High-performance parallel systems for data-intensive computing

Thesis (Ph.D.)--University of Washington, 2016-08

Dynamic load balancing of distributed memory parallel computational mechanics using unstructured meshes for multi-physical modelling

As the complexity of parallel applications increase, the performance limitations resulting from computational load imbalance become dominant. Mapping the problem space to the processors in a parallel machine in a manner that balances the workload of each processors will typically reduce the run-time. In many cases the computation time required for a given calculation cannot be predetermined even at run-time and so static partition of the problem returns poor performance. For problems in which the computational load across the discretisation is dynamic and inhomogeneous, for example multi-physics problems involving fluid and solid mechanics with phase changes, the workload for a static subdomain will change over the course of a computation and cannot be estimated beforehand. For such applications the mapping of loads to process is required to change dynamically, at run-time in order to maintain reasonable efficiency. The issue of dynamic load balancing are examined in the context of PHYSICA, a three dimensional unstructured mesh multi-physics continuum mechanics computational modelling code.

Mesh partitioning and load balancing for distributed memory parallel systems

A method is outlined for optimising graph partitions which arise in mapping unstructured mesh calculations to parallel computers. The method employs a relative gain iterative technique to both evenly balance the workload and minimise the number and volume of interprocessor communications. A parallel graph reduction technique is also briefly described and can be used to give a global perspective to the optimisation. The algorithms work efficiently in parallel as well as sequentially and when combined with a fast direct partitioning technique (such as the Greedy algorithm) to give an initial partition, the resulting two-stage process proves itself to be both a powerful and flexible solution to the static graph-partitioning problem. Experiments indicate that the resulting parallel code can provide high quality partitions, independent of the initial partition, within a few seconds. The algorithms can also be used for dynamic load-balancing, reusing existing partitions and in this case the procedures are much faster than static techniques, provide partitions of similar or higher quality and, in comparison, involve the migration of a fraction of the data.

Improving CAS Capabilities: New Rules for Computing Improper Integrals

There are diferent applications in Engineering that require to compute improper integrals of the first kind (integrals defined on an unbounded domain) such as: the work required to move an object from the surface of the earth to in nity (Kynetic Energy), the electric potential created by a charged sphere, the probability density function or the cumulative distribution function in Probability Theory, the values of the Gamma Functions(wich useful to compute the Beta Function used to compute trigonometrical integrals), Laplace and Fourier Transforms (very useful, for example in Differential Equations).

Probability and Common-Sense: Tandem Towards Robust Robotic Object Recognition in Ambient Assisted Living

The suitable operation of mobile robots when providing Ambient Assisted Living (AAL) services calls for robust object recognition capabilities. Probabilistic Graphical Models (PGMs) have become the de-facto choice in recognition systems aiming to e ciently exploit contextual relations among objects, also dealing with the uncertainty inherent to the robot workspace. However, these models can perform in an inco herent way when operating in a long-term fashion out of the laboratory, e.g. while recognizing objects in peculiar con gurations or belonging to new types. In this work we propose a recognition system that resorts to PGMs and common-sense knowledge, represented in the form of an ontology, to detect those inconsistencies and learn from them. The utilization of the ontology carries additional advantages, e.g. the possibility to verbalize the robot's knowledge. A primary demonstration of the system capabilities has been carried out with very promising results.

Code pattern analysis of object-oriented programming languages

Code patterns, including programming patterns and design patterns, are good references for programming language feature improvement and software re-engineering. However, to our knowledge, no existing research has attempted to detect code patterns based on code clone detection technology. In this study, we build upon the previous work and propose to detect and analyze code patterns from a collection of open source projects using NiPAT technology. Because design patterns are most closely associated with object-oriented languages, we choose Java and Python projects to conduct our study. The tool we use for detecting patterns is NiPAT, a pattern detecting tool originally developed for the TXL programming language based on the NiCad clone detector. We extend NiPAT for the Java and Python programming languages. Then, we try to identify all the patterns from the pattern report and classify them into several different categories. In the end of the study, we analyze all the patterns and compare the differences between Java and Python patterns.

A multiagent based strategy for detecting attacks in databases in a distributed mode

This paper presents a distributed hierarchical multiagent architecture for detecting SQL injection attacks against databases. It uses a novel strategy, which is supported by a Case-Based Reasoning mechanism, which provides to the classifier agents with a great capacity of learning and adaptation to face this type of attack. The architecture combines strategies of intrusion detection systems such as misuse detection and anomaly detection. It has been tested and the results are presented in this paper.

