DEGAS at Edinburgh
Computers and the networks which link them have become ubiquitous, and
as a result new forms of applications are being developed to take
advantage of the opportunities this offers. However, as yet,
scientific investigation of tools to support this development has been
limited. The Commission of the European Union has recent launched the
Global Computing
pro-active initiative within the Future and Emerging
Technologies part of the Information Society
Technologies programme. This initiative funds several projects to
begin on January 1st 2002,
of which Design Environment for Global
Applications (DEGAS) (IST-2001-32072) is one.
This is the local Edinburgh DEGAS page: there is also a main DEGAS page.
Project abstract
DEGAS aims to combine structured (semi-formal) graphical methods for
specification of global applications with formal methods for their
analysis and verification. We will investigate the extent to which UML
is already suitable to model global applications and we will propose
extensions. We will establish formal models of these applications based
on the operational semantics of foundational process calculi for
mobility. Our development of static and dynamic analysis will
concentrate on two key-features of global computing: performance
prediction and security. We will assess the foundational work in a
prototypical proof-of-concept environment that hides the formal
treatment from the user when possible. We will tune our development by
undertaking case-studies on wireless telecommunication applications.
Partners
Academic
Industrial
Edinburgh's role
The people working on DEGAS in Edinburgh will be
Stephen Gilmore,
Jane Hillston and
Perdita Stevens (site
leader), and two research fellows, Catherine Canevet and one other to
be appointed.
There are two strands to Edinburgh's main contribution to the project:
- UML for global applications and associated tool support. Based on
discussions with the partners we will define a UML profile - that is,
a dialect of UML defined in a standard way - for the design of global
applications. We will also lead the development of the architecture
for the toolset. We hope to use a standard UML tool as a front end,
connected via XMI based extractors to formal tools as the backend.
- The dynamic characteristics of global applications are influenced
by both the computational and communication aspects of the systems.
However timely behaviour is crucial for many applications and thus it
is essential that support for performance prediction should be
included in any design environment. We will seek to define formal
modelling constructs which reflect the UML profile of global
applications. These constructs will be used to automatically derive
stochastic models suitable for performance analysis. Moreover we will
seek to exploit these constructs in an investigation of efficient
solution techniques and data structures for carrying out the analysis.
Here is a page about a very tiny
demonstrator relevant to both strands.