Computer Engineering Commons^™

Pafex: Compiler-Based Floating-Point Exception Detection For Gpu Kernels With Papispecific Software-Defined Events, Samin Islam, Shirley V. Moore, Christoph Q. Lauter

Graduate Student Papers (CS)

As high-performance computing becomes progressively heterogeneous, the dependence upon vendor specific tools for numerical correctness has become an impediment to portability. Although modern GPUs comply with the IEEE 754 standard, the lack of practical native hardware support to raise and handle exceptions (special values like ±∞ or NaN) is a well-known architectural limitation. To embed portable numerical correctness across heterogeneous systems, we propose an architecture agnostic prototype based on LLVM-compiler infrastructure. This framework detects floating-point exceptions in GPU kernels at the Intermediate Representation (IR) level, instrumenting both device code and host code, strictly complying with the 2019 IEEE 754 standard. …

Gpu-Based Electromagnetic Microwave Tomography For Brain Imaging And Stroke Detection, Pablo Sotelo Torres

Open Access Theses & Dissertations

Each year, an estimated 795,000 people in the U.S. suffer a stroke, with approximately 610,000 being first-time cases. Of these, 87% are ischemic strokes, while the remaining 13% are hemorrhagic. Current imaging methods, such as Computed Tomography (CT), Positron Emission Tomography (PET), and Magnetic Resonance Imaging (MRI), provide useful information into brain tissue properties; while each technique has its advantages, they remain expensive, non-portable, and often too slow for emergency bedside or in-ambulance use. Electromagnetic Microwave Tomography (EMT) offers a promising alternative: an affordable, portable, rapid, and safe method for stroke detection. By contrasting dielectric properties between healthy and affected …

3d Face Modeling From 2d Images Using Deep Neural Networks, Mario Alberto De La Cruz Armendariz

Open Access Theses & Dissertations

Applications of 3D face reconstruction include biometric authentication, personalized avatars and digital identity, medical visualization, forensic analysis, and broader human-computer interaction. We propose an approach to 3D face reconstruction that can generate a fully textured 3D facial model using only two grayscale images: a front view and a profile view of the subject. Once trained, the system can perform the reconstruction autonomously without manual intervention. Unlike traditional methods requiring multi-camera setups, depth sensors, or cloud-based processing, the proposed approach runs fully offline on a standard CPU, supporting dynamic execution across CPU cores and eliminating the need for a dedicated GPU. …

Efficient Adaptive Spline-Based Path Planning For In-Space Servicing, Assembly, And Manufacturing Applications, Christian Lozoya

Open Access Theses & Dissertations

Autonomous robotic systems operating in cluttered and partially observed environments require trajectory generation methods that produce smooth and dynamically feasible motion while reacting to locally sensed obstacles. This requirement is especially pronounced for free-flyer and in-space servicing, assembly, and manufacturing (ISAM) platforms, where onboard sensing is sparse, global environmental information is unavailable, and communication or computational resources are constrained. In such settings, motion plans must be updated online using incomplete and rapidly changing local observations, while avoiding excessive replanning that can lead to oscillatory or unstable behavior. Many existing approaches either rely on dense optimization over extended horizons, which is …

Further Insights Into The Network Link Outlier Factor's (Nlof) Light-Load Penalty, Sunday Oluwaleke Ogundele

Open Access Theses & Dissertations

This research investigates the performance of the Network Link Outlier Factor with Most Likely Links (NLOF:MLL), under varying network load conditions. Earlier studies reported that the NLOF:MLL algorithm experienced a noticeable drop in fault-localization accuracy when operating in lightly loaded networks. To further examine this limitation, 240 experiments were carried out to observe how the algorithm responds as overall network load increases. The evaluation focused on the classification performance metrics: precision, recall, and F1-score. The results show that NLOF:MLL’s effectiveness improves as network load increases but that the rate of improvement slows progressively, eventually stabilizing in a pattern consistent with …

Digital Twin For Real-Time Monitoring And Control Of Conveyor Systems Using Flexsim, Ai And Plc Integration, Jose Francisco Arvizu Astorga