Techniques for distributed theory synthesis in multiagent systems

Data sources are often dispersed geographically in real life applications. Finding a knowledge model may require to join all the data sources and to run a machine learning algorithm on the joint set. We present an alternative based on a Multi Agent System (MAS): an agent mines one data source in order to extract a local theory (knowledge model) and then merges it with the previous MAS theory using a knowledge fusion technique. This way, we obtain a global theory that summarizes the distributed knowledge without spending resources and time in joining data sources. New experiments have been executed including statistical significance analysis. The results show that, as a result of knowledge fusion, the accuracy of initial theories is significantly improved as well as the accuracy of the monolithic solution.

Improving surgery operations by means of cloud systems and distributed user interfaces

Surgical interventions are usually performed in an operation room; however, access to the information by the medical team members during the intervention is limited. While in conversations with the medical staff, we observed that they attach significant importance to the improvement of the information and communication direct access by queries during the process in real time. It is due to the fact that the procedure is rather slow and there is lack of interaction with the systems in the operation room. These systems can be integrated on the Cloud adding new functionalities to the existing systems the medical expedients are processed. Therefore, such a communication system needs to be built upon the information and interaction access specifically designed and developed to aid the medical specialists. Copyright 2014 ACM.

A new approach for computing a flood vulnerability index using cluster analysis

A Flood Vulnerability Index (FloodVI) was developed using Principal Component Analysis (PCA) and a new aggregation method based on Cluster Analysis (CA). PCA simplifies a large number of variables into a few uncorrelated factors representing the social, economic, physical and environmental dimensions of vulnerability. CA groups areas that have the same characteristics in terms of vulnerability into vulnerability classes. The grouping of the areas determines their classification contrary to other aggregation methods in which the areas' classification determines their grouping. While other aggregation methods distribute the areas into classes, in an artificial manner, by imposing a certain probability for an area to belong to a certain class, as determined by the assumption that the aggregation measure used is normally distributed, CA does not constrain the distribution of the areas by the classes. FloodVI was designed at the neighbourhood level and was applied to the Portuguese municipality of Vila Nova de Gaia where several flood events have taken place in the recent past. The FloodVI sensitivity was assessed using three different aggregation methods: the sum of component scores, the first component score and the weighted sum of component scores. The results highlight the sensitivity of the FloodVI to different aggregation methods. Both sum of component scores and weighted sum of component scores have shown similar results. The first component score aggregation method classifies almost all areas as having medium vulnerability and finally the results obtained using the CA show a distinct differentiation of the vulnerability where hot spots can be clearly identified. The information provided by records of previous flood events corroborate the results obtained with CA, because the inundated areas with greater damages are those that are identified as high and very high vulnerability areas by CA. This supports the fact that CA provides a reliable FloodVI.

Interoperability Challenges in Internet of Things Systems: a Service-Oriented Computing Approach

Internet of Things systems are pervasive systems evolved from cyber-physical to large-scale systems. Due to the number of technologies involved, software development involves several integration challenges. Among them, the ones preventing proper integration are those related to the system heterogeneity, and thus addressing interoperability issues. From a software engineering perspective, developers mostly experience the lack of interoperability in the two phases of software development: programming and deployment. On the one hand, modern software tends to be distributed in several components, each adopting its most-appropriate technology stack, pushing programmers to code in a protocol- and data-agnostic way. On the other hand, each software component should run in the most appropriate execution environment and, as a result, system architects strive to automate the deployment in distributed infrastructures. This dissertation aims to improve the development process by introducing proper tools to handle certain aspects of the system heterogeneity. Our effort focuses on three of these aspects and, for each one of those, we propose a tool addressing the underlying challenge. The first tool aims to handle heterogeneity at the transport and application protocol level, the second to manage different data formats, while the third to obtain optimal deployment. To realize the tools, we adopted a linguistic approach, i.e.\ we provided specific linguistic abstractions that help developers to increase the expressive power of the programming language they use, writing better solutions in more straightforward ways. To validate the approach, we implemented use cases to show that the tools can be used in practice and that they help to achieve the expected level of interoperability. In conclusion, to move a step towards the realization of an integrated Internet of Things ecosystem, we target programmers and architects and propose them to use the presented tools to ease the software development process.

