Context-aware computing requires software that can adapt to changes in context. When contextual circumstances trigger multiple adaptations, software must also understand the relationships between these adaptations and react according to the rules governing these relationships. Adaptable software needs a means to establish and interpret these rules in order to avoid any undesirable and potentially catastrophic conflicts. This dissertation designs and implements the Adaptation Rules Management API (ArmAPI). ArmAPI has been demonstrated to work with a Context-Oriented Programming variation for Java called ContextJ* to execute conflict-free adaptations in two software applications. ArmAPI allows programmers to define relationship types between adaptations, and transfers these definitions to Prolog facts and rules. The Prolog engine, encapsulated within ArmAPI, then works with imperative algorithms to determine the appropriate adaptations to execute based on the current set of facts, rules, and contextual circumstances. Context represents all of the conditions for all of the entities known to an observing device. In any environment, context represents a large amount of data that can influence a multitude of conflicting adaptations. This research provides an incremental step towards overcoming the problem of adaptation conflict by constructing an API that considers the relationship types of inclusion, exclusion, ordering, conditional dependency, and independence. The API has been validated via two prototypes that provide typical scenarios.

This dissertation is composed of research studies that contribute to three research areas including social context-aware computing, internal context-aware computing, and human behavioral data mining. In social context-aware computing, four studies are conducted. First, mobile phone user calling behavioral patterns are characterized in forms of randomness level where relationships among them are then identified. Next, a study is conducted to investigate the relationship between the calling behavior and organizational groups. Third, a method is presented to quantitatively define mobile social closeness and social groups, which are then used to identify social group sizes and scaling ratio. Last, based on the mobile social grouping framework, the significant role of social ties in communication patterns is revealed. In internal context-aware computing, two studies are conducted where the notions of internal context are intention and situation. For intentional context, the goal is to sense the intention of the user in placing calls. A model is thus presented for predicting future calls envisaged as a call predicted list (CPL), which makes use of call history to build a probabilistic model of calling behavior. As an incoming call predictor, CPL is a list of numbers/contacts that are the most likely to be the callers within the next hour(s), which is useful for scheduling and daily planning. As an outgoing call predictor, CPL is generated as a list of numbers/contacts that are the most likely to be dialed when the user attempts to make an outgoing call (e.g., by flipping open or unlocking the phone). This feature helps save time from having to search through a lengthy phone book. For situational context, a model is presented for sensing the user's situation (e.g., in a library, driving a car, etc.) based on embedded sensors. The sensed context is then used to switch the phone into a suitable alert mode accordingly (e.g., vibrate mode while in a library, handsfree mode while driving, etc.). Inferring (social and internal) context introduces a challenging research problem in human behavioral data mining. Context is determined by the current state of mind (internal), relationship (social), and surroundings (physical). Thus, the current state of context is important and can be derived from the recent behavior and pattern. In data mining research area, therefore, two frameworks are developed for detecting recent patterns, where one is a model-driven approach and the other is a data-driven approach.

Exploiting context-aware environments, where sensors scattered in a environment update presence servers to indicate the environmental changes can be used to enable new services. Such systems have become feasible both in terms of technical di±culties and their cost. A current focus in this area of research is how a context-aware system should be designed so that it reduces both the cost and complexity of the infrastructure, but still provides the desired services. One of the key components of many context-aware systems is location sensing, because a user's location is one of the most used elements of information in context-aware services. In this paper, we address cost e®ective location services by utilizing measurements of WLAN signal strength. We derive from these measurements an estimate of a device's location, and make this location information available via a SIP Presence User Agent, thus making location information readily available to services that might wish to use this information - while hiding details of how this information is acquired from these services. / Genom att utnyttja kontextmedvetna miljöer, där sensorer i en miljö uppdaterar närvarande servrar med information omändringar i omgivningen, så kan man öppna upp vägar för nya tjänster. Sådana system har blivit utförbara bade när det gäller tekniska svarigheter och deras kostnader. Inom forskning som rör sådana här system ägnas mycket uppmärksamhetåt hur en kontextmedveten miljö borde designas för att minimera både kostnaden och komplexiteten av infrastrukturen, men fortfarande tillhandahålla den önskade tjänsten. En av huvudkomponenterna i många kontextmedvetna system är platsuppfattning, eftersom en användares position är en av de mest använda elementen av information i kontextmedvetna tjänster. I den här uppsatsen ägnar vi oss åt kostnadseffektiva platstjänster genom att mäta signalstyrkan av ett WLAN. Genom dessa mätningar uppskattar vi en enhets position och gör denna information tillgänglig via en SIP Presence User Agent, och gör på så vis platsinformationen tillgänglig för tjänster som kan vilja ha den { utan att avslöja detaljer om hur informationen har skaffats.

