Department: Engineering and Applied Science: Computer Science and Engineering ![[clear]](close-x.png "Remove this limiter")
18 matches in the database.
These are records: 1 - 18.
Did you mean instcode:ucii?
1.
Avakian, Annie.
Reducing Cache Access Time in Multicore Architectures Using Hardware and Software Techniques.
Degree: PhD, Engineering and Applied Science: Computer Science and Engineering, 2012, University of Cincinnati
► One of the challenges of multicore design is providing data quickly to…
(more)
▼ One of the challenges of multicore design is providing data quickly to all the processor cores running on a system. This has led to the development of architectures based on Network on Chips (NoC) due to their flexibility and scalability. In the NoC architecture, data is distributed among cache banks connected to different routers, therefore data access time varies by location. If data is accessed by two or more cores at the same time, then placing that data in the vicinity of those cores significantly improves overall access time. Recent proposals of hybrid and reconfigurable interconnect architectures try to take advantage of data locality to a certain extent by grouping processors that work on the same data set. In this dissertation, we propose migrating processor cores instead of data lines to take advantage of data locality. This is done either at static time where cores are assigned to routers before the runtime of the process with the introduction of the Reconfigurable Architecture for Multicore Systems (RAMS). In the static architecture, previous knowledge of the process characteristics is essential in determining a good configuration. We extend the idea further with the introduction of a dynamic reconfiguration entitled Dynamically Reconfigurable Multicore Architecture (DyaReMA) where cores are reassigned on the fly based on the pattern of data requests. A good test platform is essential to verify the validity of proposed architectures. We developed a complete modular hardware model that can be used to synthesize and simulate a range of architectures. The architectures proposed in literature have a homogeneous distribution of caches to routers. We propose a heterogeneous configuration of cache blocks to routers and show that the performance is significantly improved. We take this idea one step further and introduce the reconfigurable cache architecture that reassigns cache blocks to neighboring routers based on data access patterns. Hardware implementation alone cannot extract the full potential of multiocre architectures. Efficient data migration and cache coherency protocols are needed that further improve cache access time. Since data access is determined at run time, data migration has become an important part of multicore design. However, traditional data migration is expensive both due to the high cost in keeping track of all data accesses and the actual migration of data lines across multiple routers. In this dissertation, we propose a stepwise data migration scheme that improves efficiency reducing hardware cost. Data sharing introduces the need for cache coherency protocols. We propose a hybrid cache coherency protocol that combines the advantages of both broadcast and directory protocol to implement an efficient and scalable coherency method. Two coherency protocols are proposed, the first takes advantage of the characteristics of the hybrid/bus architecture. The second is targeted for NoC architectures. The proposed architectures along with data migration and cache coherency can improve data access time for L2 cache misses and help to significantly reduce the runtime of different processes in the realm of multicore architecture.
Advisors/Committee Members: Vemuri, Ranganadha.
Subjects: Computer Engineering
Keywords: Multicore; Hybrid NoC; Reconfigurable Architecture; Computer Architecture
More Like This
2.
Fang, Chunsheng.
Novel Frameworks for Mining Heterogeneous and Dynamic Networks.
Degree: PhD, Engineering and Applied Science: Computer Science and Engineering, 2011, University of Cincinnati
► Graphs serve as an important tool for discrete data representation. Recently, graph…
(more)
▼ Graphs serve as an important tool for discrete data representation. Recently, graph representations have made possible very powerful machine learning algorithms, such as manifold learning, kernel methods, semi-supervised learning. With the advent of large-scale real world networks, such as biological networks (disease network, drug target network, etc.), social networks (DBLP Co-authorship network, Facebook friendship, etc.), machine learning and data mining algorithms have found new application areas and have contributed to advance our understanding of properties, and phenomena governing real world networks. When dealing with real world data represented as networks, two problems arise quite naturally: I) How to integrate and align the knowledge encoded in multiple and heterogeneous networks? For instance, how to find out the similar genes in co-disease and protein-protein interaction networks? II) How to model and predict the evolution of a dynamic network? A real world example is, given N years snapshots of an evolving social network, how to build a model that can capture the temporal evolution and make reliable prediction? In this dissertation, we present an innovative graph embedding framework, which identifies the key components of modeling the evolution in time of a dynamic graph. Different from the many state-of-the-art graph link prediction and modeling algorithms, it formulates the link prediction problem from a geometric perspective that can capture the dynamics of the intrinsic continuous graph manifold evolution. It is attractive due to its simplicity and the potential to relax the mining problem into a feasible domain which enables standard machine learning and regression models to utilize historical graph time series data. To address the first problem, we first propose a novel probability-based similarity measure which led to promising applications in content based image retrieval and image annotation, followed by a manifold alignment framework to align multiple heterogeneous networks, which demonstrate its power in mining biological networks. Finally, the dynamic graph mining framework generalizes most of the current graph embedding dynamic link prediction algorithms. Comprehensive experimental results on both synthesized and real-world datasets demonstrate that our proposed algorithmic framework for multiple heterogeneous networks and dynamic networks, can lead to better and more insightful understanding of real world networks. Scalability of our algorithms is also considered by employing MapReduce cloud computing architecture.
