Research Seminar - December 14, 2000

Abstract

Most systems, whether hardware, software, mechanical or a hybrid, exhibit complex behaviour. This is a consequence of their construction from many interacting, concurrently active component parts. It is this concurrency-of-activity which causes such systems to be behaviorally complex, in contrast to, say, a sequential computation which can naturally be modelled by a function.

Ongoing research has resulted in the development of the Circal formalism for modelling concurrent systems. It has been used extensively to specify and formally verify synchronous and asynchronous digital systems (which are inherently concurrent), communication protocols, subtle timing properties of systems and the dynamics of highly concurrent spatial systems. A specification and verification environment, the Circal System, allows us to automatically verify the correctness of system properties.

Recent research addresses systems whose connective topology changes through time. This is in contrast to most systems whose structure remains fixed during its lifetime. Such dynamically reconfigurable systems may have agents becoming active, agents dying, and communication links forming and being destroyed at "run time". While my interests in this area arose from a need to provide a theory for dynamic reconfiguration of programmable chip-based architectures, and hence a basis for new programming languages and compilation techniques, these ideas also relate directly to mobility as found in the Internet and mobile, wireless systems.

The talk will discuss the significance of research into creating languages, formalisms and models for the complex communication oriented computing systems that are becoming commonplace today.