Algorithm-oriented design of efficient many-core architectures applied to dense matrix multiplication

Recent integrated circuit technologies have opened the possibility to design parallel architectures with hundreds of cores on a single chip. The design space of these parallel architectures is huge with many architectural options. Exploring the design space gets even more difficult if, beyond performance and area, we also consider extra metrics like performance and area efficiency, where the designer tries to design the architecture with the best performance per chip area and the best sustainable performance. In this paper we present an algorithm-oriented approach to design a many-core architecture. Instead of doing the design space exploration of the many core architecture based on the experimental execution results of a particular benchmark of algorithms, our approach is to make a formal analysis of the algorithms considering the main architectural aspects and to determine how each particular architectural aspect is related to the performance of the architecture when running an algorithm or set of algorithms. The architectural aspects considered include the number of cores, the local memory available in each core, the communication bandwidth between the many-core architecture and the external memory and the memory hierarchy. To exemplify the approach we did a theoretical analysis of a dense matrix multiplication algorithm and determined an equation that relates the number of execution cycles with the architectural parameters. Based on this equation a many-core architecture has been designed. The results obtained indicate that a 100 mm(2) integrated circuit design of the proposed architecture, using a 65 nm technology, is able to achieve 464 GFLOPs (double precision floating-point) for a memory bandwidth of 16 GB/s. This corresponds to a performance efficiency of 71 %. Considering a 45 nm technology, a 100 mm(2) chip attains 833 GFLOPs which corresponds to 84 % of peak performance These figures are better than those obtained by previous many-core architectures, except for the area efficiency which is limited by the lower memory bandwidth considered. The results achieved are also better than those of previous state-of-the-art many-cores architectures designed specifically to achieve high performance for matrix multiplication.

Intelligence by Design: Principles of Modularity and Coordination for Engineerin

All intelligence relies on search --- for example, the search for an intelligent agent's next action. Search is only likely to succeed in resource-bounded agents if they have already been biased towards finding the right answer. In artificial agents, the primary source of bias is engineering. This dissertation describes an approach, Behavior-Oriented Design (BOD) for engineering complex agents. A complex agent is one that must arbitrate between potentially conflicting goals or behaviors. Behavior-oriented design builds on work in behavior-based and hybrid architectures for agents, and the object oriented approach to software engineering. The primary contributions of this dissertation are: 1.The BOD architecture: a modular architecture with each module providing specialized representations to facilitate learning. This includes one pre-specified module and representation for action selection or behavior arbitration. The specialized representation underlying BOD action selection is Parallel-rooted, Ordered, Slip-stack Hierarchical (POSH) reactive plans. 2.The BOD development process: an iterative process that alternately scales the agent's capabilities then optimizes the agent for simplicity, exploiting tradeoffs between the component representations. This ongoing process for controlling complexity not only provides bias for the behaving agent, but also facilitates its maintenance and extendibility. The secondary contributions of this dissertation include two implementations of POSH action selection, a procedure for identifying useful idioms in agent architectures and using them to distribute knowledge across agent paradigms, several examples of applying BOD idioms to established architectures, an analysis and comparison of the attributes and design trends of a large number of agent architectures, a comparison of biological (particularly mammalian) intelligence to artificial agent architectures, a novel model of primate transitive inference, and many other examples of BOD agents and BOD development.

Estado situacional de los modelos basados en agentes y su impacto en la investigación organizacional

En un mundo hiperconectado, dinámico y cargado de incertidumbre como el actual, los métodos y modelos analíticos convencionales están mostrando sus limitaciones. Las organizaciones requieren, por tanto, herramientas útiles que empleen tecnología de información y modelos de simulación computacional como mecanismos para la toma de decisiones y la resolución de problemas. Una de las más recientes, potentes y prometedoras es el modelamiento y la simulación basados en agentes (MSBA). Muchas organizaciones, incluidas empresas consultoras, emplean esta técnica para comprender fenómenos, hacer evaluación de estrategias y resolver problemas de diversa índole. Pese a ello, no existe (hasta donde conocemos) un estado situacional acerca del MSBA y su aplicación a la investigación organizacional. Cabe anotar, además, que por su novedad no es un tema suficientemente difundido y trabajado en Latinoamérica. En consecuencia, este proyecto pretende elaborar un estado situacional sobre el MSBA y su impacto sobre la investigación organizacional.