Advisors/Committee Members: Ralescu, Anca.
Subjects: Computer Science
Keywords: machine learning; social network; data mining; manifold learning; graph embedding; dynamic graph
More Like This
3.
Fu, Weihuang.
Analytical Model for Capacity and Delay Optimization in Wireless Mesh Networks.
Degree: PhD, Engineering and Applied Science: Computer Science and Engineering, 2010, University of Cincinnati
► Motivated by ubiquitous communication, both wireless network theory and technology have vigorously…
(more)
▼ Motivated by ubiquitous communication, both wireless network theory and technology have vigorously developed in the past decades that could support broadband wireless access (BWA), and the current trend continues to replace wired network backbone. Conventional network access is served by network infrastructure, which is deployed at fixed locations and acts as "bridge", i.e., gateway, between wired backbone and mobile clients (MCs) with equipped wired-interface and air-interface. Infrastructures have to be placed at the locations where cables available, including network and power cables, which poses strong constraints on deployment locations, and high cost in cable deployment and maintenance. Wireless mesh networks (WMNs) are comprised of multi-radio mesh routers (MRs), which interconnect each other using wireless links to form a mesh backbone. This also forms a multi-cell architecture to provide network service for MCs, where Internet gateways (IGWs) are special MRs having wired connection to the Internet. The deployment of MRs is flexible, cost-efficient, self-organizing, etc. Mobile MRs even form a mobile mesh backbone. Due to its advantages, WMN could be one of the promising case of the next generation Internet. However, developing such a network also needs to address many fundamental issues inherited from two-tier network architecture, wireless multi-hop transmission, multi-cell structure, etc. In this dissertation, we analytically model a two-tier WMN and derive the asymptotic bounds of network capacity and delay, which are essential and tightly related factors in developing a WMN to support delay-sensitive applications such as voice over IP (VoIP), video conference, etc. This dissertation performs the analysis on a WMN backbone formed by self-organizing ad hoc MRs and shows how the network capacity is dominated by the network delay constraints, and the numbers of MCs, MRs and IGWs. We find that the network delay scales to either the number of MRs or the number of IGWs, and dominating factors depends on the type of routing strategy. Some of our results are also applicable to ad hoc networks, which can be equivalent to special case of a self-organizing WMN. We optimize the backbone capacity by introducing two types of channel assignment schemes in managing spectrum resource and mitigating backbone interference. One is a centralized channel assignment scheme, which is suitable to WMNs deployed by Internet service providers (ISPs), and the other is a distributed channel assignment scheme, which applies to static or mobile self-organizing WMN. In addition, we propose a clustering based fractional frequency reuse for multi-cell coverage of WMNs, which offers resource allocation higher flexibility and better fairness with additional spatial dimension. Our work is to analyze and solve the fundamental problems in developing WMNs for ubiquitous and pervasive access. The results in our dissertation can serve as the guideline in research and design of practical WMNs. We conclude with dissertation with some discussion on future area of research.
Advisors/Committee Members: Agrawal, Dharma.
Subjects: Computer Science
Keywords: Channel Assignment; Interference Mitigation; Multi-channel and Multi-radio; Multi-hop; Resource Management; Wireless Mesh Networks
More Like This
4.
Ghosh, Krishnendu.
Formal Analysis of Automated Model Abstractions under Uncertainty: Applications in Systems Biology.
Degree: PhD, Engineering and Applied Science: Computer Science and Engineering, 2012, University of Cincinnati
► In this dissertation, three fundamental problems in modeling of large scale biological…
(more)
▼ In this dissertation, three fundamental problems in modeling of large scale biological systems are addressed. 1. Modeling of chemical reaction under imprecise rate of reactions: A framework is created to model chemical reactions with an interval based approach, incorporating imprecision as well as creating a finite space. Algorithms are presented to construct model abstraction efficiently. The results of the algorithms on a prototype elucidate the model. The formalism presents a novel way to represent continuous data of concentrations for the chemicals and quantitative analysis of temporal behavior of the system. 2. Multiscale formalism in discrete domains: Biological processes are multiscale. We formalize the definition of multiscale modeling in discrete domains. A polynomial algorithm is constructed to compute identifiability of multiscale systems. 3. Formal analysis of gene regulatory network: A formalism that incorporates noise in the data is presented to study gene regulation. Computational efficiency of the formalism is evaluated on a prototype constructed from biological experimental data.
Advisors/Committee Members: Schlipf, John.
Subjects: Computer Science
Keywords: Formal Analysis; Systems Biology; Modeling; Uncertainty; Algorithms; Multiscale
More Like This
5.
Henkener, Kevin.
Two-Hop f-Factors and a Fair and Trustworthy P2P Storage Model.