A distributed middleware for IT/OT convergence in modern industrial environments

The modern industrial environment is populated by a myriad of intelligent devices that collaborate for the accomplishment of the numerous business processes in place at the production sites. The close collaboration between humans and work machines poses new interesting challenges that industry must overcome in order to implement the new digital policies demanded by the industrial transition. The Industry 5.0 movement is a companion revolution of the previous Industry 4.0, and it relies on three characteristics that any industrial sector should have and pursue: human centrality, resilience, and sustainability. The application of the fifth industrial revolution cannot be completed without moving from the implementation of Industry 4.0-enabled platforms. The common feature found in the development of this kind of platform is the need to integrate the Information and Operational layers. Our thesis work focuses on the implementation of a platform addressing all the digitization features foreseen by the fourth industrial revolution, making the IT/OT convergence inside production plants an improvement and not a risk. Furthermore, we added modular features to our platform enabling the Industry 5.0 vision. We favored the human centrality using the mobile crowdsensing techniques and the reliability and sustainability using pluggable cloud computing services, combined with data coming from the crowd support. We achieved important and encouraging results in all the domains in which we conducted our experiments. Our IT/OT convergence-enabled platform exhibits the right performance needed to satisfy the strict requirements of production sites. The multi-layer capability of the framework enables the exploitation of data not strictly coming from work machines, allowing a more strict interaction between the company, its employees, and customers.

6G enabled IoV scenarios with distributed intelligence mechanisms

The Internet of Vehicles (IoV) paradigm has emerged in recent times, where with the support of technologies like the Internet of Things and V2X , Vehicular Users (VUs) can access different services through internet connectivity. With the support of 6G technology, the IoV paradigm will evolve further and converge into a fully connected and intelligent vehicular system. However, this brings new challenges over dynamic and resource-constrained vehicular systems, and advanced solutions are demanded. This dissertation analyzes the future 6G enabled IoV systems demands, corresponding challenges, and provides various solutions to address them. The vehicular services and application requests demands proper data processing solutions with the support of distributed computing environments such as Vehicular Edge Computing (VEC). While analyzing the performance of VEC systems it is important to take into account the limited resources, coverage, and vehicular mobility into account. Recently, Non terrestrial Networks (NTN) have gained huge popularity for boosting the coverage and capacity of terrestrial wireless networks. Integrating such NTN facilities into the terrestrial VEC system can address the above mentioned challenges. Additionally, such integrated Terrestrial and Non-terrestrial networks (T-NTN) can also be considered to provide advanced intelligent solutions with the support of the edge intelligence paradigm. In this dissertation, we proposed an edge computing-enabled joint T-NTN-based vehicular system architecture to serve VUs. Next, we analyze the terrestrial VEC systems performance for VUs data processing problems and propose solutions to improve the performance in terms of latency and energy costs. Next, we extend the scenario toward the joint T-NTN system and address the problem of distributed data processing through ML-based solutions. We also proposed advanced distributed learning frameworks with the support of a joint T-NTN framework with edge computing facilities. In the end, proper conclusive remarks and several future directions are provided for the proposed solutions.

Machine learning "as a service" for high energy physics (MLaaS4HEP): evolution of a framework for ML-based physics analyses

The scientific success of the LHC experiments at CERN highly depends on the availability of computing resources which efficiently store, process, and analyse the amount of data collected every year. This is ensured by the Worldwide LHC Computing Grid infrastructure that connect computing centres distributed all over the world with high performance network. LHC has an ambitious experimental program for the coming years, which includes large investments and improvements both for the hardware of the detectors and for the software and computing systems, in order to deal with the huge increase in the event rate expected from the High Luminosity LHC (HL-LHC) phase and consequently with the huge amount of data that will be produced. Since few years the role of Artificial Intelligence has become relevant in the High Energy Physics (HEP) world. Machine Learning (ML) and Deep Learning algorithms have been successfully used in many areas of HEP, like online and offline reconstruction programs, detector simulation, object reconstruction, identification, Monte Carlo generation, and surely they will be crucial in the HL-LHC phase. This thesis aims at contributing to a CMS R&D project, regarding a ML "as a Service" solution for HEP needs (MLaaS4HEP). It consists in a data-service able to perform an entire ML pipeline (in terms of reading data, processing data, training ML models, serving predictions) in a completely model-agnostic fashion, directly using ROOT files of arbitrary size from local or distributed data sources. This framework has been updated adding new features in the data preprocessing phase, allowing more flexibility to the user. Since the MLaaS4HEP framework is experiment agnostic, the ATLAS Higgs Boson ML challenge has been chosen as physics use case, with the aim to test MLaaS4HEP and the contribution done with this work.

Saude coletiva e aptdão fisica de escolares de segundo grau : estudo a partis do Colegio Tecnico Industrial - UNESP, Bauru

Universidade Estadual de Campinas . Faculdade de Educação Física