On intentional and social agents with graded attitudes

La principal contribución de esta Tesis es la propuesta de un modelo de agente BDI graduado (g-BDI) que permita especificar una arquitetura de agente capaz de representar y razonar con actitudes mentales graduadas. Consideramos que una arquitectura BDI más exible permitirá desarrollar agentes que alcancen mejor performance en entornos inciertos y dinámicos, al servicio de otros agentes (humanos o no) que puedan tener un conjunto de motivaciones graduadas. En el modelo g-BDI, las actitudes graduadas del agente tienen una representación explícita y adecuada. Los grados en las creencias representan la medida en que el agente cree que una fórmula es verdadera, en los deseos positivos o negativos permiten al agente establecer respectivamente, diferentes niveles de preferencias o de rechazo. Las graduaciones en las intenciones también dan una medida de preferencia pero en este caso, modelan el costo/beneficio que le trae al agente alcanzar una meta. Luego, a partir de la representación e interacción de estas actitudes graduadas, pueden ser modelados agentes que muestren diferentes tipos de comportamiento. La formalización del modelo g-BDI está basada en los sistemas multi-contextos. Diferentes lógicas modales multivaluadas se han propuesto para representar y razonar sobre las creencias, deseos e intenciones, presentando en cada caso una axiomática completa y consistente. Para tratar con la semántica operacional del modelo de agente, primero se definió un calculus para la ejecución de sistemas multi-contextos, denominado Multi-context calculus. Luego, mediante este calculus se le ha dado al modelo g-BDI semántica computacional. Por otra parte, se ha presentado una metodología para la ingeniería de agentes g-BDI en un escenario multiagente. El objeto de esta propuesta es guiar el diseño de sistemas multiagentes, a partir de un problema del mundo real. Por medio del desarrollo de un sistema recomendador en turismo como caso de estudio, donde el agente recomendador tiene una arquitectura g-BDI, se ha mostrado que este modelo es valioso para diseñar e implementar agentes concretos. Finalmente, usando este caso de estudio se ha realizado una experimentación sobre la flexibilidad y performance del modelo de agente g-BDI, demostrando que es útil para desarrollar agentes que manifiesten conductas diversas. También se ha mostrado que los resultados obtenidos con estos agentes recomendadores modelizados con actitudes graduadas, son mejores que aquellos alcanzados por los agentes con actitudes no-graduadas.

An energy budget agent-based model of earthworm populations and its application to study the effects of pesticides

Earthworms are important organisms in soil communities and so are used as model organisms in environmental risk assessments of chemicals. However current risk assessments of soil invertebrates are based on short-term laboratory studies, of limited ecological relevance, supplemented if necessary by site-specific field trials, which sometimes are challenging to apply across the whole agricultural landscape. Here, we investigate whether population responses to environmental stressors and pesticide exposure can be accurately predicted by combining energy budget and agent-based models (ABMs), based on knowledge of how individuals respond to their local circumstances. A simple energy budget model was implemented within each earthworm Eisenia fetida in the ABM, based on a priori parameter estimates. From broadly accepted physiological principles, simple algorithms specify how energy acquisition and expenditure drive life cycle processes. Each individual allocates energy between maintenance, growth and/or reproduction under varying conditions of food density, soil temperature and soil moisture. When simulating published experiments, good model fits were obtained to experimental data on individual growth, reproduction and starvation. Using the energy budget model as a platform we developed methods to identify which of the physiological parameters in the energy budget model (rates of ingestion, maintenance, growth or reproduction) are primarily affected by pesticide applications, producing four hypotheses about how toxicity acts. We tested these hypotheses by comparing model outputs with published toxicity data on the effects of copper oxychloride and chlorpyrifos on E. fetida. Both growth and reproduction were directly affected in experiments in which sufficient food was provided, whilst maintenance was targeted under food limitation. Although we only incorporate toxic effects at the individual level we show how ABMs can readily extrapolate to larger scales by providing good model fits to field population data. The ability of the presented model to fit the available field and laboratory data for E. fetida demonstrates the promise of the agent-based approach in ecology, by showing how biological knowledge can be used to make ecological inferences. Further work is required to extend the approach to populations of more ecologically relevant species studied at the field scale. Such a model could help extrapolate from laboratory to field conditions and from one set of field conditions to another or from species to species.