Open Access Theses & Dissertations

Modern manufacturing is making significant advancements by innovating and automating most processes. However, a major challenge remains: systems are constantly evolving and becoming more complex to analyze. Fortunately, a powerful tool can help, Digital Twin (DT) technology. This technology enables the analysis and optimization of processes like never before. A Digital Twin is a real-time virtual model of a physical system that continuously up dates with live data. One of its greatest features is the ability to create infinite scenarios, allowing hundreds of configurations to be tested virtually, risk-free, and without making any real-world changes that could disrupt ongoing operations. …

Multi-Material Cell Clipping On The Gpu, Melanie Cassidy Walsmith

Open Access Theses & Dissertations

The process of clipping a multi-material cell finds diverse applications across fields such as numerical simulations and computer graphics visualization. Computational fluid dynamics problem often combines multiple materials with different physical properties. The interfaces between those materials may be a part of the solution and evolve in time and can be non-aligned with the mesh. When volume conservation is crucial, interface reconstruction methods are used to approximate such material interfaces. They involve multiple steps, one of which is the process known as clipping. Clipping consists of intersecting and cutting a given cell with a material interface (represented by a line …

Towards Mitigation Of The Light Network Load Performance Penalty Of The Network Link Outlier Factor (Nlof), Jaime Merin Guzman

Open Access Theses & Dissertations

Detecting and localizing faults in communication networks is critical to maintaining reliable and efficient network operations. The Network Link Outlier Factor with Most Likely Link (NLOF: MLL) algorithm has demonstrated its potential to automate this task but suffers from significant performance degradation under low network load conditions, where limited network flow data reduces its ability to localize faults. This thesis proposes and evaluates the performance of a synthetic traffic generation algorithm to be used with NLOF:MLL. This algorithm strategically injects synthetic flows that supplement the insufficient real network flows, thereby improving NLOF:MLL's performance under low-load conditions. Specifically, we select network …

Inference Per Joule: A Performance Metric For Artificial Intelligence In Space Applications, Eduardo Macias Zugasti

Open Access Theses & Dissertations

The use of artificial intelligence (AI) has grown exponentially in recent years. This growth is driven in part by the significant advancements in computing capabilities, which have also increased exponentially. Computers have not only become more powerful but also smaller in size, thanks to the evolution of transistor technology. These developments have enabled AI to become a widely accessible tool, even in recreational activities such as image creation and entertainment videos.

More recently, the use of AI has extended to space applications, where it can enhance and optimize various tasks. However, space conditions pose significant challenges for conventional computers due …

Control Of Industrial Robots Based On Artificial Intelligence, Bryan Lara Medrano

Open Access Theses & Dissertations

Industrial robots are vital in developing smart factories, creating the need for more efficient and modern control systems. As a result, investigators and scholars are dedicating great effort to advancing this field et al. [27]. Literature showcases significant progress in various areas, including the control of articulated arms and advancements in human-robot interfaces, self-decision-making, object recognition, decision-making, and routing planning. This manuscript describes a novel technique for predicting the movement of a robotic arm based on artificial neural networks. We have implemented an artificial intelligence method based on artificial neural networks to analyze the possible routing of a robotic arm …

General Purpose Gpu Benchmarks For Neural Networks, Jose Maria Granados

Open Access Theses & Dissertations

Neural networks are a field of computing experiencing a rise in popularity in recent years due to the utilization of graphics processing units as their computational centerpiece. The lack of neural network benchmarks for the open-source Nyuzi architecture, a developing general-purpose processor with graphical processing capabilities, is the focus of this thesis. This work aims to determine whether Nyuziâ??s performance counters and traceable events suffice for performance tuning of neural network implementations. Given the mathematical intensity of neural networks, a strong emphasis is placed on events related to arithmetic instructions. Experimenting with neural network implementations in C and C++, existent …

Mav Localization In Gps-Denied Environments And Synthetic Data Collection In Challenging Simulated Conditions, Julio A. Reyes Munoz

Open Access Theses & Dissertations

The development of unmanned aerial systems presents an opportunity for conducting industrial inspections in environments where traditional navigation systems, such as the Global Navigation Satellite System (GNSS), are compromised. This dissertation investigates the implementation of a micro aerial vehicle (MAV) capable of autonomous data acquisition in complex, GNSS-degraded industrial settings. The primary challenge addressed is the robust localization of MAVs, a critical aspect in ensuring reliable operation under varying and uncertain environmental conditions.

