Interoperable resource brokering with policy-based provisioning and job allocation

The increasing needs for computational power in areas such as weather simulation, genomics or Internet applications have led to sharing of geographically distributed and heterogeneous resources from commercial data centers and scientific institutions. Research in the areas of utility, grid and cloud computing, together with improvements in network and hardware virtualization has resulted in methods to locate and use resources to rapidly provision virtual environments in a flexible manner, while lowering costs for consumers and providers. ^ However, there is still a lack of methodologies to enable efficient and seamless sharing of resources among institutions. In this work, we concentrate in the problem of executing parallel scientific applications across distributed resources belonging to separate organizations. Our approach can be divided in three main points. First, we define and implement an interoperable grid protocol to distribute job workloads among partners with different middleware and execution resources. Second, we research and implement different policies for virtual resource provisioning and job-to-resource allocation, taking advantage of their cooperation to improve execution cost and performance. Third, we explore the consequences of on-demand provisioning and allocation in the problem of site-selection for the execution of parallel workloads, and propose new strategies to reduce job slowdown and overall cost.^

Methods for modeling and analyzing concurrent software

Concurrent software executes multiple threads or processes to achieve high performance. However, concurrency results in a huge number of different system behaviors that are difficult to test and verify. The aim of this dissertation is to develop new methods and tools for modeling and analyzing concurrent software systems at design and code levels. This dissertation consists of several related results. First, a formal model of Mondex, an electronic purse system, is built using Petri nets from user requirements, which is formally verified using model checking. Second, Petri nets models are automatically mined from the event traces generated from scientific workflows. Third, partial order models are automatically extracted from some instrumented concurrent program execution, and potential atomicity violation bugs are automatically verified based on the partial order models using model checking. Our formal specification and verification of Mondex have contributed to the world wide effort in developing a verified software repository. Our method to mine Petri net models automatically from provenance offers a new approach to build scientific workflows. Our dynamic prediction tool, named McPatom, can predict several known bugs in real world systems including one that evades several other existing tools. McPatom is efficient and scalable as it takes advantage of the nature of atomicity violations and considers only a pair of threads and accesses to a single shared variable at one time. However, predictive tools need to consider the tradeoffs between precision and coverage. Based on McPatom, this dissertation presents two methods for improving the coverage and precision of atomicity violation predictions: 1) a post-prediction analysis method to increase coverage while ensuring precision; 2) a follow-up replaying method to further increase coverage. Both methods are implemented in a completely automatic tool.

Tool cutting force modeling and wear estimation of micro-end-milling operations

The applications of micro-end-milling operations have increased recently. A Micro-End-Milling Operation Guide and Research Tool (MOGART) package has been developed for the study and monitoring of micro-end-milling operations. It includes an analytical cutting force model, neural network based data mapping and forecasting processes, and genetic algorithms based optimization routines. MOGART uses neural networks to estimate tool machinability and forecast tool wear from the experimental cutting force data, and genetic algorithms with the analytical model to monitor tool wear, breakage, run-out, cutting conditions from the cutting force profiles. The performance of MOGART has been tested on the experimental data of over 800 experimental cases and very good agreement has been observed between the theoretical and experimental results. The MOGART package has been applied to the micro-end-milling operation study of Engineering Prototype Center of Radio Technology Division of Motorola Inc.

Docking and simulation study of the enantiospecificity of chiral epoxidation reactions catalyzed by chloroperoxidase

Chloroperoxidase (CPO), a 298-residue glycosylated protein from the fungus Caldariomyces fumago, is probably the most versatile heme enzyme yet discovered. Interest in CPO as a catalyst is based on its power to produce enantiomerically enriched products. Recent research has focused its attention on the ability of CPO to epoxidize alkenes in high regioselectivity and enantioselectivity as an efficient and environmentally benign alternative to traditional synthetic routes. There has been little work on the nature of ligand binding, which probably controls the regio- and enantiospecifity of CPO. Consequently it is here that we focus our work. We report docking calculations and computer simulations aimed at predicting the enantiospecificity of CPO-catalyzed epoxidation of three model substrates. On the basis of this work candidate mutations to improve the efficiency of CPO are predicted. In order to accomplish these aims, a simulated annealing and molecular dynamics protocol is developed to sample potentially reactive substrate/CPO complexes.