Massive parallelism combined with complex memory hierarchies and heterogeneity in high-performance computing (HPC) systems form a barrier to efficient application and architecture design. The performance achievements of the past must continue over the next decade to address the needs of scientific simulations. However, building an exascale system by 2022 that uses less than 20 megawatts will require significant innovations in power and performance efficiency. A key limitation of past approaches is a lack of power-performance policies allowing users to quantitatively bound the effects of power management on the performance of their applications and systems. Existing controllers and predictors use policies fixed by a knowledgeable user to opportunistically save energy and minimize performance impact. While the qualitative effects are often good and the aggressiveness of a controller can be tuned to try to save more or less energy, the quantitative effects of tuning and setting opportunistic policies on performance and power are unknown. In other words, the controller will save energy and minimize performance loss in many cases but we have little understanding of the quantitative effects of controller tuning. This makes setting power-performance policies a manual trial and error process for domain experts and a black art for practitioners. To improve upon past approaches to high-performance power management, we need to quantitatively understand the effects of power and performance at scale. In this work, I have developed theories and techniques to quantitatively understand the relationship between power and performance for high performance systems at scale. For instance, our system-level, iso-energy-efficiency model analyzes, evaluates and predicts the performance and energy use of data intensive parallel applications on multi-core systems. This model allows users to study the effects of machine and application dependent characteristics on system energy efficiency. Furthermore, this model helps users isolate root causes of energy or performance inefficiencies and develop strategies for scaling systems to maintain or improve efficiency.  I have also developed methodologies which can be extended and applied to model modern heterogeneous architectures such as GPU-based clusters to improve their efficiency at scale. / Ph. D.

Multicore architectures impose great pressure on resource management. The exploration spaces available for resource management increase explosively, especially for large-scale high end computing systems. The availability of abundant parallelism causes scalability concerns at all levels. Multicore architectures also impose pressure on power management. Growth in the number of cores causes continuous growth in power. In this dissertation, we introduce methods and techniques to enable scalable and energy efficient execution of parallel applications on multicore architectures. We study strategies and methodologies that combine DCT and DVFS for the hybrid MPI/OpenMP programming model. Our algorithms yield substantial energy saving (8.74% on average and up to 13.8%) with either negligible performance loss or performance gain (up to 7.5%). To save additional energy for high-end computing systems, we propose a power-aware MPI task aggregation framework. The framework predicts the performance effect of task aggregation in both computation and communication phases and its impact in terms of execution time and energy of MPI programs. Our framework provides accurate predictions that lead to substantial energy saving through aggregation (64.87% on average and up to 70.03%) with tolerable performance loss (under 5%). As we aggregate multiple MPI tasks within the same node, we have the scalability concern of memory registration for high performance networking. We propose a new memory registration/deregistration strategy to reduce registered memory on multicore architectures with helper threads. We investigate design polices and performance implications of the helper thread approach. Our method efficiently reduces registered memory (23.62% on average and up to 49.39%) and avoids memory registration/deregistration costs for reused communication memory. Our system enables the execution of application input sets that could not run to the completion with the memory registration limitation. / Ph. D.

