Computer Engineering Commons^™

Real-Time Analysis Of Aerosol Size Distributions With The Fast Integrated Mobility Spectrometer (Fims), Daisy Wang

McKelvey School of Engineering Theses & Dissertations

The Fast Integrated Mobility Spectrometer (FIMS) has emerged as an innovative instrument in the aerosol science domain. It employs a spatially varying electric field to separate charged aerosol particles by their electrical mobilities. These separated particles are then enlarged through vapor condensation and imaged in real time by a high-speed CCD camera. FIMS achieves near 100% detection efficiency for particles ranging from 10 nm to 600 nm with a temporal resolution of one second. However, FIMS’ real-time capabilities are limited by an offline data analysis process. Deferring analysis until hours or days after measurement makes FIMS' capabilities less valuable for …

Watch: A Distributed Clock Time Offset Estimation Tool On The Platform For Open Wireless Data-Driven Experimental Research, Cassie Jeng

McKelvey School of Engineering Theses & Dissertations

The synchronization of the clocks used at different devices across space is of critical importance in wireless communications networks. Each device’s local clock differs slightly, affecting the times at which packets are transmitted from different nodes in the network. This thesis provides experimentation and software development on POWDER, the Platform for Open, Wireless Data-driven Experimental Research, an open wireless testbed across the University of Utah campus. We build upon Shout, a suite of Python scripts that allow devices to iteratively transmit and receive with each other and save the collected data. We introduce WATCH, an experimental method to estimate clock …

Mirror Position Detection In A Catoptric Surface, Run Zhang

McKelvey School of Engineering Theses & Dissertations

The Catoptric Surface research project is a pioneering exploration of controlling daylight effects within built environments. In this thesis, we focus on the mirror position detection problem, which plays a vital role in achieving dynamic control over the direction of reflected light within a space. To address the challenge of mirror position detection, we employ computer vision techniques, specifically edge detection and the RANdom SAmple Consensus (RANSAC) algorithm. Edge detection is utilized to identify significant changes in intensity or color, corresponding to object boundaries, while RANSAC is applied for ellipse fitting. By iteratively selecting minimal subsets of points and fitting …

Targeted Adversarial Attacks Against Neural Network Trajectory Predictors, Kaiyuan Tan

McKelvey School of Engineering Theses & Dissertations

Trajectory prediction is an integral component of modern autonomous systems as it allows for envisioning future intentions of nearby moving agents. Due to the lack of other agents' dynamics and control policies, deep neural network (DNN) models are often employed for trajectory forecasting tasks. Although there exists an extensive literature on improving the accuracy of these models, there is a very limited number of works studying their robustness against adversarially crafted input trajectories. To bridge this gap, in this paper, we propose a targeted adversarial attack against DNN models for trajectory forecasting tasks. We call the proposed attack TA4TP for …

Adversarial Patch Attacks On Deep Reinforcement Learning Algorithms, Peizhen Tong

McKelvey School of Engineering Theses & Dissertations

Adversarial patch attack has demonstrated that it can cause the misclassification of deep neural networks to the target label when the size of patch is relatively small to the size of input image; however, the effectiveness of adversarial patch attack has never been experimented on deep reinforcement learning algorithms. We design algorithms to generate adversarial patches to attack two types of deep reinforcement learning algorithms, including deep Q-networks (DQN) and proximal policy optimization (PPO). Our algorithms of generating adversarial patch consist of two parts: choosing attack position and training adversarial patch on that position. Under the same bound of total …

Applying Hls To Fpga Data Preprocessing In The Advanced Particle-Astrophysics Telescope, Meagan Konst

