Research

We shall build on our expertise in terms semantic models respectively for timed and synchronous systems (Aoste), distributed and asynchronous systems (Oasis), and build extensions of these models that can be combine in a coherent framework. The idea is not to reinvent from scratch a new framework that would replace existing and well-adopted theories, but rather to keep from each domain their main features and strength, build links between them, and build a coherent framework that will address the Cyber-Physical system requirements.

Concretely, this will give birth to extensions of the respective meta-models (at the level of abstract syntax or meta-models) and of their behavioural semantics. In a second step, extensions or combination of the corresponding modeling tools will be addressed.

For real-time systems, many temporal logics have been proposed (timed automata, Timed CCS, Timed CSP, TCTL, Real-Duration Calculus, etc.), good at specifying temporal restrictions for real time systems. A proposal for a clock specification language (CCSL) was made and standardized at OMG as part of the MARTE UML profile; it allows representing such kind of models with engineering model-driven techniques.

This project will extend one or more of these logic systems so that they can really describe the integration of computation and physical processes of E&DS. In a second step, prototype tools implementing these formalisms will be developed and connected to our existing analysis tools.

More precisely, we want to establish an integrated formal analysis method for specifying the temporal and spatial consistency, combining calculative and physical components, considering collaboration, mobility and heterogeneity. Besides considering constraints of hard real-time and soft-time, we emphasize spatial constraints, namely, how to express the location and how to determine the spatial restrictions thus integrate temporal and spatial constraints to characterize the essential of E&DS.

The E&DS are designed in such a way that collection of various (embedded) computational resources, connected by an interconnection network, collaborated as a single system. In this project, we consider the task scheduling and load balance in the distributed environment with soft real-time limit and pre-emptive priority based on the ASP (Asynchronous Sequential Processes) calculus. These two requirements in distributed systems are now emerging for cyber physical systems (CPS). We want to develop formalized primitives and related scheduling algorithms for task allocation and load balancing with integration of machine performance evaluation, task attributes and scheduling strategy.

An important aspect of the aforementioned CCSL formalism is that it allows explicit representation of static regular schedule, as syntactic objects which provide the explicit activation patterns for events and operations. Such features have been proved to match many important cases in classical scheduling theory, such as modulo and affine scheduling as found in process network, nested loop programs with affine bounds, and software pipelining optimization. The periodic subdomain of real-time scheduling can be covered as well (with possibly multirate periods). Dynamic settings with pre-emptive tasks and dynamic priority issues remain problematic, as the ordering now must be established at runtime and cannot be computed and easily encoded into a syntactic pattern beforehand. Opportunities for generalization in dedicated specific cases will be investigated in the course of the project, and form a whole area of research therein.