thesisPitch.md

A thesis

<!-- Context -->
<!-- Problem -->
<!-- Vision -->
My thesis is: a modeling framework should encapsulate time and uncertainty as first-class concepts and that enables efficient accesses (read, write), efficient process and easy manipulation.
in order to abstract the knowledge, i.e., the environment, the system and, the adaption process, of an dynamically adaptive system for improving the adaption process by taking more optimal decision with better performance.

Keywords: dynamically adaptive systems, knowledge, model-driven engineering, uncertainty modeling, time modeling

<!-- Open with a concrete problem  -->
<!-- Generalize -->

• DAS

• MAPE-k loop / control loop

• Knoweledge representation

• Inconsistent knoweledge

  • uncertainty
  • different time paces
  • contradictory information

• Creos smart grid

• Suboptimal decisions are taken
• Among the possible root causes, we focus oon two:
  • the adaptation process manipulates uncertain data
  • the time to execute the actions and to measure their effects is greater than the adaptation frequency

• Definition and representation of the knowledge for DAS
  • specify characteristic of data manipulated
  • efficient read/write/process
  • ease of manipulation

• Unified modeling framework
  • abstracts information related to
    • the adaptation process
      • actions
        • triggers
        • effects/impacts
      • specification
        • requirements
        • constraints
    • the environment of the system
      • behavior (rules which define how it evolves)
      • structure
    • the system itself
      • structure
      • behavior
      • specification
      • actuators
      • sensors
  • manages
    • temporal data
    • uncertain data
  • allows
    • efficient read
    • efficient write
    • easy read
    • easy write

• Definition of a language to define and manipulate uncertain model

  • allow to define uncertain data
  • allow to combine uncertain data and propagate the uncertainty throw this

• Definition of a temporal context model

  • can represent context over time
  • can represent past/on going and future actions
  • can represent action effects

• Error handling

• Expressiveness

• Performance evaluation

  • time to read/write/process data

• Validation industrial

  • implementation of (some ?) of the contributions into an industrial use case (Creos)

• Push the contrib. into GreyCat