McKelvey School of Engineering Theses & Dissertations

The Advanced Particle-astrophysics Telescope (APT) and its preliminary iteration the Antarctic Demonstrator for APT (ADAPT) are highly collaborative projects that seek to capture gamma-ray emissions. Along with dark matter and ultra-heavy cosmic ray nuclei measurements, APT will provide sub-degree localization and polarization measurements for gamma-ray transients. This will allow for devices on Earth to point to the direction from which the gamma-ray transients originated in order to collect additional data. The data collection process is as follows. A scintillation occurs and is detected by the wavelength-shifting fibers. This signal is then read by an ASIC and stored in an ADC …

Model-Based Deep Learning For Computational Imaging, Xiaojian Xu

McKelvey School of Engineering Theses & Dissertations

This dissertation addresses model-based deep learning for computational imaging. The motivation of our work is driven by the increasing interests in the combination of imaging model, which provides data-consistency guarantees to the observed measurements, and deep learning, which provides advanced prior modeling driven by data. Following this idea, we develop multiple algorithms by integrating the classical model-based optimization and modern deep learning to enable efficient and reliable imaging. We demonstrate the performance of our algorithms by validating their performance on various imaging applications and providing rigorous theoretical analysis.

The dissertation evaluates and extends three general frameworks, plug-and-play priors (PnP), regularized …

Machine Learning For Analog/Mixed-Signal Integrated Circuit Design Automation, Weidong Cao

McKelvey School of Engineering Theses & Dissertations

Analog/mixed-signal (AMS) integrated circuits (ICs) play an essential role in electronic systems by processing analog signals and performing data conversion to bridge the analog physical world and our digital information world.Their ubiquitousness powers diverse applications ranging from smart devices and autonomous cars to crucial infrastructures. Despite such critical importance, conventional design strategies of AMS circuits still follow an expensive and time-consuming manual process and are unable to meet the exponentially-growing productivity demands from industry and satisfy the rapidly-changing design specifications from many emerging applications. Design automation of AMS IC is thus the key to tackling these challenges and has been …

A Collaborative Knowledge-Based Security Risk Assessments Solution Using Blockchains, Tara Thaer Salman

McKelvey School of Engineering Theses & Dissertations

Artificial intelligence and machine learning have recently gained wide adaptation in building intelligent yet simple and proactive security risk assessment solutions. Intrusion identification, malware detection, and threat intelligence are examples of security risk assessment applications that have been revolutionized with these breakthrough technologies. With the increased risk and severity of cyber-attacks and the distributed nature of modern threats and vulnerabilities, it becomes critical to pose a distributed intelligent assessment solution that evaluates security risks collaboratively. Blockchain, as a decade-old successful distributed ledger technology, has the potential to build such collaborative solutions. However, in order to be used for such solutions, …

Domain Specific Computing In Tightly-Coupled Heterogeneous Systems, Anthony Michael Cabrera

McKelvey School of Engineering Theses & Dissertations

Over the past several decades, researchers and programmers across many disciplines have relied on Moores law and Dennard scaling for increases in compute capability in modern processors. However, recent data suggest that the number of transistors per square inch on integrated circuits is losing pace with Moores laws projection due to the breakdown of Dennard scaling at smaller semiconductor process nodes. This has signaled the beginning of a new “golden age in computer architecture” in which the paradigm will be shifted from improving traditional processor performance for general tasks to architecting hardware that executes a class of applications in a …

Investigating Single Precision Floating General Matrix Multiply In Heterogeneous Hardware, Steven Harris

McKelvey School of Engineering Theses & Dissertations

The fundamental operation of matrix multiplication is ubiquitous across a myriad of disciplines. Yet, the identification of new optimizations for matrix multiplication remains relevant for emerging hardware architectures and heterogeneous systems. Frameworks such as OpenCL enable computation orchestration on existing systems, and its availability using the Intel High Level Synthesis compiler allows users to architect new designs for reconfigurable hardware using C/C++. Using the HARPv2 as a vehicle for exploration, we investigate the utility of several of the most notable matrix multiplication optimizations to better understand the performance portability of OpenCL and the implications for such optimizations on this and …

