The Event-based semantics workshop April 3, 2007 assoicated to RTAS 2007
Summary of discussion.


Need:  

There is a shift towards openness, federation, for real time, multi-scale, wide
area, critical systems.  All this still with certification, 
a major verification challenge.  One aim of a theory of event-based systems
is to support specification and verification of such systems.

The workshop consisted of a series of tutorial talks covering a wide range of
concepts, techniques and issues related to event-based semantics, with
a concluding discussion session and substantial discussion during the talks.

The slides will be posted on the workshop website  
   (http://blackforest.stanford.edu/eventsemantics/)

The following is a very high-level summary of the topics of the talks.

* Compositional event-based specification and semantics of actor systems (events
  are message receives) at multiple levels of abstraction, with mappings between
  different semantics.

* A game theoretic approach to synthesizing resource managers (more generally of
  coordinators) from interface specifications of resource related events.

* Semantic units (models of computation specified in a common formal framework) in
  the context of model-base development used to ground composition of heterogeneous
  components and support prototyping and analyses.

* A mathemathical theory based on topological concepts to model components and
  functions mapping signals (abstract time series of events) and their wiring
  together with feedback loops.

* A methodology for specifying distributed systems in a Logic of Events, a refining
  to a conjunction of realizable constraints from which code can be generated.

* Languages to express events at the sensor level and map these to events and
  higher levels of abstraction to bridge the gap between physical and logical
  worlds.

* Probability distributions on intervals to manage uncertainty in observation and
  requirements for raising alerts.


Several notions of event, component, interface, and composition 
emerged in the talks and discussion.  Each with their own relevance.
One challenge is to have a coherent framework that can accomodate
the different viewpoints, combine and move between them in 
semantically meaningful ways.

Some challenges brought out in the windup discussion.

  System stability -- what are the boundaries?

  Time resolution of observation -- must be communicated in time to be useful.
  An example is subsecond control on a power grid

  Change of event structure over time, for example traffic control for
  high density landing of aircraft.   Weather and traffic patterns can
  change things.

  Practical problem -- systems are (usually) built on top of an OS!
  What are the principled `quarks' can we build upon.

  Event based observations for diagnostics. Can system structure be inferred?

  Developing the theory of event-based systems, which currently is less
  well-developed than the theory of state based systems.


Possible focus points for the next workshop 
  How to tie the formal to the practical
    traceability of properties is important
  Open verification technology --- what can be done automatically, online/offline
  Formally based testing
  Challenge problems
    maybe aim for something modest but doable such as a distributed sensor system?