Degree: PhD, Engineering and Applied Science: Computer Science and Engineering, 2010, University of Cincinnati
► In this dissertation, we present a novel approach to the problem of…
(more)
▼ In this dissertation, we present a novel approach to the problem of distributed (peer-to-peer) backup. Our approach requires that data not be transferred more than two-hops from it source and that each peer store exactly the same amount of data as it distributes to be backed up. These two requirements address two import features of any distributed backup solution - trust and fairness. In a social network, the hop distance requirement means that in the worst case, a peer's data is backed up in the local storage of a friend of a friend (FoaF). Our assumption is that this offers a higher degree of trust than simply choosing a random peer. We achieve fairness through the requirement that peers store exactly the same amount of data that they distribute for backup. To facilitate this requirement, our approach uses symmetric exchanges of data. This not only supports fairness, but also enhances trust by introducing a vested interest between peers to preserve the data that they are storing. We call our approach the fair two-hop exchange scheme, or FTHES. We show that existing f-factor theory and algorithms can be used to compute an FTHES. Then we introduce and prove a fundamental existence theorem which states that an FTHES always exists under two fairly weak conditions. This theorem leads to a linear time sequential algorithm and an efficient distributed algorithm. We also prove a theorem stating that at most 2n-3 exchanges are needed to backup all of the data in our scheme and later conjecture that this may actually have a lower bound of n. Finally, we present an application of the FTHES in a content management system.
Advisors/Committee Members: Berman, Kenneth.
Subjects: Computer science
Keywords: f-factor; peer-to-peer; distributed backup; p2p storage
More Like This
6.
Horvitz, Richard P.
Symbol Grounding Using Neural Networks.
Degree: PhD, Engineering and Applied Science: Computer Science and Engineering, 2012, University of Cincinnati
► The classical approach to artificial intelligence (i.e. symbol manipulation) and the connectionist…
(more)
▼ The classical approach to artificial intelligence (i.e. symbol manipulation) and the connectionist approach (artificial neural networks) have been criticized for their inadequacies. The philosopher John Searle's Chinese room thought experiment argued that symbolic systems have no understanding of the meaning contained in their representations. The philosophers Jerry Fodor and Zenon Pylyshyn argued that artificial neural networks could not exhibit certain features of human cognition, such as systematicity and composition of representations. We take the view that both of these problems can be solved by a suitable integration of connectionist and symbolic systems. In this work we investigate methods of using artificial neural networks to produce descriptions in propositional and predicate logic. Artificial neural networks are stuctured such that, upon training, simple features of the network correspond directly to either propositional variables in one case, and objects and predicates in the other. In both cases, the methods were tested on character recognition tasks.
Advisors/Committee Members: Bhatnagar, Raj.
Subjects: Computer Science
Keywords: symbol grounding; artificial neural networks; neural networks
More Like This
7.
Hu, Zhen.
Multi-Domain Clustering on Real-Valued Datasets.
Degree: PhD, Engineering and Applied Science: Computer Science and Engineering, 2011, University of Cincinnati
► Clustering is an important research problem for knowledge discovery from databases. It…
(more)
▼ Clustering is an important research problem for knowledge discovery from databases. It focuses on finding hidden structures embedded in datasets. It is non-trivial to arrive at a clustering in a dataset such that each pair of data points within the same cluster is similar to each other, and each pair in different clusters is distinct from each other. This is due to the multiplicity of meanings of similarity between data points and also from criteria determining the number, shape, and boundaries of clusters. Despite a large body of published research, new clustering problems keep arising requiring novel solutions. Such a situation is evolving in the field of biomedical research which is generating a large number of interrelated and interdependent datasets, and also in many other domains of science and business. We have developed three novel methodologies for clustering to meet these newly emerging needs. The first problem we have solved relates to the grouping of data points with “similar” density in the data space into distinct clusters, using full dimensional clustering. Based on the pair-wise similarity matrix among data points, we define a new type of relationship among them - that of the point pairs being Mutual K-Nearest Neighbors (MKNN) of each other, and design clustering algorithms based on this new notion to capture the data density. Compared with traditional Euclidean distance based clustering algorithms for datasets having different densities, our MKNN-based clustering algorithm allows users to form density-based clusters with significantly lower sensitivity to parameters . We have analytically and empirically demonstrated, using both synthetic and real-world datasets, the increased capability, precision, efficiency, and robustness of our algorithm. The second clustering algorithm which we have developed incorporates prior domain knowledge, provided as pair-wise similarity matrix in one dataset, into the clustering performed for data in another dataset. The data objects in “prior knowledge” data source and the second data source are the same. By adopting a semi-supervised clustering procedure, our algorithm, called Semi-supervised Gaussian Infinite Mixture Model (SGIMM), balances information from two data sources and generates clusters enforcing precise pair-wise relationships. SGIMM accommodates many types of prior knowledge and from the empirical studies done with both the synthetic data and the real-world data; SGIMM generates high quality clusters regardless of the quality of prior knowledge. The third type of problem we have solved relates to the discovery of subspace clusters. Numerous real world applications focus on selecting subsets of data points and feature subspaces having desirable characteristics specified in terms of properties such as low variance, high distinction, low residue value, etc. We use lattice structured search spaces to identify low variance subspace clusters from one dataset (bicluster), two datasets (3-Cluster), and high discrepancy subspace clusters from a single dataset (polarized bicluster). The results on both synthetic datasets and genomic datasets have been shown for all these types of clustering tasks and they show performance better than what is shown by most of the existing algorithms.
