their geometry is typically long and slender their particularshape can be used in the design of computational methods. Onlyspecial modes are transmitted and eigenvalue and eigenvectoranalysis becomes important. We develop a .nite-element code for solving theelectromagnetic .eld problem in closed waveguides .lled withvarious materials. By discretizing the cross-section of thewaveguide into a number of triangles, an eigenvalue problem isderived. A general program based on Arnoldi’s method andARPACK has been written using node and edge elements toapproximate the .eld. A serious problem in the FEM was theoccurrence of spurious solution, that was due to impropermodeling of the null space of the curl operator. Therefore edgeelements has been chosen to remove non physical spurioussolutions that arises. Numerical examples are given for homogeneous andinhomogeneous waveguides, in the homogeneous case the resultsare compared to analytical solutions and the right order ofconvergence is achieved. For the more complicated inhomogeneouswaveguides with and without striplines, comparison has beendone with results found in literature together with gridconvergence studies. The code has been implemented to be used in an industrialenvironment, together with full 3-D time and frequency domainsolvers. The2-D simulations has been used as input for full3-D time domain simulations, and the results have been comparedto what an analytical input would give. / NR 20140805

In the progress of city development What's your positon in a city What's an architecture's position in a city What's the condition of the city's edge / Master of Architecture

Let G=(V,E) be a graph with vertex set V and edge set E. A vertex v∈V ve-dominates every edge incident to it as well as every edge adjacent to these incident edges. The vertex–edge degree of a vertex v is the number of edges ve-dominated by v. Similarly, an edge e=uv ev-dominates the two vertices u and v incident to it, as well as every vertex adjacent to u or v. The edge–vertex degree of an edge e is the number of vertices ev-dominated by edge e. In this paper we introduce these types of degrees and study their properties.

At the crossing of a strong architectural edge and an axis line, it is necessary to articulate the intersection and acknowledge the moment. But what if, at the point of this intersection, other contextual factors work against the articulation? What if there is an opportunity to not only mark the intersection, but in doing so strengthen the edge condition, elevate the importance of the axis line and provide a celebrated threshold experience? This project looks at this very condition as it exists within the context of the Virginia Tech campus in Blacksburg, Virginia. At the intersection of the axis of symmetry for the campus and the building edge along Perry Street, there is no acknowledgment of this crossing. In fact, in its present condition, the intersection is beset by a breakdown in the edge condition and only a trace of the powerful axis line. In addressing the challenges that plague this existing condition, this project will seek to achieve four things with respect to the Virginia Tech campus, at large: 1. Articulate the termination point of the axis of symmetry for the campus by strengthening the pedestrian path that runs along the axis providing a clearly defined route to the Drill Field. 2. A redefinition of the edge along Perry Street, repairing the breech in the building wall and connecting the components that make up the edge. 3. Strengthen intersection of the edge and the axis/path line by developing a new pedestrian portal into the heart of campus thereby providing a formal entry point along an edge that currently does not articulate the entry points into campus. 4. Develop the architectural context within the site, bridging the divide between the architectural traditions of the campus core with the modernist vernacular of the Perry Street Edge. / Master of Architecture

Several flow control schemes were designed and tested to determine the most suitable method for reducing the momentum deficit in a rotor wake and thus attenuate rotor-stator interaction noise. A secondary concern of the project was to reduce the amount of blowing required air for wake filling and thus limit the efficiency penalty in an aircraft engine environment. Testing was performed in a linear blow down cascade wind tunnel, which produced an inlet Mach number of 0.345. The cascade consisted of five blades with the stagger angle, pitch, and airfoil cross-section representative of 90% span of the rotor geometry for NASA's Active Noise Control Fan (ANCF) test rig. The Reynolds number for the tests was based on inlet conditions and a chord length of 4 inches. Trailing edge jets, trailing edge slots, ejector pumps, and pressure/suction side jets were among the configurations tested for wake filling. A range of mass flow percentages were applied to each configuration and a pressure loss coefficient was determined for each. Considerable reduction in wake losses took place for discrete jet blowing techniques as well as pressure side and suction side jets. In the case of the pressure and suction side jets, near full wake filling occurred at 0.75% of the total mass flow. In terms of loss coefficients and calculated momentum coefficients, the suction/pressure surface jets were the most successful. Jets located upstream of the trailing edge helped to re-energize the momentum deficits in the wake region by using a flow pattern capable of mixing the region while also adding momentum to the wake. The slotted configuration was modeled after NASA's current blowing scheme and served as a baseline for comparison for all data. Digital particle image velocimetry was performed for flow visualizations as well as velocity analysis in the wake region. / Master of Science

