Computer Engineering Commons^™

What Broke Where For Distributed And Parallel Applications — A Whodunit Story, Subrata Mitra

Open Access Dissertations

Detection, diagnosis and mitigation of performance problems in today's large-scale distributed and parallel systems is a difficult task. These large distributed and parallel systems are composed of various complex software and hardware components. When the system experiences some performance or correctness problem, developers struggle to understand the root cause of the problem and fix in a timely manner. In my thesis, I address these three components of the performance problems in computer systems. First, we focus on diagnosing performance problems in large-scale parallel applications running on supercomputers. We developed techniques to localize the performance problem for root-cause analysis. Parallel applications, …

Approximate Computing: An Integrated Cross-Layer Framework, Swagath Venkataramani

Open Access Dissertations

A new design approach, called approximate computing (AxC), leverages the flexibility provided by intrinsic application resilience to realize hardware or software implementations that are more efficient in energy or performance. Approximate computing techniques forsake exact (numerical or Boolean) equivalence in the execution of some of the application’s computations, while ensuring that the output quality is acceptable. While early efforts in approximate computing have demonstrated great potential, they consist of ad hoc techniques applied to a very narrow set of applications, leaving in question the applicability of approximate computing in a broader context.

The primary objective of this thesis is to …

Improving Programmability And Performance For Scientific Applications, Chenyang Liu

Open Access Dissertations

With modern advancements in hardware and software technology scaling towards new limits, our compute machines are reaching new potentials to tackle more challenging problems. While the size and complexity of both the problems and solutions increases, the programming methodologies must remain at a level that can be understood by programmers and scientists alike. In our work, this problem is encountered when developing an optimized framework to best exploit the semantic properties of a finite-element solver. In efforts to address this problem, we explore programming and runtime models which decouple algorithmic complexity, parallelism concerns, and hardware mapping. We build upon these …

Interactive Logical Analysis Of Planning Domains, Rajesh Kalyanam

Open Access Dissertations

Humans exhibit a significant ability to answer a wide range of questions about previously unencountered planning domains, and leverage this ability to construct “general-purpose'' solution plans for the domain.

The long term vision of this research is to automate this ability, constructing a system that utilizes reasoning to automatically verify claims about a planning domain. The system would use this ability to automatically construct and verify a generalized plan to solve any planning problem in the domain. The goal of this thesis is to start with baseline results from the interactive verification of claims about planning domains and develop the …

Three Dimensional Moving Pictures With A Single Imager And Microfluidic Lens, Chao Liu

Open Access Dissertations

Three-dimensional movie acquisition and corresponding depth data is commonly generated from multiple cameras and multiple views. This technology has high cost and large size which are limitations for medical devices, military surveillance and current consumer products such as small camcorders and cell phone movie cameras. This research result shows that a single imager, equipped with a fast-focus microfluidic lens, produces a highly accurate depth map. On test material, the depth is found to be an average Root Mean Squared Error (RMSE) of 3.543 gray level steps (1.38\%) accuracy compared to ranging data. The depth is inferred using a new Extended …

Exploring Spin-Transfer-Torque Devices And Memristors For Logic And Memory Applications, Zoha Pajouhi

Open Access Dissertations

As scaling CMOS devices is approaching its physical limits, researchers have begun exploring newer devices and architectures to replace CMOS.

Due to their non-volatility and high density, Spin Transfer Torque (STT) devices are among the most prominent candidates for logic and memory applications. In this research, we first considered a new logic style called All Spin Logic (ASL). Despite its advantages, ASL consumes a large amount of static power; thus, several optimizations can be performed to address this issue. We developed a systematic methodology to perform the optimizations to ensure stable operation of ASL.

Second, we investigated reliable design of …

Improving The Resilience Of Cyber-Physical Systems Under Strategic Adversaries, Paul Wood

Open Access Dissertations

Renewable energy resources challenge traditional energy system operations by substituting the stability and predictability of fossil fuel based generation with the unreliability and uncertainty of wind and solar power. Rising demand for green energy drives grid operators to integrate sensors, smart meters, and distributed control to compensate for this uncertainty and improve the operational efficiency of the grid. Real-time negotiations enable producers and consumers to adjust power loads during shortage periods, such as an unexpected outage or weather event, and to adapt to time-varying energy needs. While such systems improve grid performance, practical implementation challenges can derail the operation of …

A Holistic Approach To Lowering Latency In Geo-Distributed Web Applications, Shankaranarayanan Puzhavakath Narayanan

Open Access Dissertations

User perceived end-to-end latency of web applications have a huge impact on the revenue for many businesses. The end-to-end latency of web applications is impacted by: (i) User to Application server (front-end) latency which includes downloading and parsing web pages, retrieving further objects requested by javascript executions; and (ii) Application and storage server(back-end) latency which includes retrieving meta-data required for an initial rendering, and subsequent content based on user actions.