Advisors/Committee Members: Bhatnagar, Raj.
Subjects: Computer Science
Keywords: Clustering; Subspace Clustering
More Like This
8.
Iyer, Laxmi R.
CANDID - A Neurodynamical Model of Idea Generation.
Degree: PhD, Engineering and Applied Science: Computer Science and Engineering, 2012, University of Cincinnati
► Idea generation is a central cognitive activity in humans, and studying the…
(more)
▼ Idea generation is a central cognitive activity in humans, and studying the mechanisms of idea generation is important both to understand the creative process better and to produce applications that mimic human creativity. The goal of this research is to explore the neural basis of idea generation in individuals through computational connectionist modeling, and to use the resulting framework to study broader aspects of higher level cognition. The product of this is a model called Context-Adaptive NeuroDynamical IDeation (CANDID). While there have been other models of ideation, CANDID attempts to incorporate known information about the actual structures and processes of the brain - at least at an abstract level. Following widely accepted theories of ideation, the model postulates that ideas are conceptual combinations, and that the combinations arise naturally from the dynamics of the neurocognitive system under the influence of contextual information. Concepts, which constitute the fundamental semantic elements of the model, are represented in the system in two ways: 1) amodally via the activity of neural units in a Concept Network (CN); and 2) in terms of their sensory, functional and abstract attributes, or features, which are encoded in a neural network termed the Feature Layer (FL). Concepts are grouped together into categories based on their functional and/or attribute similarity. The categories are represented as distributed patterns of neural activity in the Dynamic Selector Network (DSN), which confines the ideational dynamics to a context-appropriate cognitive space through a dynamic biasing mechanism – termed “neurocognitive spotlights” due to its usage. The system receives information about the context as input, which interacts with the intrinsic dynamics of the DSN-CN-FL idea generation system to generate an itinerant sequence of ideas. These are evaluated by a critic, which models both internal and external evaluative processes. Based on its evaluation, the critic generates a reward signal, which feeds back to the generation system to improve ideation by reinforcing connections and modulating the dynamics. The proposed mechanism for the generation of ideas involves three concurrent and interacting processes: 1) Selecting a context-specific subspace of the overall concept space within which ideas will be sought; 2) Searching productively through this subspace via itinerant neural dynamics; and 3) Modulating and reconfiguring the search process through learning based on evaluative feedback. The system is driven by a context input representing the context and/or goal of ideation, which activates appropriate categories in the DSN, biasing the associated concepts in the CN to create a context-specific search space. Itinerant dynamics in the biased CN generate a productive search path to produce ideas, which are evaluated by the critic. The research in this thesis makes two main contributions: 1) The first comprehensive, biologically plausible neural model of context-dependent ideation - and thinking in general; and 2) A neurodynamical model for constructing context-specific cognitive spaces through the spotlight mechanism. In addition to these, the work also addresses other important issues such as the neural representation of semantic knowledge, the emergence of ideas as metastable attractors, and the formation of category representation.
Advisors/Committee Members: Minai, Ali.
Subjects: Computer Science
Keywords: ideation; cognitive control; working memory; neural networks; semantic cognition; computational neuroscience
More Like This
9.
Jun, Jung Hyun.
Analysis of Optimal Strategies to Minimize Message Delay in Mobile Opportunistic Sensor Networks.
Degree: PhD, Engineering and Applied Science: Computer Science and Engineering, 2011, University of Cincinnati
► Wireless sensor networks (WSNs) are autonomous and self-healing networks of small battery…
(more)
▼ Wireless sensor networks (WSNs) are autonomous and self-healing networks of small battery powered sensors. Besides sensing their physical environments, these sensors are capable of communicate wirelessly, store, and compute data locally. The small size and its capability attract academics as well as industry for real-time monitoring of an area for potential events like wild fire, intruders, and hazardous gas. Since multi-hop communications from sources to sink node were unavoidable in WSNs, it is hard to achieve a longer life time. To reduce multi-hop communication, the idea of using mobile nodes as relay nodes which collect data and deliver it to sink, was introduced to WSNs. In addition to life time improvement, the mobile relay nodes can also keep wireless bandwidth capacity to a constant level while the node density is high. The mobile relay nodes moves independently and random from the perspective of WSNs. Mobile Opportunistic sensor Network (MOsN) specifically denotes overlays of the mobile opportunistic network on top of a static wireless sensor network where the time taken for relay nodes to deliver the data from static sensors to sink is completely opportunistic and unbounded. However, many applications related to security, emergency, and bio-hazard cannot tolerate this unbounded message delay. So we begin by analyzing the average time taken for relay nodes to deliver the message to sink in MOsN by modeling the delivery of message as a randomly moving particle with certain biasness towards the sink. The result shows this delay is a function of the message bias level and distance d from the sink to origin of message. The sink as placed in the center of tilted grid with the radius of D and the message bias level α which varies [0, 1], the delay can vary from Ο(d) to Ο(D log d) to Ο(D² log d). Based on this result, we propose an heuristic algorithm which forward the data to other relay nodes if it has the bias level larger than a threshold 1/(2x+1), where x is distance of relay node carrying message to the sink. Finally, we deduced that the lower average message delay can be achieved by a static sensor wait for a relay node with its bias level higher than some threshold for its message delivery. Before we delve into optimal message handover policy, the movement of relay nodes modeled as a directed random walk with biases in their mobility. This bias random mobility model well represents the network of multiple independent but not identically moving relay. The relationship between the message bias level and mobility bias level is derived from a help of the simulation results. The optimal message handover policy based on the observed mobility bias level is proposed for static sensors at different locations when the inter-arrival time of relay nodes to a given static sensor is close to a constant. We also propose an optimal relay node selection algorithm in the case of inter-arriving time of relay nodes are linearly increasing.