Investigating Patterns In Convolution Neural Network Parameters Using Probabilistic Support Vector Machines, Yuqiu Zhang

McKelvey School of Engineering Theses & Dissertations

Artificial neural networks(ANNs) are recognized as high-performance models for classification problems. They have proved to be efficient tools for many of today's applications like automatic driving, image and video recognition and restoration, big-data analysis. However, high performance deep neural networks have millions of parameters, and the iterative training procedure thus involves a very high computational cost. This research attempts to study the relationships between parameters in convolutional neural networks(CNNs). I assume there exists a certain relation between adjacent convolutional layers and proposed a machine learning model(MLM) that can be trained to represent this relation. The MLM's generalization ability is evaluated …

Decoupling Information And Connectivity Via Information-Centric Transport, Hila Ben Abraham

McKelvey School of Engineering Theses & Dissertations

The power of Information-Centric Networking architectures (ICNs) lies in their abstraction for communication --- the request for named data. This abstraction was popularized by the HyperText Transfer Protocol (HTTP) as an application-layer abstraction, and was extended by ICNs to also serve as their network-layer abstraction. In recent years, network mechanisms for ICNs, such as scalable name-based forwarding, named-data routing and in-network caching, have been widely explored and researched. However, to the best of our knowledge, the impact of this network abstraction on ICN applications has not been explored or well understood. The motivation of this dissertation is to address this …

Polarization Division Multiplexing For Optical Data Communications, Darko Ivanovich

McKelvey School of Engineering Theses & Dissertations

Multiple parallel channels are ubiquitous in optical communications, with spatial division multiplexing (separate physical paths) and wavelength division multiplexing (separate optical wavelengths) being the most common forms. In this research work, we investigate the viability of polarization division multiplexing, the separation of distinct parallel optical communication channels through the polarization properties of light. We investigate polarization division multiplexing based optical communication systems in five distinct parts. In the first part of the work, we define a simulation model of two or more linearly polarized optical signals (at different polarization angles) that are transmitted through a common medium (e.g., air), filtered …

Management And Security Of Multi-Cloud Applications, Lav Gupta

McKelvey School of Engineering Theses & Dissertations

Single cloud management platform technology has reached maturity and is quite successful in information technology applications. Enterprises and application service providers are increasingly adopting a multi-cloud strategy to reduce the risk of cloud service provider lock-in and cloud blackouts and, at the same time, get the benefits like competitive pricing, the flexibility of resource provisioning and better points of presence. Another class of applications that are getting cloud service providers increasingly interested in is the carriers' virtualized network services. However, virtualized carrier services require high levels of availability and performance and impose stringent requirements on cloud services. They necessitate the …

Toward Controllable And Robust Surface Reconstruction From Spatial Curves, Zhiyang Huang

McKelvey School of Engineering Theses & Dissertations

Reconstructing surface from a set of spatial curves is a fundamental problem in computer graphics and computational geometry. It often arises in many applications across various disciplines, such as industrial prototyping, artistic design and biomedical imaging. While the problem has been widely studied for years, challenges remain for handling different type of curve inputs while satisfying various constraints. We study studied three related computational tasks in this thesis. First, we propose an algorithm for reconstructing multi-labeled material interfaces from cross-sectional curves that allows for explicit topology control. Second, we addressed the consistency restoration, a critical but overlooked problem in applying …

Real-Time Reliable Middleware For Industrial Internet-Of-Things, Chao Wang

McKelvey School of Engineering Theses & Dissertations

This dissertation contributes to the area of adaptive real-time and fault-tolerant systems research, applied to Industrial Internet-of-Things (IIoT) systems. Heterogeneous timing and reliability requirements arising from IIoT applications have posed challenges for IIoT services to efficiently differentiate and meet such requirements. Specifically, IIoT services must both differentiate processing according to applications' timing requirements (including latency, event freshness, and relative consistency of each other) and enforce the needed levels of assurance for data delivery (even as far as ensuring zero data loss). It is nontrivial for an IIoT service to efficiently differentiate such heterogeneous IIoT timing/reliability requirements to fit each application, …