The work is divided into two main parts. The first part focuses on the design and integration of a MAV system specifically for power plant inspections in …

Understanding The Limits Of Deep Packet Inspection For Network Traffic Classification, Herman Ramey

Open Access Theses & Dissertations

We present our human network application labeling system that contributes a new level of distinction between the network traffic that should be labeled from the network traffic that should not be labeled. This distinction improves the label accuracy of the training data set produced from the human labeled data and will subsequently improve the performance of supervised machine learning classifiers used for network traffic classification. This system also allows for the human network user to label traffic, with little effort, in a manner consistent with normal network usage, i.e., no need for a contrived experiment. Lastly, we use human supplied …

Communication Security In Smart Home Iot And Scada Applications, Ismael Holguin

Open Access Theses & Dissertations

The advancements in technology have enabled continuous integration of consumer electronicdevices and sensors to the Internet, creating Internet of Things (IoT) environments for various applications. IoT communication protocols play a significant role in these environments and greatly vary depending on the type of application. In this work, we consider communication protocols in Smart Home IoT and SCADA environments, focus on their security features, and investigate methods to enhance their security. Most of these communication protocols were designed to meet specific communication goals, where security was not a requirement or priority during design. However, due to a shift in the security …

Addressing Current Security And Privacy Limitations In Iot Applications With Differential Privacy, Nicholas Anthony Lopez

Open Access Theses & Dissertations

Over the last decade, billions of devices have been developed to run on the Internet of Things (IoT) domain. IoT has been embedded into everyday tasks so much that the number of devices has outpaced the global population. Some of the main areas of IoT include healthcare, industry, communications, transportation, logistics, and environmental awareness, to name a few. With this greater demand for devices comes an even greater demand for privacy and security. The three main principles of security include confidentiality, integrity, and availability. However, there is no “one size fits all” solution regarding privacy or security. When applied across …

Optimized Learning Using Fuzzy-Inference-Assisted Algorithms For Deep Learning, Miroslava Barua

Open Access Theses & Dissertations

For years, researchers in Artificial Intelligence (AI) and Deep Learning (DL) observed that performance of a Deep Learning Network (DLN) could be improved by using larger and larger datasets coupled with complex network architectures. Although these strategies yield remarkable results, they have limits, dictated by data quantity and quality, rising costs by the increased computational power, or, more frequently, by long training times on networks that are very large. Training DLN requires laborious work involving multiple layers of densely connected neurons, updates to millions of network parameters, while potentially iterating thousands of times through millions of entries in a big …

Online/Incremental Learning To Mitigate Concept Drift In Network Traffic Classification, Alberto R. De La Rosa

Open Access Theses & Dissertations

Communication networks play a large role in our everyday lives. COVID19 pandemic in 2020 highlighted their importance as most jobs had to be moved to remote work environments. It is possible that the spread of the virus, the death toll, and the economic consequences would have been much worse without communication networks. To remove sole dependence on one equipment vendor, networks are heterogeneous by design. Due to this, as well as their increasing size, network management has become overwhelming for network managers. For this reason, automating network management will have a significant positive impact. Machine learning and software defined networking …

Intelligent Autonomous Inspections Using Deep Learning And Detection Markers, Alejandro Martinez Acosta

Open Access Theses & Dissertations

Inspection of industrial and scientific facilities is a crucial task that must be performed regularly. These inspections tasks ensure that the facilityâ??s structure is in safe operational conditions for humans. Furthermore,the safe operation of industrial machinery, is dependent on the conditions of the environment. For safety reasons, inspections for both structural integrity and equipment is often manually performed by operators or technicians. Naturally, this is often a tedious and laborious task. Additionally, buildings and structures frequently contain hard to reach or dangerous areas, which leads to the harm, injury or death of humans. Autonomous robotic systems offer an attractive solution …