Advisors/Committee Members: Agrawal, Dharma.
Subjects: Computer Science
Keywords: opportunistic network; mobile relay node; average message delay; optimal handover policy; wireless sensor network; grid topology
More Like This
10.
Kim, Hyoung-Kook.
Defect-oriented fault analysis of a two-D-flip-flop synchronizer and test method for its application.
Degree: PhD, Engineering and Applied Science: Computer Science and Engineering, 2012, University of Cincinnati
► This thesis presents defect-oriented fault modeling and analysis of a two-D-flip-flop synchronizer…
(more)
▼ This thesis presents defect-oriented fault modeling and analysis of a two-D-flip-flop synchronizer and provides a test method for its application circuits. Bridging (open) defects are injected into any possible pair of internal nodes of the synchronizer. Then, HSPICE is used to perform the circuit analysis of each defect. The major purpose of this analysis is to acquire all possible faults that might occur in the synchronizer by each injected bridging (open) defect. Simulation results show that bridging and open defects can cause the synchronizer to generate stuck-at fault, functional timing fault, pulse output fault, one-time pulse fault, internal oscillation fault, and undefined output fault. Moreover, fault behaviors of the synchronizer depend on the location and resistance value of each defect, the input signal pattern (rising and falling), the input signal application time, and the applied clock frequency. The issues of fault behavior under the consideration of process variation, and the relationship between defects and the synchronizer failure mechanisms are also discussed. After dealing with failure analysis, an asynchronous First-In-First-Out (FIFO) interface (for multi-clock domain circuits) as an application of the two-D-flip-flop synchronizer is implemented. The number of synchronizers in the asynchronous FIFO interface depends on the width of the address lines. A general test method for the asynchronous FIFO interface is proposed. The proposed general test method evolves to the several test methods to detect the observed faults of all synchronizers in the asynchronous FIFO interface. Programmable delay generation and calibration are used to accomplish the pseudo at-speed delay testing for the FIFO circuit. Results demonstrate that the fault modeling and test methods developed in this research are effective, and can greatly enhance the reliability of a circuit which contains multiple clock domains.
Advisors/Committee Members: Jone, Wen Ben.
Subjects: Computer Engineering
Keywords: fault modeling; fault analysis; synchronizer; asynchronous FIFO
More Like This
11.
Liu, Jianxun.
Pseudo-Exhaustive Built-in Self-Testing for Signal Integrity of High-Speed SoC Interconnects.
Degree: PhD, Engineering and Applied Science: Computer Science and Engineering, 2011, University of Cincinnati
► As technology approaches deep sub-micron and clock frequency approaches Giga Hertz, the…
(more)
▼ As technology approaches deep sub-micron and clock frequency approaches Giga Hertz, the signal integrity problem of high-speed interconnects is becoming a more and more serious issue. In this work, we propose a pseudo-exhaustive testing scheme for signal integrity faults of high-speed SoC interconnects. We first validate the applicability of traditional pseudo-exhaustive testing methods to high-speed interconnect testing by validating the crosstalk locality. Base on the concept of crosstalk locality, a PE-BIST testing scheme for simple interconnect bus structures is proposed. The scheme uses a serial scan chain interface, and thus can be easily integrated with existing boundary scan architectures. Special boundary scan cells and instructions to support such integration are also discussed. The proposed PE-BIST method is then extended to arbitrary interconnect structures. With the aid of a Net Interference Graph (NIG), we can easily identify the PE-BIST test cone size and assign individual nets into PE-BIST channels. The test architecture for arbitrary interconnects is also very simple, largely reusing existing BIST components built on the chip. The hardware overhead can therefore be minimized. In order to control the test cone size for PE-BIST, shield canbe inserted into the interconnect structure to control the test time. We also present a post global routing track placement method to reduce shielding overhead. Simulation results show that the interconnect signal integrity problem can be dealt with by PE-BIST with minimum shielding overhead and reasonable test time. Finally, PE-BIST uses a parallel testing scheme and excites many aggressor nets to do the transitions which may lead to excessive power dissipation during testing. Power limit is usually considered in current SoC design, and thus the power dissipation for PE-BIST cannot be negligible. We use an efficient high level power modeling scheme to partition a PE-BIST solution into small child PE-BIST solutions so that each child PE-BIST solution can be tested within a given test power limit.
Advisors/Committee Members: Jone, Wen Ben.