Nanopower Analog Frontends For Cyber-Physical Systems, Kenji Aono

McKelvey School of Engineering Theses & Dissertations

In a world that is increasingly dominated by advances made in digital systems, this work will explore the exploiting of naturally occurring physical phenomena to pave the way towards a self-powered sensor for Cyber-Physical Systems (CPS). In general, a sensor frontend can be broken up into a handful of basic stages: transduction, filtering, energy conversion, measurement, and interfacing. One analog artifact that was investigated for filtering was the physical phenomenon of hysteresis induced in current-mode biquads driven near or at their saturation limit. Known as jump resonance, this analog construct facilitates a higher quality factor to be brought about without …

Self-Powered Time-Keeping And Time-Of-Occurrence Sensing, Liang Zhou

McKelvey School of Engineering Theses & Dissertations

Self-powered and passive Internet-of-Things (IoT) devices (e.g. RFID tags, financial assets, wireless sensors and surface-mount devices) have been widely deployed in our everyday and industrial applications. While diverse functionalities have been implemented in passive systems, the lack of a reference clock limits the design space of such devices used for applications such as time-stamping sensing, recording and dynamic authentication. Self-powered time-keeping in passive systems has been challenging because they do not have access to continuous power sources. While energy transducers can harvest power from ambient environment, the intermittent power cannot support continuous operation for reference clocks. The thesis of this …

Development Of Scalable Simulator For Spiking Neural Network, Jae Sang Ha

McKelvey School of Engineering Theses & Dissertations

A neural network simulator for Spiking Neural Network (SNN) is a useful research tool to model brain functions with a computer. With this tool, different parameters can be explored easily compared to using a real brain. For several decades, researchers have developed many software packages and simulators to accelerate research in computational neuroscience. However, despite their advantages, different neural simulators possess different limitations, such as flexibility of choosing different neuron models and scalability of simulators for large numbers of neurons. This paper demonstrates an efficient and scalable spiking neural simulator that is based on growth transform neurons and runs on …

Bio-Inspired Multi-Spectral And Polarization Imaging Sensors For Image-Guided Surgery, Nimrod Missael Garcia

McKelvey School of Engineering Theses & Dissertations

Image-guided surgery (IGS) can enhance cancer treatment by decreasing, and ideally eliminating, positive tumor margins and iatrogenic damage to healthy tissue. Current state-of-the-art near-infrared fluorescence imaging systems are bulky, costly, lack sensitivity under surgical illumination, and lack co-registration accuracy between multimodal images. As a result, an overwhelming majority of physicians still rely on their unaided eyes and palpation as the primary sensing modalities to distinguish cancerous from healthy tissue. In my thesis, I have addressed these challenges in IGC by mimicking the visual systems of several animals to construct low power, compact and highly sensitive multi-spectral and color-polarization sensors. I …

Efficiently And Transparently Maintaining High Simd Occupancy In The Presence Of Wavefront Irregularity, Stephen V. Cole

McKelvey School of Engineering Theses & Dissertations

Demand is increasing for high throughput processing of irregular streaming applications; examples of such applications from scientific and engineering domains include biological sequence alignment, network packet filtering, automated face detection, and big graph algorithms. With wide SIMD, lightweight threads, and low-cost thread-context switching, wide-SIMD architectures such as GPUs allow considerable flexibility in the way application work is assigned to threads. However, irregular applications are challenging to map efficiently onto wide SIMD because data-dependent filtering or replication of items creates an unpredictable data wavefront of items ready for further processing. Straightforward implementations of irregular applications on a wide-SIMD architecture are prone …