Security Analysis And Implementation Of Dnp3 Multilayer Protocol For Secure And Safe Communication In Scada Systems, Isaac Monroy

Open Access Theses & Dissertations

When SCADA systems were first introduced into society, a lot of manpower was required for monitoring and controlling devices within critical infrastructures. With the increasing demand for services and growing systems, a need arose to automate the monitoring and controlling tasks. This led to introduction of networks into SCADA systems to enhance monitoring and control capabilities, that can scale with system size and requirements. But this introduction of network layer along with its advantages, also introduced a new threat surface which exposed multiple vulnerabilities within the system that can exploited to launch attacks, that led to the integration of security …

Productivity And Quality Evaluation In Assembly Using Collaborative Robots, Carlos F. Manzanares Vega

Open Access Theses & Dissertations

In Industry 4.0, various technologies have been applied to achieve automation for traditional manufacturing and practices. For this reason, Smart Manufacturing (SM) environments utilize collaborative robots for process optimization by integrating the Internet of Things (IoT). Cobots are equipped with sensors and/or other devices to be able to transmit data in real-time while performing their tasks. Consequently, such SM implementations improves the decision making and business development, such as supply chain and operations, by sharing real-time data from a plant operational level. The collaborative robots are also designed to safely interact and collaborate with humans to perform tasks and optimize …

Miner-Town: Self-Driving Robotics Testbed For Vehicle-To-Grid Simulation, Carlos Adolfo Cortes Pliego

Open Access Theses & Dissertations

Autonomous vehicles and Vehicle-to-Grid (V2G) technology bring promising implications in boosting energy efficiency, helping the environment, improving our productivity, and have the potential to stabilize the grid during peak times and reduce car accidents. However, implementing and testing these complex novel technologies in the real world comes with high risks and investment. For these reasons, there is the need to research, test, and validate these theories in a compact and controlled environment at minimal cost. This thesis presents a modular autonomous vehicle testbed for the exploration of Vehicle-to-Grid and charging activities in pedestrian filled environments such as a University campus. …

Efficient Approaches To Steady State Detection In Multivariate Systems, Honglun Xu

Open Access Theses & Dissertations

Steady state detection is critically important in many engineering fields such as fault detection and diagnosis, process monitoring and control. However, most of the existing methods are designed for univariate signals. In this dissertation, we proposed an efficient online steady state detection method for multivariate systems through a sequential Bayesian partitioning approach. The signal is modeled by a Bayesian piecewise constant mean and covariance model, and a recursive updating method is developed to calculate the posterior distributions analytically. The duration of the current segment is utilized to test the steady state. Insightful guidance is provided for hyperparameter selection. The effectiveness …

Development Of An Automated Electronic Prototyping System, Cesar Yahir Sanchez Zambrano

Open Access Theses & Dissertations

Prototyping systems with interconnected components can be a time and resource expensive process. The process consists of three main phases (design, build and analysis) with each having their own associated cost. For the case of electronic circuits, the building phase is the costliest phase among the three, being prone to human errors which causes the circuit to fail. All three phases of the prototyping process are important. However, often a disproportionate amount of time is spent on the build phase due to the difficulty of making and troubleshooting circuits by hand. In this thesis we will discuss a system that …

Addressing Security And Privacy Issues By Analyzing Vulnerabilities In Iot Applications, Francsico Javier Candelario Burgoa

Open Access Theses & Dissertations

The Internet of Things (IoT) environment has been expanding rapidly for the past few years into several areas of our lives, from factories, to stores and even into our own homes. All these new devices in our homes make our day-to-day lives easier and more comfortable with less effort on our part, converting our simple houses into smart homes. This increase in inter-connectivity brings multiple benefits including the improvement in energy efficiency in our homes, however it also brings with it some potential dangers since more points of connection mean more potential vulnerabilities in our grid. These vulnerabilities bring security …

The Network Link Outlier Factor (Nlof) For Localizing Network Faults, Christopher Mendoza