Improving the user-perceived performance of web applications is challenging, given their complex operating environments involving user-facing web servers, content distribution network (CDN) servers, multi-tiered application servers, and storage servers. Further, …

Energy Efficiency In Data Collection Wireless Sensor Networks, Miquel Andres Navarro Patino

Open Access Dissertations

This dissertation studies the problem of energy efficiency in resource constrained and heterogeneous wireless sensor networks (WSNs) for data collection applications in real-world scenarios. The problem is addressed from three different perspectives: network routing, node energy profiles, and network management. First, the energy efficiency in a WSN is formulated as a load balancing problem, where the routing layer can diagnose and exploit the WSN topology redundancy to reduce the data traffic processed in critical nodes, independent of their hardware platform, improving their energy consumption and extending the network lifetime. We propose a new routing strategy that extends traditional cost-based routing …

Representation And Control Of Coordinated-Motion Tasks For Human-Robot Systems, Hyungju Andy Park

Open Access Dissertations

It is challenging for robots to perform various tasks in a human environment. This is because many human-centered tasks require coordination in both hands and may often involve cooperation with another human. Although human-centered tasks require different types of coordinated movements, most of the existing methodologies have focused only on specific types of coordination. This thesis aims at the description and control of coordinated-motion tasks for human-robot systems; i.e., humanoid robots as well as multi-robot and human-robot systems.

First, for bimanually coordinated-motion tasks in dual-manipulator systems, we propose the Extended-Cooperative-Task-Space (ECTS) representation, which extends the existing Cooperative-Task-Space (CTS) representation based …

Learning In Vision And Robotics, Daniel P. Barrett

Open Access Dissertations

I present my work on learning from video and robotic input. This is an important problem, with numerous potential applications. The use of machine learning makes it possible to obtain models which can handle noise and variation without explicitly programming them. It also raises the possibility of robots which can interact more seamlessly with humans rather than only exhibiting hard-coded behaviors. I will present my work in two areas: video action recognition, and robot navigation. First, I present a video action recognition method which represents actions in video by sequences of retinotopic appearance and motion detectors, learns such models automatically …

Grounding Robot Motion In Natural Language And Visual Perception, Scott Alan Bronikowski

Open Access Dissertations

The current state of the art in military and first responder ground robots involves heavy physical and cognitive burdens on the human operator while taking little to no advantage of the potential autonomy of robotic technology. The robots currently in use are rugged remote-controlled vehicles. Their interaction modalities, usually utilizing a game controller connected to a computer, require a dedicated operator who has limited capacity for other tasks.

I present research which aims to ease these burdens by incorporating multiple modes of robotic sensing into a system which allows humans to interact with robots through a natural-language interface. I conduct …

Content-Based Image Analysis With Applications To The Multifunction Printer Imaging Pipeline And Image Databases, Cheng Lu

Open Access Dissertations

Image understanding is one of the most important topics for various applications. Most of image understanding studies focus on content-based approach while some others also rely on meta data of images. Image understanding includes several sub-topics such as classification, segmentation, retrieval and automatic annotation etc., which are heavily studied recently. This thesis proposes several new methods and algorithms for image classification, retrieval and automatic tag generation. The proposed algorithms have been tested and verified in multiple platforms. For image classification, our proposed method can complete classification in real-time under hardware constraints of all-in-one printer and adaptively improve itself by online …

Generalized Techniques For Using System Execution Traces To Support Software Performance Analysis, Thelge Manjula Peiris

Open Access Dissertations

This dissertation proposes generalized techniques to support software performance analysis using system execution traces in the absence of software development artifacts such as source code. The proposed techniques do not require modifications to the source code, or to the software binaries, for the purpose of software analysis (non-intrusive). The proposed techniques are also not tightly coupled to the architecture specific details of the system being analyzed. This dissertation extends the current techniques of using system execution traces to evaluate software performance properties, such as response times, service times. The dissertation also proposes a novel technique to auto-construct a dataflow model …

Architectural Techniques To Extend Multi-Core Performance Scaling, Hamza Bin Sohail

Open Access Dissertations

Multi-cores have successfully delivered performance improvements over the past decade; however, they now face problems on two fronts: power and off-chip memory bandwidth. Dennard's scaling is effectively coming to an end which has lead to a gradual increase in chip power dissipation. In addition, sustaining off-chip memory bandwidth has become harder due to the limited space for pins on the die and greater current needed to drive the increasing load . My thesis focuses on techniques to address the power and off-chip memory bandwidth challenges in order to avoid the premature end of the multi-core era. ^ In the first …

Captured Open Book Image De-Warping And Shading Correction Using 3d Depth Information, Chyuan-Tyng Wu