Subjects: Computer Engineering
Keywords: Interconnect Testing; Pseudo-Exhaustive Testing; PE-BIST; Signal Integrity; SoC; High Speed Interconnect
More Like This
12.
Mayfield, James L. IV.
A Parameterized Framework for Quantum Computation.
Degree: PhD, Engineering and Applied Science: Computer Science and Engineering, 2012, University of Cincinnati
► The primary focus of this dissertation is the development of a framework…
(more)
▼ The primary focus of this dissertation is the development of a framework for the design and analysis of quantum circuits and quantum algorithms based on a parameterization of primitive quantum operators. Parameterized operators are characterized by the linear combination of two logically opposing operators. Single qubit operators can be viewed as a partial identification and negation. Similarly, conditional operators act partially in terms of a given condition and partially in terms of the logical negation of that condition. In support of this logic, several sets of basis operators are defined for n qubit computation and a hierarchy of these operators is established. A set of elementary n qubit operators is then placed within the general hierarchy of n qubit operators. Within the framework, operators are further characterized by the encoding of binary information about their input to the phase of their output. This encoding is expressed as a boolean function on the n bit strings that correspond to the basis states of an n qubit state. Furthermore, interference between operators is characterized in terms of these encoding functions with specific relationships between the encoding functions producing a special case of interference dubbed decoding. When decoding occurs between two operators then the decoding operator effectively interprets the information encoded in the phase of a quantum state and produces a new state relative to the decoded information. The encoding and decoding properties of single qubit operators, elementary n qubit operators, and basic n qubit composite operators is developed. Finally, Deutsch's algorithm is analyzed and generalized by way of the parameterized operator and phase encoding/decoding framework.
Advisors/Committee Members: Ralescu, Anca.
Subjects: Computer Science
Keywords: Quantum Computing; Theoretical Computer Science
More Like This
13.
Mefford, Greg.
Side Channel Analysis Research Framework (SCARF).
Degree: MS, Engineering and Applied Science: Computer Science and Engineering, 2012, University of Cincinnati
► Getting started in the research of side-channel information leakage from cryptographic systems…
(more)
▼ Getting started in the research of side-channel information leakage from cryptographic systems is difficult and time-consuming. In addition to the time spent learning about the various mechanisms through which information can be leaked and the metrics used to quantify and analyze the leakage, the researcher must develop tools to collect the necessary data and process it in a standard way in order to compare results across multiple experiments. The Side-Channel Analysis Research Framework (SCARF) is presented as a solution to the complexities of designing a software infrastructure capable of managing both simulation and measurement data to produce relevant metrics and plots relating to cryptographic side-channel analysis research. SCARF consists of a software framework built on top of the Ruby programming language and using the open-source Redis database as an efficient in-memory object store. It is distributed with fully-working example code to demonstrate its use in side-channel analysis attacks against a simplified version of the Data Encryption Standard (DES) and against a physical implementation of Microchip's KeeLoq technology, often used in remote keyless-entry applications for vehicles and garage door openers. Through the implementation of multiple example attacks, we demonstrate SCARF as a useful tool for researching and comparing side-channel attacks on both physical and proposed cryptographic hardware.
Advisors/Committee Members: Vemuri, Ranga.
Subjects: Computer Engineering
Keywords: cryptography; side-channel; power-analysis; security; software; research
More Like This
14.
Nsang, Augustine S.
An Empirical Study of Novel Approaches to Dimensionality Reduction and Applications.
Degree: PhD, Engineering and Applied Science: Computer Science and Engineering, 2011, University of Cincinnati
► Dimensionality reduction is becoming increasingly important in the field of machine learning.…
(more)
▼ Dimensionality reduction is becoming increasingly important in the field of machine learning. In this thesis, we examine several traditional methods of dimensionality reduction, which include random projections, principal component analysis, singular value decomposition, kernel principal component analysis and discrete cosine transform. We also examine several existing applications of random projections (or dimensionality reduction, in general). In their paper, Random projections in dimensionality reduction: Applications to image and text data (2001), Bingham and Manilla suggest the use of random projections for query matching in a situation where a set of documents, instead of one particular one, were searched for. This suggests another application of random projections, namely to reduce the complexity of the query process. In this thesis, we explain why this approach fails, and suggest three alternative approaches to reducing the complexity of the query process using dimensionality reduction. We also outline query-based dimensionality reduction methods that can be used for image and web data. In each of the traditional approaches to dimensionality reduction (named above), each attribute in the reduced set is actually a linear combination of the attributes in the original data set. In this thesis, we take the position that true dimensionality reduction is obtained when the set of attributes in the reduced set is a proper subset of the attributes in the original data set, and we discuss seven novel approaches which satisfy this requirement. Using these seven approaches, as well as the RP and PCA approaches, we discuss several ways in which dimensionality reduction can be used for high dimensional clustering and classification.
Advisors/Committee Members: Ralescu, Anca.
Subjects: Computer Science
Keywords: dimensionality reduction; random projections; clustering; classification; queries; web data
More Like This
15.
Uti, Ngozi V.
Real-Time Mobile Video Compression and Streaming: Live Video from Mobile Devices over Cell Phone Networks.