The scalability of parallel applications executing on multithreaded and multicore multiprocessors is often quite limited due to large degrees of contention over shared resources on these systems. In fact, negative scalability frequently occurs such that a non-negligable performance loss is observed through the use of more processors and cores. In this dissertation, we present a prediction model for identifying efficient operating points of concurrency in multithreaded scientific applications in terms of both performance as a primary objective and power secondarily. We also present a runtime system that uses live analysis of hardware event rates through the prediction model to optimize applications dynamically. We discuss a dynamic, phase-aware performance prediction model (DPAPP), which combines statistical learning techniques, including multivariate linear regression and artificial neural networks, with runtime analysis of data collected from hardware event counters to locate optimal operating points of concurrency. We find that the scalability model achieves accuracy approaching 95%, sufficiently accurate to identify improved concurrency levels and thread placements from within real parallel scientific applications. Using DPAPP, we develop a prediction-driven runtime optimization scheme, called ACTOR, which throttles concurrency so that power consumption can be reduced and performance can be set at the knee of the scalability curve of each parallel execution phase in an application. ACTOR successfully identifies and exploits program phases where limited scalability results in a performance loss through the use of more processing elements, providing simultaneous reductions in execution time by 5%-18% and power consumption by 0%-11% across a variety of parallel applications and architectures. Further, we extend DPAPP and ACTOR to include support for runtime adaptation of DVFS, allowing for the synergistic exploitation of concurrency throttling and DVFS from within a single, autonomically-acting library, providing improved energy-efficiency compared to either approach in isolation. / Ph. D.

Power has become a primary concern for HPC systems. Dynamic voltage and frequency scaling (DVFS) and dynamic concurrency throttling (DCT) are two software tools (or knobs) for reducing the dynamic power consumption of HPC systems. To date, few works have considered the synergistic integration of DVFS and DCT in performance-constrained systems, and, to the best of our knowledge, no prior research has developed application-aware simultaneous DVFS and DCT controllers in real systems and parallel programming frameworks. We present a multi-dimensional, online performance prediction framework, which we deploy to address the problem of simultaneous runtime optimization of DVFS, DCT, and thread placement on multi-core systems. We present results from an implementation of the prediction framework in a runtime system linked to the Intel OpenMP runtime environment and running on a real dual-processor quad-core system as well as a dual-processor dual-core system. We show that the prediction framework derives near-optimal settings of the three power-aware program adaptation knobs that we consider. Our overall runtime optimization framework achieves significant reductions in energy (12.27% mean) and ED² (29.6% mean), through simultaneous power savings (3.9% mean) and performance improvements (10.3% mean). Our prediction and adaptation framework outperforms earlier solutions that adapt only DVFS or DCT, as well as one that sequentially applies DCT then DVFS. Further, our results indicate that prediction-based schemes for runtime adaptation compare favorably and typically improve upon heuristic search-based approaches in both performance and energy savings. / Master of Science

Green computing, an emerging field of research that seeks to reduce excess power consumption in high performance computing (HPC), is gaining popularity among researchers. Research in this field often relies on simulation or only uses a small cluster, typically 8 or 16 nodes, because of the lack of hardware support. In contrast, System G at Virginia Tech is a 2592 processor supercomputer equipped with power aware components suitable for large scale green computing research. DIRECT is a deterministic global optimization algorithm, implemented in the mathematical software package VTDIRECT95. This thesis explores the potential energy savings for the parallel implementation of DIRECT, called pVTdirect, when used with a large scale computational biology application, parameter estimation for a budding yeast cell cycle model, on System G. Two power aware approaches for pVTdirect are developed and compared against the CPUSPEED power saving system tool. The results show that knowledge of the parallel workload of the underlying application is beneficial for power management. / Master of Science

