UML-B Specification for Proven Embedded Systems Design

UML-B Specification for Proven Embedded Systems Design presents the perspective of the project on a Paradigm Unifying System Specification Environments for proven Electronic design (PUSSEE) as conceived in the course of the research during 2002 - 2003.

The goal of the research in this project was to introduce the formal proof of system properties throughout a modular system design methodology that integrates sub-systems co-verification with system refinement and reusability of virtual system components. This is done by combining the formal and informal means of specification by way of the UML and B languages to allow the verification of system specifications through the composition of proven sub-systems (with some particular attention to interfaces, in line with the VSIA/SLIF approach). The link of B with C, VHDL and SystemC extends the correct-by-construction design process to lower system-on-chip (SoC) development stages. The production of proven embedded software is therefore complemented by the production of proven hardware. Prototype tools have been developed for the code generation from UML and B, and existing B verification tools will be extended to support IP reuse, according to the VSIA recommendations. The methodology and tools were validated through the development of three industrial applications: a wireless mobile terminal a telecom system-on-chip based on HIPERLAN/2 protocol and an anti-collision module for automobiles.

The problem was known to be hard and the scope ambitious. But the seventeen chapters of UML-B Specification for Proven Embedded Systems Design, describing the main results obtained demonstrate the success of the research, acknowledged by the European reviewers. They are released to allow the largest audience including the various sectors of industry's system design engineers, university teachers and researchers. They will hopefully convince skeptical professionals that formal approaches can now cope with industrial strength problems.