Degree: PhD, Engineering and Applied Science: Computer Science and Engineering, 2011, University of Cincinnati
► The limited computing resources on mobile phones, the demands of real-time requirements,…
(more)
▼ The limited computing resources on mobile phones, the demands of real-time requirements, and the variable and error-prone nature of the bandwidth of cell phone networks make the task of streaming live video from cell phones very challenging. As such, computational simplicity and efficiency are a requirement for video encoders on mobile devices. This research presents core components of a mobile video compression algorithm that has been developed in this project to compress real-time video from cell phones. This work shows how the careful selection of video compression components can be used to strike a delicate balance between the computationally complex nature of video compression and the efficient utilization of the limited computing resources available on cell phones. Although optimality is never claimed, a method for compressing and streaming real-time video of 15 frames per second has been developed. The video encoder uses 5-3 wavelet transformation and a new subband aligned integer run-length encoding technique to compress video in real-time on mobile devices. The wavelet video encoder is adaptive, highly scalable, and can gracefully adjust video compression levels to match changing cell phone network bandwidth conditions. Further, because of the variability of the bandwidth of cell phone networks, the efficient streaming of real-time video over cell phone networks requires the ability to adapt the quality and amount of video being streamed to the available bandwidth. This research shows that without such adaptability, video frames will be dropped. Experiments presented herein show that without an adaptive framework over 50% of the video frames can be dropped. In response to this challenge, this research implements an application layer framework for the control of real-time streaming video originating from mobile devices to better utilize available bandwidth. The approach taken here aims to align the quality and transmission rate of live streaming video with the capabilities of cell phone networks. Using decision making and feedback from the receiving video decoder, this real-time mobile streaming video framework is able to sense network conditions and effectively predict the available bandwidth. This adaptive framework utilizes the scalable wavelet video encoder for video compression. In conjunction with the wavelet video encoder on the mobile device, the framework adapts in real-time the video quality and video frames transmitted per second to achieve a near 100% delivery rate. This work provides a thorough description of this framework along with numerous experimental results. Presented is a detailed examination of the features of the adaptive framework and how they relate to cell phone network conditions, the video being streamed, and the mobile computing resources available on the mobile device.
Advisors/Committee Members: Cheng, Yizong.
Subjects: Computer Science
Keywords: Real-Time Mobile Video; Mobile Video Compression; Video Streaming Challenges; Mobile Computational Resources; Real-Time Video Compression; Cell Phone Bandwidth
More Like This
16.
Wang, Junfang.
Efficient Positioning Technique for Multi-Interface Multi-Rate Wireless Mesh Networks.
Degree: PhD, Engineering and Applied Science: Computer Science and Engineering, 2010, University of Cincinnati
► Wireless mesh network (WMN) is a strong candidate for the next generation…
(more)
▼ Wireless mesh network (WMN) is a strong candidate for the next generation wireless network. A WMN is made up with three types of entities: Internet Gateways (IGWs), mesh routers (MRs), and mesh clients (MCs). IGWs provide interfaces to both the Internet and MRs. MRs together with IGWs form a mesh backbone by interconnecting each other via multi-hop wireless links. MCs access the Internet by setting up connections with MRs. The placement of three entities is one of the fundamental issues that could greatly affect the performance of a WMN. Positioning technique in WMNs aims at optimizing the positions of these nodes to improve network performance and could be further categorized into: IGW placement, MR placement and MC association. The first part of this dissertation introduces our work over MR placement. MR placement is a strategy that determines the minimal number and positions of MRs that satisfies various constraints such as network coverage, connectivity, Internet traffic demand, etc., for a given network area to be covered by a WMN. Some MR placement strategies may also indicate the appropriate number of interfaces each MR needs as well. A systematic MR placement is the first important step for establishing a WMN with desired network performance efficiently. Our study starts from modeling and formulating the MR placement problem. Then, we analyze the problem in an ideal homogeneous network model, which is characterized by single IGW, identical transmission rate, and MRs could be positioned anywhere in the network region. Hence, we extend the discussion into a more realistic constraint network model: MRs can only be placed in the pre-decided candidate positions; traffic demands is non-uniformly distributed. Furthermore, we deepen our study by taking into accounts the nature of multiple transmission ranges/rates of commercial MRs. We propose a heuristic placement algorithm called ILSearch, which considers both multiple transmission rates and co-channel interference in the constraint network with multiple IGWs. In addition, we develop a virtual force based algorithm: VFPlace, to place MRs in a special constraint network where only candidate areas, rather than specific positions, are known in advance. The numeric results shows the correctness and effectives of our analytical models and proposed algorithms. The second part of this dissertation presents our work about MC association. MC association strategy targets at helping MCs choose best MRs to establish the connections in the limited space and time. MC association can influence the performance of a WMN to a large extent since it determines how well the mesh backbone, i.e., the associated MRs, can serve MCs. MR's capability for serving MCs varies dynamically due to the changing channel condition, traffic load, and other network parameters. Mobile MCs should attach to MRs that are able to provide best service so as to maximize their own performance. But the determination procedure is constrained by time and space, which makes MCs only able to check partial MRs of the WMN. We map MC association problem as the modified secretary problem. Then, we propose three probabilistic strategies to enhance the possibility that MCs choose the best MRs: opportunistic association, conservative association, and hybrid association. Both statistical analysis and simulations show the conservative strategy outperforms the other two.