An aspect-oriented reference architecture for Software Engineering Environments

Reusable and evolvable Software Engineering Environments (SEES) are essential to software production and have increasingly become a need. In another perspective, software architectures and reference architectures have played a significant role in determining the success of software systems. In this paper we present a reference architecture for SEEs, named RefASSET, which is based on concepts coming from the aspect-oriented approach. This architecture is specialized to the software testing domain and the development of tools for that domain is discussed. This and other case studies have pointed out that the use of aspects in RefASSET provides a better Separation of Concerns, resulting in reusable and evolvable SEEs. (C) 2011 Elsevier Inc. All rights reserved.

Agent-Based Hybrid Intelligent Systems: An Agent-Based Framework for Complex Problem Solving

Pt. I. Fundamentals of hybrid intelligent systems and agents -- 1. Introduction -- 2. Basics of hybrid intelligent systems -- 3. Basics of agents and multi-agent systems -- Pt. II. Methodology and framework -- 4. Agent-oriented methodologies -- 5. Agent-based framework for hybrid intelligent systems --6. Matchmaking in middle agents -- Pt. III. Application systems -- 7. Agent-based hybrid intelligent system for financial investment
planning -- 8. Agent-based hybrid intelligent system for data mining -- Pt. IV. Concluding remarks -- 9. The less the more -- App. Sample source codes of the agent-based financial planning system

Towards a Methodology for Coordination Mechanism Selection in Open Systems

Agent-oriented software engineering (AOSE) is a promising approach to developing applications for dynamic open systems. If well developed, these applications can be opportunistic, taking advantage of services implemented by other developers at appropriate times. However, methodologies are needed to aid the development of systems that are both flexible enough to be opportunistic and tightly defined by the application requirements. In this paper, we investigate how developers can choose the coordination mechanisms of agents so that the agents will best fulfil application requirements in an open system.

Agents and the Grid: Service Discovery

The Grid is a large-scale computer system that is capable of coordinating resources that are not subject to centralised control, whilst using standard, open, general-purpose protocols and interfaces, and delivering non-trivial qualities of service. In this chapter, we argue that Grid applications very strongly suggest the use of agent-based computing, and we review key uses of agent technologies in Grids: user agents, able to customize and personalise data; agent communication languages offering a generic and portable communication medium; and negotiation allowing multiple distributed entities to reach service level agreements. In the second part of the chapter, we focus on Grid service discovery, which we have identified as a prime candidate for use of agent technologies: we show that Grid-services need to be located via personalised, semantic-rich discovery processes, which must rely on the storage of arbitrary metadata about services that originates from both service providers and service users. We present UDDI-MT, an extension to the standard UDDI service directory approach that supports the storage of such metadata via a tunnelling technique that ties the metadata store to the original UDDI directory. The outcome is a flexible service registry which is compatible with existing standards and also provides metadata-enhanced service discovery.

From SMART to agent systems development

n order for agent-oriented software engineering to prove effective it must use principled notions of agents and enabling specification and reasoning, while still considering routes to practical implementation. This paper deals with the issue of individual agent specification and construction, departing from the conceptual basis provided by the smart agent framework. smart offers a descriptive specification of an agent architecture but omits consideration of issues relating to construction and control. In response, we introduce two new views to complement smart: a behavioural specification and a structural specification which, together, determine the components that make up an agent, and how they operate. In this way, we move from abstract agent system specification to practical implementation. These three aspects are combined to create an agent construction model, actsmart, which is then used to define the AgentSpeak(L) architecture in order to illustrate the application of actsmart.