Open Access Theses & Dissertations

This work presents the Network Link Outlier Factor (NLOF), a data analytics pipeline for network fault detection and localization solution that consists of four stages. In the first stage, flow record throughput values are clustered in two sub-stages: using Density-Based Spatial Clustering of Applications with Noise (DBSCAN) and then a novel domain-specific ThroughPut Cluster (TPCluster) technique. In the second stage, Flow Outlier Factor (FOF) scores are computed for each flow. In the third stage, flows are traced onto the network. Finally, in the fourth stage, each link is given a Network Link Outlier Factor (NLOF) score which is the ratio …

Hardware For Quantized Mixed-Precision Deep Neural Networks, Andres Rios

Open Access Theses & Dissertations

Recently, there has been a push to perform deep learning (DL) computations on the edge rather than the cloud due to latency, network connectivity, energy consumption, and privacy issues. However, state-of-the-art deep neural networks (DNNs) require vast amounts of computational power, data, and energyâ??resources that are limited on edge devices. This limitation has brought the need to design domain-specific architectures (DSAs) that implement DL-specific hardware optimizations. Traditionally DNNs have run on 32-bit floating-point numbers; however, a body of research has shown that DNNs are surprisingly robust and do not require all 32 bits. Instead, using quantization, networks can run on …

Geometric Analysis Leads To Adversarial Teaching Of Cybersecurity, Christian Servin, Olga Kosheleva, Vladik Kreinovich

Departmental Technical Reports (CS)

As time goes, our civilization becomes more and more dependent on computers and therefore, more and more vulnerable to cyberattacks. Because of this threat, it is very important to make sure that computer science students -- tomorrow's computer professionals -- are sufficiently skilled in cybersecurity. In this paper, we analyze the need for teaching cybersecurity from the geometric viewpoint. We show that the corresponding geometric analysis leads to adversarial teaching -- an empirically effective but not-well-theoretically-understood approach, when the class is divided into sparring mini-teams that try their best to attack each other and defend from each other. Thus, our …

We Need Fuzzy Techniques To Design Successful Human-Like Robots, Vladik Kreinovich, Olga Kosheleva, Laxman Bokati

Departmental Technical Reports (CS)

In this chapter, we argue that to make sure that human-like robots exhibit human-like behavior, we need to use fuzzy techniques -- and we also provide details of this usage. The chapter is intended both for researchers and practitioners who are very familiar with fuzzy techniques and also for researchers and practitioners who do not know these techniques -- but who are interested in designing human-like robots.

How To Train A-To-B And B-To-A Neural Networks So That The Resulting Transformations Are (Almost) Exact Inverses, Paravee Maneejuk, Torben Peters, Claus Brenner, Vladik Kreinovich

Departmental Technical Reports (CS)

In many practical situations, there exist several representations, each of which is convenient for some operations, and many data processing algorithms involve transforming back and forth between these representations. Many such transformations are computationally time-consuming when performed exactly. So, taking into account that input data is usually only 1-10% accurate anyway, it makes sense to replace time-consuming exact transformations with faster approximate ones. One of the natural ways to get a fast-computing approximation to a transformation is to train the corresponding neural network. The problem is that if we train A-to-B and B-to-A networks separately, the resulting approximate transformations are …

A Fully Lexicographic Extension Of Min Or Max Operation Cannot Be Associative, Olga Kosheleva, Vladik Kreinovich

Departmental Technical Reports (CS)

In many applications of fuzzy logic, to estimate the degree of confidence in a statement A&B, we take the minimum min(a,b) of the expert's degrees of confidence in the two statements A and B. When a < b, then an increase in b does not change this estimate, while from the commonsense viewpoint, our degree of confidence in A&B should increase. To take this commonsense idea into account, Ildar Batyrshin and colleagues proposed to extend the original order in the interval [0,1] to a lexicographic order on a larger set. This idea works for expressions of the type A&B, so maybe we can extend it to more general expressions? In this paper, we show that such an extension, while theoretically possible, would violate another commonsense requirement -- associativity of the "and"-operation. A similar negative result is proven for lexicographic extensions of the maximum operation -- that estimates the expert's degree of confidence in a statement A\/B.