Parallel Real-Time Scheduling For Latency-Critical Applications, Jing Li

McKelvey School of Engineering Theses & Dissertations

In order to provide safety guarantees or quality of service guarantees, many of today's systems consist of latency-critical applications, e.g. applications with timing constraints. The problem of scheduling multiple latency-critical jobs on a multiprocessor or multicore machine has been extensively studied for sequential (non-parallizable) jobs and different system models and different objectives have been considered. However, the computational requirement of a single job is still limited by the capacity of a single core. To provide increasingly complex functionalities of applications and to complete their higher computational demands within the same or even more stringent timing constraints, we must exploit the …

Easier Parallel Programming With Provably-Efficient Runtime Schedulers, Robert Utterback

McKelvey School of Engineering Theses & Dissertations

Over the past decade processor manufacturers have pivoted from increasing uniprocessor performance to multicore architectures. However, utilizing this computational power has proved challenging for software developers. Many concurrency platforms and languages have emerged to address parallel programming challenges, yet writing correct and performant parallel code retains a reputation of being one of the hardest tasks a programmer can undertake.

This dissertation will study how runtime scheduling systems can be used to make parallel programming easier. We address the difficulty in writing parallel data structures, automatically finding shared memory bugs, and reproducing non-deterministic synchronization bugs. Each of the systems presented depends …

Underwater Celestial Navigation Using The Polarization Of Light Fields, Samuel Bear Powell

McKelvey School of Engineering Theses & Dissertations

Global-scale underwater navigation presents challenges that modern technology has not solved. Current technologies drift and accumulate errors over time (inertial measurement), are accurate but short-distance (acoustic), or do not sufficiently penetrate the air-water interface (radio and GPS). To address these issues, I have developed a new mode of underwater navigation based on the passive observation of patterns in the polarization of in-water light. These patterns can be used to infer the sun__s relative position, which enables the use of celestial navigation in the underwater environment. I have developed an underwater polarization video camera based on a bio-inspired polarization image sensor …

Learning In The Real World: Constraints On Cost, Space, And Privacy, Matt J. Kusner

McKelvey School of Engineering Theses & Dissertations

The sheer demand for machine learning in fields as varied as: healthcare, web-search ranking, factory automation, collision prediction, spam filtering, and many others, frequently outpaces the intended use-case of machine learning models. In fact, a growing number of companies hire machine learning researchers to rectify this very problem: to tailor and/or design new state-of-the-art models to the setting at hand.

However, we can generalize a large set of the machine learning problems encountered in practical settings into three categories: cost, space, and privacy. The first category (cost) considers problems that need to balance the accuracy of a machine learning model …

Global Edf Scheduling For Parallel Real-Time Tasks, Jing Li

McKelvey School of Engineering Theses & Dissertations

As multicore processors become ever more prevalent, it is important for real-time programs to take advantage of intra-task parallelism in order to support computation-intensive applications with tight deadlines. In this thesis, we consider the Global Earliest Deadline First (GEDF) scheduling policy for task sets consisting of parallel tasks. Each task can be represented by a directed acyclic graph (DAG) where nodes represent computational work and edges represent dependences between nodes. In this model, we prove that GEDF provides a capacity augmentation bound of 4-2/m and a resource augmentation bound of 2-1/m. The capacity augmentation bound acts as a linear-time schedulability …

Atomic Transfer For Distributed Systems, Haraldur Darri Thorvaldsson

McKelvey School of Engineering Theses & Dissertations

Building applications and information systems increasingly means dealing with concurrency and faults stemming from distribution of system components. Atomic transactions are a well-known method for transferring the responsibility for handling concurrency and faults from developers to the software's execution environment, but incur considerable execution overhead. This dissertation investigates methods that shift some of the burden of concurrency control into the network layer, to reduce response times and increase throughput. It anticipates future programmable network devices, enabling customized high-performance network protocols.

We propose Atomic Transfer (AT), a distributed algorithm to prevent race conditions due to messages crossing on a path of …