With the recent advancement in deep learning, there has been increasing interest to apply deep learning algorithms to mobile edge devices (e.g. wireless access points, mobile phones, and self-driving vehicles). Such devices are closer to end-users and data sources compared to cloud data centers, therefore deep learning on the edge leads to several merits: 1) reduce communication overhead (e.g. latency), 2) preserve data privacy (e.g. not leaking sensitive information to cloud service providers), and 3) promote autonomy without the need of continuous network connectivity. However, it also comes with a trade-off that deep learning on the edge often results in less prediction accuracy or longer inference time. How to optimize such a trade-off has drawn a lot of attention among the machine learning and systems research communities. Those communities have explored three main directions: partitioning, caching, and compression to solve the problem. Deep learning model partitioning works in distributed and parallel computing by leveraging computation units (e.g. edge nodes and end devices) of different capabilities to achieve the best of both worlds (accuracy and latency), but the inference time of partitioning is nevertheless lower bounded by the smallest of inference times on edge nodes (or end devices). In contrast, model caching is not limited by such a lower bound. There are two trends of studies in caching, 1) caching the prediction results on the edge node or end device, and 2) caching a partition or less complex model on the edge node or end device. Caching the prediction results usually compromises accuracy, since a mapping function (e.g. a hash function) from the inputs to the cached results often cannot match a complex function given by a full-size neural network. On the other hand, caching a model's partition does not sacrifice accuracy, if we employ a proper partition selection policy. Model compression reduces deep learning model size by e.g. pruning neural network edges or quantizing network parameters. A reduced model has a smaller size and fewer operations to compute on the edge nodes or end device. However, compression usually sacrifices prediction accuracy in exchange for shorter inference time. In this thesis, our contributions to partitioning, caching, and compression are covered with experiments on state-of-the-art deep learning models. In partitioning, we propose TeamNet based on competitive and selective learning schemes. Experiments using MNIST and CIFAR-10 datasets show that on Raspberry Pi and Jetson TX2 (with TensorFlow), TeamNet shortens neural network inference as much as 53% without compromising predictive accuracy. In caching, we propose CacheNet, which caches low-complexity models on end devices and high-complexity (or full) models on edge or cloud servers. Experiments using CIFAR-10 and FVG have shown on Raspberry Pi, Jetson Nano, and Jetson TX2 (with TensorFlow Lite and NCNN), CacheNet is 58-217% faster than baseline approaches that run inference tasks on end devices or edge servers alone. In compression, we propose the logographic subword model for compression in machine translation. Experiments demonstrate that in the tasks of English-Chinese/Chinese-English translation, logographic subword model reduces training and inference time by 11-77% with Theano and Torch. We demonstrate our approaches are promising for applying deep learning models on the mobile edge. / Thesis / Doctor of Philosophy (PhD) / Edge artificial intelligence (EI) has attracted much attention in recent years. EI is a new computing paradigm where artificial intelligence (e.g. deep learning) algorithms are distributed among edge nodes and end devices of computer networks. There are many merits in EI such as shorter latency, better privacy, and autonomy. These advantages motivate us to contribute to EI by developing intelligent solutions including partitioning, caching, and compression.

On a manifold with edge we construct a specific class of (edgedegenerate) elliptic differential operators. The ellipticity refers to the principal symbolic structure σ = (σψ, σ^) of the edge calculus consisting of the interior and edge symbol, denoted by σψ and σ^, respectively. For our choice of weights the ellipticity will not require additional edge conditions of trace or potential type, and the operators will induce isomorphisms between the respective edge spaces.

The Internet of Things (IoT) has deeply changed how we interact with our world. Today, smart homes, self-driving cars, connected industries, and wearables are just a few mainstream applications where IoT plays the role of enabling technology. When IoT became popular, Cloud Computing was already a mature technology able to deliver the computing resources necessary to execute heavy tasks (e.g., data analytic, storage, AI tasks, etc.) on data coming from IoT devices, thus practitioners started to design and implement their applications exploiting this approach. However, after a hype that lasted for a few years, cloud-centric approaches have started showing some of their main limitations when dealing with the connectivity of many devices with remote endpoints, like high latency, bandwidth usage, big data volumes, reliability, privacy, and so on. At the same time, a few new distributed computing paradigms emerged and gained attention. Among all, Edge Computing allows to shift the execution of applications at the edge of the network (a partition of the network physically close to data-sources) and provides improvement over the Cloud Computing paradigm. Its success has been fostered by new powerful embedded computing devices able to satisfy the everyday-increasing computing requirements of many IoT applications. Given this context, how can next-generation IoT applications take advantage of the opportunity offered by Edge Computing to shift the processing from the cloud toward the data sources and exploit everyday-more-powerful devices? This thesis provides the ingredients and the guidelines for practitioners to foster the migration from cloud-centric to novel distributed design approaches for IoT applications at the edge of the network, addressing the issues of the original approach. This requires the design of the processing pipeline of applications by considering the system requirements and constraints imposed by embedded devices. To make this process smoother, the transition is split into different steps starting with the off-loading of the processing (including the Artificial Intelligence algorithms) at the edge of the network, then the distribution of computation across multiple edge devices and even closer to data-sources based on system constraints, and, finally, the optimization of the processing pipeline and AI models to efficiently run on target IoT edge devices. Each step has been validated by delivering a real-world IoT application that fully exploits the novel approach. This paradigm shift leads the way toward the design of Edge Intelligence IoT applications that efficiently and reliably execute Artificial Intelligence models at the edge of the network.

The degradation of bridge edge beam systems in Sweden entailed the study of new alternativedesigns, which may become more optimal from a life-cycle perspective than the currenttypical solution used (concrete integrated). Subsequently, a U-shaped steel edge beamproposed by the consulting engineering group Ramböll was considered by the SwedishTransport Administration for its use in a real bridge project. This thesis follows theimplementation of this alternative in a bridge project.The goals of the thesis are to study the development of the U-shaped steel edge beam solutionin the case study, and to identify the key factors behind it. The case study consists of a roadframe bridge where a heavily damaged bridge edge beam system is going to be replaced.For the structural design of the solution, a static linear analysis of a vehicle collision has beencarried out with the help of Finite Element Modelling and current codes. The report shows themodelling of the design solution throughout different development phases in the project. Thecommercial software used has been LUSAS.As an outcome of the project, four models have been designed and analysed, two of themdeveloped by the author as proposed solutions. The factors behind the different changes in thedesign have been identified as: (1) structural resistance, (2) constructability and (3) the use ofstainless steel. Moreover, the connection between the steel edge beam and the concrete slabhas been the main critical part for the structural resistance. Finally, the current preliminarymodel at the moment this thesis is written, which was proposed in the project meetings, meetsthe requirements from a structural point of view.