Domain Specific Modelling for Reconfigurable Distributed Embedded Control Systems

distributed, embedded, control systems, domain specific languages, model driven software engineering, executable models, Petri nets, formal analysis, software methodologies

Construction process of Distributed Embedded Control Systems (DECS) is typically based on direct programming in low-level languages and compilation of binary executable intended to run on target platforms or operating systems. That makes the construction process very rigid and expensive. In addition to that, todays DECSs consist of tens or hundreds of interrelated nodes and thus represent a very complex environment for system programmers to maintain. Model-Driven Software Engineering (MDSE) for Embedded Systems solve the problem of complexity by a higher level of abstraction and formal definitions approach. In this paper we describe a methodology that bases on the usage of Domain Specific Modelling Languages (DSMLs) and formal methods for DECSs specification and their further transformation into the Reference Petri Nets (RPN) model directly runnable as a system simulation. The RPN implementation is finally translated into the interpretable form of the model and deployed on physical nodes of the system and run by a specialized virtual machine called Petri Nets Virtual Machine (PNVM). Among the major advantages of our approach, that will be explained in this paper belong: (i) the ability of domain experts to understand the DSML specification of the system and therefore directly participate within its construction process, (ii) formally defined, but still Touring complete intermediate language - reflecting the distributed, concurrent and synchronised features of the system, as well as (iii) the actor properties of every part of the system functionality and the possibility of its migration across the running model, and also the (iv) platform independence of each part of the system and the possibility to modify parts at run-time, as well as (v) the possibility of selected automated formal analysis of the constructed model of the system.