Open Access Dissertations

Various three dimensional (3D) measuring or capturing devices are introduced to the society recently, and there are abundant possibilities that we can take advantage of this new technology. In this research, we worked on one useful application: to correct the distortion due to the curved shape of the pages of an open book in captured images using of depth information. This work is relevant to camera-based capture devices that can use a projector to cast structured light patterns to provide depth information. In order to improve the visual quality of captured documents, we established our algorithm from two perspectives. First, …

Semcache: Semantics-Aware Caching For Efficient Gpu Offloading, Nabeel Al-Saber

Open Access Dissertations

Graphical Processing Units (GPUs) offer massive, highly-efficient parallelism, making them an attractive target for computation-intensive applications. However, GPUs have a separate memory space which introduces the complexity of manually handling explicit data movements between GPU and CPU memory spaces. Although GPU kernels/libraries have made it easy to improve application performance by offloading computation to GPUs, unfortunately it is very difficult to manually optimize CPU-GPU communication between multiple kernel invocations to avoid redundant communication when using these kernels with complex applications. ^ In this thesis, we introduce SemCache, a semantics-aware GPU cache that automatically manages CPU-GPU communication in addition to optimizing …

Black-Box Printer Models And Their Applications, Yanling Ju

Open Access Dissertations

In the electrophotographic printing process, the deposition of toner within the area of a given printer addressable pixel is strongly influenced by the values of its neighboring pixels. The interaction between neighboring pixels, which is commonly referred to as dot-gain, is complicated. The printer models which are developed according to a pre-designed test page can either be embedded in the halftoning algorithm, or used to predict the printed halftone image at the input to an algorithm being used to assess print quality. In our research, we examine the potential influence of a larger neighborhood (45?45) of the digital halftone image …

Improving Capacity-Performance Tradeoffs In The Storage Tier, Eric P. Villasenor

Open Access Dissertations

Data-set sizes are growing. New techniques are emerging to organize and analyze these data-sets. There is a key access pattern emerging with these new techniques, large sequential file accesses. The trend toward bigger files exists to help amortize the cost of data accesses from the storage layer, as many workloads are recognized to be I/O bound. The storage layer is widely recognized as the slowest layer in the system. This work focuses on the tradeoff one can make with that storage capacity to improve system performance. ^ Capacity can be leveraged for improved availability or improved performance. This tradeoff is …

Assessment Of High-Fidelity Collision Models In The Direct Simulation Monte Carlo Method, Andrew Brian Weaver

Open Access Dissertations

Advances in computer technology over the decades has allowed for more complex physics to be modeled in the DSMC method. Beginning with the first paper on DSMC in 1963, 30,000 collision events per hour were simulated using a simple hard sphere model. Today, more than 10 billion collision events can be simulated per hour for the same problem. Many new and more physically realistic collision models such as the Lennard-Jones potential and the forced harmonic oscillator model have been introduced into DSMC. However, the fact that computer resources are more readily available and higher-fidelity models have been developed does not …

Accelerating Mpi Collective Communications Through Hierarchical Algorithms With Flexible Inter-Node Communication And Imbalance Awareness, Benjamin Scott Parsons

Open Access Dissertations

This work presents and evaluates algorithms for MPI collective communication operations on high performance systems. Collective communication algorithms are extensively investigated, and a universal algorithm to improve the performance of MPI collective operations on hierarchical clusters is introduced. This algorithm exploits shared-memory buffers for efficient intra-node communication while still allowing the use of unmodified, hierarchy-unaware traditional collectives for inter-node communication. The universal algorithm shows impressive performance results with a variety of collectives, improving upon the MPICH algorithms as well as the Cray MPT algorithms. Speedups average 15x - 30x for most collectives with improved scalability up to 65536 cores.^ Further …

Digital Provenance - Models, Systems, And Applications, Salmin Sultana

Open Access Dissertations

Data provenance refers to the history of creation and manipulation of a data object and is being widely used in various application domains including scientific experiments, grid computing, file and storage system, streaming data etc. However, existing provenance systems operate at a single layer of abstraction (workflow/process/OS) at which they record and store provenance whereas the provenance captured from different layers provide the highest benefit when integrated through a unified provenance framework. To build such a framework, a comprehensive provenance model able to represent the provenance of data objects with various semantics and granularity is the first step. In this …

Shape From Inconsistent Silhouette: Reconstruction Of Objects In The Presence Of Segmentation And Camera Calibration Error, Amy Tabb

Open Access Dissertations

Silhouettes are useful features to reconstruct the object shape when the object is textureless or the shape classes of objects are unknown. In this dissertation, we explore the problem of reconstructing the shape of challenging objects from silhouettes under real-world conditions such as the presence of silhouette and camera calibration error. This problem is called the Shape from Inconsistent Silhouettes problem. A psuedo-Boolean cost function is formalized for this problem, which penalizes differences between the reconstruction images and the silhouette images, and the Shape from Inconsistent Silhouette problem is cast as a psuedo-Boolean minimization problem. We propose a memory and …