Dynamically Adapting BDI Agent Architectures based on High-level User Specifications

Users are facing an increasing challenge of managing information and being available anytime anywhere, as the web exponentially grows. As a consequence, assisting them in their routine tasks has become a relevant issue to be addressed. In this paper, we introduce a software framework that supports the development of Personal Assistance Software (PAS). It relies on the idea of exposing a high level user model in order to increase user trust in the task delegation process as well as empowering them to manage it. The framework provides a synchronization mechanism that is responsible for dynamically adapting an underlying BDI agent-based running implementation in order to keep this high-level view of user customizations consistent with it.

Incorporating design explanation within formal object-oriented method (FOOM)

Requirements engineering is a commencing phase in the development of either software applications or information systems. It is concerned with understanding and specifying the customer's requirements of the system to be delivered. Throughout the literature, this is agreed to be one of the most crucial and, unfortunately, problematic phases in development. Despite the diversity of research directions, approaches and methods, the question of process understanding and management is still limited. Among contemporary approaches to the improvement of the current practice of Requirements Engineering, Formal Object-Oriented Method (FOOM) has been introduced as a new promising solution. The FOOM approach to requirements engineering is based on a synthesis of socio-organisational theory, the object-oriented approach, and mathematical formal specification. The entire FOOM specification process is evolutionary and involves a large volume of changes in requirements. During this process, requirements evolve through various forms of informal, semi-formal, and formal while maintaining a semantic link between these forms and, most importantly, conforming to the customer's requirements. A deep understanding of the complexity of the requirements model and its dynamics is critical in improving requirements engineering process management. This thesis investigates the benefits of documenting both the evolution of the requirements model and the rationale for that evolution. Design explanation explains and justifies the deliberations of, and decisions made during, the design activity. In this thesis, design explanation is used to describe the requirements engineering process in order to improve understandability of, and traceability within, the evolving requirements specification. The design explanation recorded during this research project is also useful in assisting the researcher in gaining insights into the creativity and opportunistic characteristics of the requirements engineering process. This thesis offers an interpretive investigation into incorporating design explanation within FOOM in order to extend and advantage the method. The researcher's interpretation and analysis of collected data highlight an insight-driven and opportunistic process rather than a strictly and systematically predefined one. In fact, the process was not smoothly evolutionary, but involved occasional 'crisis' points at which the model was reconceptualised, simplified and restructured. Therefore, contributions of the thesis lie not only in an effective incorporation of design explanation within FOOM, but also a deep understanding of the dynamic process of requirements engineering. The new understanding of the complexity of the requirements model and its dynamics suggests new directions for future research and forms a basis for a new approach to process management.

Service-oriented wireless sensor networks and an energy-aware mesh routing algorithm

Service-oriented wireless sensor networks (WSNs) are being paid more and more attention because service computing can hide complexity of WSNs and enables simple and transparent access to individual sensor nodes. Existing WSNs mainly use IEEE 802.15.4 as their communication specification, however, this protocol suite cannot support IP-based routing and service-oriented access because it only specifies a set of physical- and MAC-layer protocols. For inosculating WSNs with IP networks, IEEE proposed a 6LoWPAN (IPv6 over LoW Power wireless Area Networks) as the adaptation layer between IP and MAC layers. However, it is still a challenging task how to discover and manage sensor resources, guarantee the security of WSNs and route messages over resource-restricted sensor nodes. This paper is set to address such three key issues. Firstly, we propose a service-oriented WSN architectural model based on 6LoWPAN and design a lightweight service middleware SOWAM (service-oriented WSN architecture middleware), where each sensor node provides a collection of services and is managed by our SOWAM. Secondly, we develop a security mechanism for the authentication and secure connection among users and sensor nodes. Finally, we propose an energyaware mesh routing protocol (EAMR) for message transmission in a WSN with multiple mobile sinks, aiming at prolonging the lifetime of WSNs as long as possible. In our EAMR, sensor nodes with the residual energy lower than a threshold do not forward messages for other nodes until the threshold is leveled down. As a result, the energy consumption is evened over sensor nodes significantly. The experimental results demonstrate the feasibility of our service-oriented approach and lightweight middleware SOWAM, as well as the effectiveness of our routing algorithm EAMR.