By 2004, microprocessor design focused on multicore scaling"increasing the number of cores per die in each generation "as the primary strategy for improving performance. These multicore processors typically equip multiple memory subsystems to improve data throughput. In addition, these systems employ heterogeneous processors such as GPUs and heterogeneous memories like non-volatile memory to improve performance, capacity, and energy efficiency. With the increasing volume of hardware resources and system complexity caused by heterogeneity, future systems will require intelligent ways to manage hardware resources. Early research to improve performance and energy efficiency on heterogeneous, multi-core, multi-memory systems focused on tuning a single primitive or at best a few primitives in the systems. The key limitation of past efforts is their lack of a holistic approach to resource management that balances the tradeoff between performance and energy consumption. In addition, the shift from simple, homogeneous systems to these heterogeneous, multicore, multi-memory systems requires in-depth understanding of efficient resource management for scalable execution, including new models that capture the interchange between performance and energy, smarter resource management strategies, and novel low-level performance/energy tuning primitives and runtime systems. Tuning an application to control available resources efficiently has become a daunting challenge; managing resources in automation is still a dark art since the tradeoffs among programming, energy, and performance remain insufficiently understood. In this dissertation, I have developed theories, models, and resource management techniques to enable energy-efficient execution of parallel applications through thread and data management in these heterogeneous multi-core, multi-memory systems. I study the effect of dynamic concurrent throttling on the performance and energy of multi-core, non-uniform memory access (NUMA) systems. I use critical path analysis to quantify memory contention in the NUMA memory system and determine thread mappings. In addition, I implement a runtime system that combines concurrent throttling and a novel thread mapping algorithm to manage thread resources and improve energy efficient execution in multi-core, NUMA systems. In addition, I propose an analytical model based on the queuing method that captures important factors in multi-core, multi-memory systems to quantify the tradeoff between performance and energy. The model considers the effect of these factors in a holistic fashion that provides a general view of performance and energy consumption in contemporary systems. Finally, I focus on resource management of future heterogeneous memory systems, which may combine two heterogeneous memories to scale out memory capacity while maintaining reasonable power use. I present a new memory controller design that combines the best aspects of two baseline heterogeneous page management policies to migrate data between two heterogeneous memories so as to optimize performance and energy. / Ph. D.

A identificação de características que sejam importantes sobre um dado estudante durante o processo de ensino-aprendizagem tem sido alvo de muitos estudos nos últimos tempos. Docentes, coordenadores e pedagogos têm defendido que cada indivíduo possui características particulares e importantes dentro do processo de desenvolvimento do conhecimento. A necessidade de adaptar as estratégias de ensino de acordo com as preferências do aprendiz é uma realidade dentro das salas de aula, sejam estas presenciais ou virtuais. Porém, isto não significa que numa sala de aula deva haver criação de um método para cada aluno, mas sim que se aponte qual a melhor forma de interação para cada um deles construindo grupos de aprendizes com características comuns. Trabalhando desta forma se torna possível identificar perfis de aprendizes dentro de um conjunto de estudantes, possibilitando que se possa trabalhar com conteúdos e ambientes de aprendizagem mais adequados aos aprendizes.O objetivo deste trabalho é apresentar a arquitetura de um sistema que possibilita a criação de cenários de aprendizagem baseados no perfil do aluno. O docente deve indicar as características que devem ser observadas durante as interações do aprendiz em um dado ambiente de aprendizagem eletrônica. Estas características serão modeladas como informações de contexto, permitindo que os pontos definidos como observáveis sejam organizados e modelados de forma a facilitar a monitoração das interações. A classificação do aluno em um determinando tipo de perfil de aprendizagem, previamente definido pelo docente, é realizado considerando-se as informações sobre o contexto da interação e os tipos de perfis de aprendizagem. Para que seja possível construir os cenários de aprendizagem considerando o perfil do aluno este trabalho propõe criar um relacionamento entre os tipos de perfis de aprendizagem e as estratégias de ensino através das categorias descritoras dos objetos de aprendizagem. / The identification of characteristics of a given student that are important during the teachinglearning process has been the focus of numerous studies in recent years. Teachers, coordinators and pedagogues have defended the notion that each individual presents particular and important characteristics in the developing knowledge process. The need to adapt teaching strategies to the student\'s preferences is a reality in classrooms, being physical or virtual classrooms. However, this does not mean that a method should be created for each student in a classroom, but that the best form of interaction for each of them be identified, building groups of learners with common characteristics. Working in this way makes it possible to identify learner profiles within a group of students, allowing one to work with learning contents and environments more suited to the learners. The objective of this work is to present the architecture of a system that allows for the creation of learning scenarios based on the studen profile. The teacher should point out the features which may be observed during the student interaction in a given e-elearning environment. These features are designed as context information, allowing defined observation points to be organized and modeled for facilitating the monitoring of interactions. The student classification in a given learning profile type, defined previously by the teacher, is realized considering information about the context of interaction and the learning profile types. For the learning environment to build the learning scenarios according to the student profile, this work proposes to create a relationship between the learning profile types and the teaching strategies through the description of learning objects categories.