Image Analysis Using Visual Saliency With Applications In Hazmat Sign Detection And Recognition, Bin Zhao

Open Access Dissertations

Visual saliency is the perceptual process that makes attractive objects "stand out" from their surroundings in the low-level human visual system. Visual saliency has been modeled as a preprocessing step of the human visual system for selecting the important visual information from a scene. We investigate bottom-up visual saliency using spectral analysis approaches. We present separate and composite model families that generalize existing frequency domain visual saliency models. We propose several frequency domain visual saliency models to generate saliency maps using new spectrum processing methods and an entropy-based saliency map selection approach. A group of saliency map candidates are then …

Functional Programming Abstractions For Weakly Consistent Systems, Sivaramakrishnan Krishnamoorthy Chandrasekaran

Open Access Dissertations

In recent years, there has been a wide-spread adoption of both multicore and cloud computing. Traditionally, concurrent programmers have relied on the underlying system providing strong memory consistency, where there is a semblance of concurrent tasks operating over a shared global address space. However, providing scalable strong consistency guarantees as the scale of the system grows is an increasingly difficult endeavor. In a multicore setting, the increasing complexity and the lack of scalability of hardware mechanisms such as cache coherence deters scalable strong consistency. In geo-distributed compute clouds, the availability concerns in the presence of partial failures prohibit strong consistency. …

Embodied Interaction With Visualization And Spatial Navigation In Time-Sensitive Scenarios, Yu-Ting Li

Open Access Dissertations

Paraphrasing the theory of embodied cognition, all aspects of our cognition are determined primarily by the contextual information and the means of physical interaction with data and information. In hybrid human-machine systems involving complex decision making, continuously maintaining a high level of attention while employing a deep understanding concerning the task performed as well as its context are essential. Utilizing embodied interaction to interact with machines has the potential to promote thinking and learning according to the theory of embodied cognition proposed by Lakoff. Additionally, the hybrid human-machine system utilizing natural and intuitive communication channels (e.g., gestures, speech, and body …

Automatic Translation Of Non-Repetitive Openmp To Mpi, Fahed Jubair

Open Access Dissertations

Cluster platforms with distributed-memory architectures are becoming increasingly available low-cost solutions for high performance computing. Delivering a productive programming environment that hides the complexity of clusters and allows writing efficient programs is urgently needed. Despite multiple efforts to provide shared memory abstraction, message-passing (MPI) is still the state-of-the-art programming model for distributed-memory architectures. ^ Writing efficient MPI programs is challenging. In contrast, OpenMP is a shared-memory programming model that is known for its programming productivity. Researchers introduced automatic source-to-source translation schemes from OpenMP to MPI so that programmers can use OpenMP while targeting clusters. Those schemes limited their focus on …

Usability Of Immersive Virtual Reality Input Devices, Christopher G. Mankey

Open Access Dissertations

This research conducts a usability analysis of human interface devices within an Immersive Virtual Reality Environment. The analysis is carried out for two different interface devices, a commercially available Intersense ^© Wand and a home built pinch glove and wireless receiver. Users were asked to carry out a series of minor tasks involving placement of shaped blocks into corresponding holes within an Immersive Virtual Reality Environment. Performance was evaluated in terms of speed, accuracy and precision via the collection of completion times, errors made and the precision of motion during the experiment.

Design Of Robust Spin-Transfer Torque Magnetic Random Access Memories For Ultralow Power High Performance On-Chip Cache Applications, Xuanyao Fong

Open Access Dissertations

Spin-transfer torque magnetic random access memories (STT-MRAMs) based on magnetic tunnel junction (MTJ) has become the leading candidate for future universal memory technology due to its potential for low power, non-volatile, high speed and extremely good endurance. However, conflicting read and write requirements exist in STT-MRAM technology because the current path during read and write operations are the same. Read and write failures of STT-MRAMs are degraded further under process variations. The focus of this dissertation is to optimize the yield of STT- MRAMs under process variations by employing device-circuit-architecture co-design techniques. A devices-to-systems simulation framework was developed to evaluate …

A Method For Clustering High-Dimensional Data Using 1d Random Projections, Sangchun Han

Open Access Dissertations

Clustering high-dimensional data is more difficult than clustering low-dimensional data. The problem is twofold. First, there is an efficiency problem related to the data size, which increases with the dimensionality. Second, there is an effectiveness problem related to the fact that the mere existence of clusters in sample sets of high dimensions is questionable, as empirical samples hardly tend to cluster together in a meaningful fashion. The current approach to addressing this issue is to seek clusters in embedded subspaces of the original space. However, as dimensionality increases, a naive exhaustive search among all subspaces becomes exponentially more complex, which …