Citation
Anne Ngu, George Chin Jr.. "Context Aware Actors". Talk or presentation, 16, February, 2011; Presented at the Ninth Biennial Ptolemy Miniconference, Berkeley, CA.

Abstract
Data-intensive scientific workflows are traditionally modeled using a dataflow-oriented model. The simplicity of a dataflow model facilitates intuitive workflow design, analysis, and optimization. However, some amount of control-flow modeling is often necessary for engineering fault-tolerant, robust, and adaptive workflows. In scientific domains and cyber physical systems, myriads of environmental information are needed for the effective control of different stages of coordination and execution. Modeling various fine-grained controls using inherently dataflow-oriented actors will quickly result in a workflow that is hard to comprehend, reuse, and maintain. We present a context annotation framework for Kepler/Ptolemy that enables actors to exploit the dynamic environmental information during runtime without introducing complex control-flow actors or low-level re-coding of existing actors to cover all possible dynamic conditions. Contexts can refer to all data or events that can be sensed automatically from the physical environment. They can affect an actor's behavior but is not part of the original actor's design. Context should be modeled separately as an entity that can be read by any actor in many different ways, for example, via global parameters, an actor's ports, or an actor's parameters. To support the context annotation process, three key components are added to the existing Kepler/Ptolemy system: 1) context modeling and provisioning, 2) context binding, and 3) context triggering. The incorporation of context-aware actors in a scientific workflow results in a scientific process that is aware of its environment during execution and that responds intelligently based on such awareness without having to hard-code or entangle complex reactionary logic in actors. A further advantage of annotating an actor with contexts is that the defined contexts can be reused and shared across other workflows via a model-driven design approach without writing any low level codes. Context annotation also hides the complexity of using complex workflows by providing high-level concepts for initializing and configuring multiple parameters that have implicit dependencies. This paper presents the motivation for and system design and implementation of a context annotation process in Kepler/Ptolemy.

Electronic downloads

• 11NguContext_Actor.pdf · application/pdf · 323 kbytes

• HTML

  Anne Ngu, George Chin Jr.. <a
  href="http://chess.eecs.berkeley.edu/pubs/815.html"><i>Context
  Aware Actors</i></a>, Talk or presentation,  16,
  February, 2011; Presented at the <a
  href="http://ptolemy.eecs.berkeley.edu/conferences/11"
  >Ninth Biennial Ptolemy Miniconference</a>,
  Berkeley, CA.

• Plain text

  Anne Ngu, George Chin Jr.. "Context Aware Actors".
  Talk or presentation,  16, February, 2011; Presented at the
  <a
  href="http://ptolemy.eecs.berkeley.edu/conferences/11"
  >Ninth Biennial Ptolemy Miniconference</a>,
  Berkeley, CA.

• BibTeX

  @presentation{NguChinJr11_ContextAwareActors,
      author = {Anne Ngu and George Chin Jr.},
      title = {Context Aware Actors},
      day = {16},
      month = {February},
      year = {2011},
      note = {Presented at the <a
                href="http://ptolemy.eecs.berkeley.edu/conferences/11"
                >Ninth Biennial Ptolemy Miniconference</a>,
                Berkeley, CA.},
      abstract = {Data-intensive scientific workflows are
                traditionally modeled using a dataflow-oriented
                model. The simplicity of a dataflow model
                facilitates intuitive workflow design, analysis,
                and optimization. However, some amount of
                control-flow modeling is often necessary for
                engineering fault-tolerant, robust, and adaptive
                workflows. In scientific domains and cyber
                physical systems, myriads of environmental
                information are needed for the effective control
                of different stages of coordination and execution.
                Modeling various fine-grained controls using
                inherently dataflow-oriented actors will quickly
                result in a workflow that is hard to comprehend,
                reuse, and maintain. We present a context
                annotation framework for Kepler/Ptolemy that
                enables actors to exploit the dynamic
                environmental information during runtime without
                introducing complex control-flow actors or
                low-level re-coding of existing actors to cover
                all possible dynamic conditions. Contexts can
                refer to all data or events that can be sensed
                automatically from the physical environment. They
                can affect an actor's behavior but is not part of
                the original actor's design. Context should be
                modeled separately as an entity that can be read
                by any actor in many different ways, for example,
                via global parameters, an actor's ports, or an
                actor's parameters. To support the context
                annotation process, three key components are added
                to the existing Kepler/Ptolemy system: 1) context
                modeling and provisioning, 2) context binding, and
                3) context triggering. The incorporation of
                context-aware actors in a scientific workflow
                results in a scientific process that is aware of
                its environment during execution and that responds
                intelligently based on such awareness without
                having to hard-code or entangle complex
                reactionary logic in actors. A further advantage
                of annotating an actor with contexts is that the
                defined contexts can be reused and shared across
                other workflows via a model-driven design approach
                without writing any low level codes. Context
                annotation also hides the complexity of using
                complex workflows by providing high-level concepts
                for initializing and configuring multiple
                parameters that have implicit dependencies. This
                paper presents the motivation for and system
                design and implementation of a context annotation
                process in Kepler/Ptolemy.},
      URL = {http://chess.eecs.berkeley.edu/pubs/815.html}
  }
  

Posted by Christopher Brooks on 18 Feb 2011.
Groups: ptolemy
For additional information, see the Publications FAQ or contact webmaster at chess eecs berkeley edu.