Advisors/Committee Members: Agrawal, Dharma.
Subjects: Computer Science
Keywords: Mesh Router Placement; Mesh Client Association; Multiply Transmission Rate; Multiply Transmission Range; Virtual Force; Wireless Mesh Network
More Like This
17.
Xu, Hao.
Runtime Leakage Control in Deep Sub-micron CMOS Technologies.
Degree: PhD, Engineering and Applied Science: Computer Science and Engineering, 2010, University of Cincinnati
► MOSFET scaling into deep sub-micro realm has resulted in significant increase in…
(more)
▼ MOSFET scaling into deep sub-micro realm has resulted in significant increase in leakage power consumption. In 45nm technology generation and beyond, leakage power consumption will catch up with, and may even dominate, dynamic power consumption. This makes leakage power reduction an indispensable component for low power designs in deep sub-micro technologies. Many leakage control techniques have been introduced and studied so far. They can be characterized into two classes: runtime techniques and design-time techniques. Design-time techniques only modify the circuit and thus have limited capability of leakage reduction. On the other hand, runtime techniques tune the circuit into low-leakage mode according to the variation of circuit workload. When a circuit or a system has substantial slackness in its workload, runtime techniques can yield significant leakage saving. Hence runtime techniques, such as power gating and reverse body biasing, are widely used in industrial practices, and extensively studied in current researches as well. Since the invention of runtime leakage control techniques (RTLC), most of they have been applied in a very crude manner. Several key questions regarding the design methodologies of RTLC remain unanswered, including how to design the optimum control policy, what is the optimum granularity of applying RTLC, and how to reduce leakage in circuit active mode. Before these questions are answered, RTLC can only be of an ad-hoc style. On top of these major questions, several other common problems in deep sub-micro technologies need to be considered before RTLC converges to a practical technique. These common problems include temperature and process variation, robustness issues in deep sub-micro technologies etc. They make the design of RTLC even more complicated and challenging. In response to all these questions and challenges, our research aims at answering the major open questions of RTLC, tackling the practical problems in deep sub-micro technologies and finally forging a systematical solution for RTLC. To this end, this thesis studies the corresponding modeling, optimization, design methodology and design automation issues. Our whole study is driven by the following two main ideas. First, we consider that aggressive idleness exploitation is the key to achieve maximal leakage control. Second, we consider that applying RTLC in a finer manner is the key to enable aggressive idleness exploitation. Our study is based on two leakage control techniques: power gating and reserve body biasing. During our analysis, one type of RBB technique, Vth hopping, turns out to be more effective to control leakage in a finer manner. Therefore it becomes the center of our final solution.
Advisors/Committee Members: Vemuri, Ranganadha.
Subjects: Electrical engineering
Keywords: CMOS digital design; Low power design; Leakage power control
More Like This
18.
Zhang, Minlu.
Discovery and Analysis of Patterns in Molecular Networks: Link Prediction, Network Analysis, and Applications to Novel Drug Target Discovery.
Degree: PhD, Engineering and Applied Science: Computer Science and Engineering, 2012, University of Cincinnati
► One of the most challenging problems in the post-genomic era for computer…
(more)
▼ One of the most challenging problems in the post-genomic era for computer scientists and bioinformaticians is to identify meaningful patterns from a huge amount of data describing a variety of molecular systems. Networks provide a unifying representation for these various molecular systems, such as protein interaction maps, transcriptional regulations, metabolites and reactions, signaling transduction pathways, and functional associations. On one hand, computational determination of molecular networks is of interest due to the tremendous labor and cost associated with large-scale wet-lab experiments. On the other hand, novel methods and approaches are in need to extract useful and meaningful patterns from established large-scale molecular networks. In this thesis, we tackle the problems of computationally predicting links to construct large-scale protein interaction maps, transcriptional regulatory networks, and disease related heterogeneous networks. In particular, we adopted a supervised learning framework for link prediction in protein interaction maps of a human pathogen, and performed network analysis to extract and identify novel drug targets for disease treatment. We developed and demonstrated a semi-supervised learning approach for link prediction in a transcriptional regulatory network, and further analyzed the biological relevance of identified links. In the thesis, we also developed and performed computational approaches to extract biologically meaningful patterns in large-scale protein interaction maps and disease- and gene-related networks. Similar to other real-life systems, molecular networks are dynamic and context-dependent. We comparatively analyzed the static conglomerate networks and context-dependent networks and systematically revealed their differences in global topological characteristics, subnetwork structure components, and functional compartments. Finally, we applied network analysis to extract interesting patterns in networks of rare human diseases and disease causing genes and identified their unique properties.
Advisors/Committee Members: Bhatnagar, Raj.
Subjects: Computer Science
Keywords: network analysis; link prediction; transcriptional regulation; orphan disease; rare disease; protein-protein interaction
More Like This