Evaluating the effectiveness of signal timing optimization based on microscopic simulation evaluation

The optimization of the timing parameters of traffic signals provides for efficient operation of traffic along a signalized transportation system. Optimization tools with macroscopic simulation models have been used to determine optimal timing plans. These plans have been, in some cases, evaluated and fine tuned using microscopic simulation tools. A number of studies show inconsistencies between optimization tool results based on macroscopic simulation and the results obtained from microscopic simulation. No attempts have been made to determine the reason behind these inconsistencies. This research investigates whether adjusting the parameters of macroscopic simulation models to correspond to the calibrated microscopic simulation model parameters can reduce said inconsistencies. The adjusted parameters include platoon dispersion model parameters, saturation flow rates, and cruise speeds. The results from this work show that adjusting cruise speeds and saturation flow rates can have significant impacts on improving the optimization/macroscopic simulation results as assessed by microscopic simulation models.

Interoperable Resource Brokering with Policy-based Provisioning and Job Allocation

The increasing needs for computational power in areas such as weather simulation, genomics or Internet applications have led to sharing of geographically distributed and heterogeneous resources from commercial data centers and scientific institutions. Research in the areas of utility, grid and cloud computing, together with improvements in network and hardware virtualization has resulted in methods to locate and use resources to rapidly provision virtual environments in a flexible manner, while lowering costs for consumers and providers. However, there is still a lack of methodologies to enable efficient and seamless sharing of resources among institutions. In this work, we concentrate in the problem of executing parallel scientific applications across distributed resources belonging to separate organizations. Our approach can be divided in three main points. First, we define and implement an interoperable grid protocol to distribute job workloads among partners with different middleware and execution resources. Second, we research and implement different policies for virtual resource provisioning and job-to-resource allocation, taking advantage of their cooperation to improve execution cost and performance. Third, we explore the consequences of on-demand provisioning and allocation in the problem of site-selection for the execution of parallel workloads, and propose new strategies to reduce job slowdown and overall cost.

Development of Physics-based Models and Design Optimization of Power Electronic Conversion Systems

The main objective for physics based modeling of the power converter components is to design the whole converter with respect to physical and operational constraints. Therefore, all the elements and components of the energy conversion system are modeled numerically and combined together to achieve the whole system behavioral model. Previously proposed high frequency (HF) models of power converters are based on circuit models that are only related to the parasitic inner parameters of the power devices and the connections between the components. This dissertation aims to obtain appropriate physics-based models for power conversion systems, which not only can represent the steady state behavior of the components, but also can predict their high frequency characteristics. The developed physics-based model would represent the physical device with a high level of accuracy in predicting its operating condition. The proposed physics-based model enables us to accurately develop components such as; effective EMI filters, switching algorithms and circuit topologies [7]. One of the applications of the developed modeling technique is design of new sets of topologies for high-frequency, high efficiency converters for variable speed drives. The main advantage of the modeling method, presented in this dissertation, is the practical design of an inverter for high power applications with the ability to overcome the blocking voltage limitations of available power semiconductor devices. Another advantage is selection of the best matching topology with inherent reduction of switching losses which can be utilized to improve the overall efficiency. The physics-based modeling approach, in this dissertation, makes it possible to design any power electronic conversion system to meet electromagnetic standards and design constraints. This includes physical characteristics such as; decreasing the size and weight of the package, optimized interactions with the neighboring components and higher power density. In addition, the electromagnetic behaviors and signatures can be evaluated including the study of conducted and radiated EMI interactions in addition to the design of attenuation measures and enclosures.

Methods for Modeling and Analyzing Concurrent Software

Concurrent software executes multiple threads or processes to achieve high performance. However, concurrency results in a huge number of different system behaviors that are difficult to test and verify. The aim of this dissertation is to develop new methods and tools for modeling and analyzing concurrent software systems at design and code levels. This dissertation consists of several related results. First, a formal model of Mondex, an electronic purse system, is built using Petri nets from user requirements, which is formally verified using model checking. Second, Petri nets models are automatically mined from the event traces generated from scientific workflows. Third, partial order models are automatically extracted from some instrumented concurrent program execution, and potential atomicity violation bugs are automatically verified based on the partial order models using model checking. Our formal specification and verification of Mondex have contributed to the world wide effort in developing a verified software repository. Our method to mine Petri net models automatically from provenance offers a new approach to build scientific workflows. Our dynamic prediction tool, named McPatom, can predict several known bugs in real world systems including one that evades several other existing tools. McPatom is efficient and scalable as it takes advantage of the nature of atomicity violations and considers only a pair of threads and accesses to a single shared variable at one time. However, predictive tools need to consider the tradeoffs between precision and coverage. Based on McPatom, this dissertation presents two methods for improving the coverage and precision of atomicity violation predictions: 1) a post-prediction analysis method to increase coverage while ensuring precision; 2) a follow-up replaying method to further increase coverage. Both methods are implemented in a completely automatic tool.