Abstract
E. Ledinot presented the evolution of avionics computer systems, which grew in size by a factor of 100 between 1980 (Mirage F1) and 2000 (Rafale 2000), with increasingly modular avionics. He detailed the problems of sequencing and parallelizing functions, and then showed how difficult it is to achieve these using conventional deterministic programming methods, which require everything to be scheduled by hand. He explained why Esterel solves this problem, automatically ensuring correct scheduling by construction and generating code with predictable performance. He showed why the composition of subsystems developed by different people leads to complex causality problems, solved in the latest versions of the Esterel v5 compiler based on constructive semantics. Finally, he showed why the new SCADE 6 system described below is an essential tool for the even greater complexity of future applications, as it unifies continuous and discrete control in a graphical formalism familiar to engineers and mathematically well-defined.
Emmanuel Ledinot
E. Ledinot is Director of Upstream Scientific Studies at Dassault Aviation. He and his team have pioneered the use of formal methods, particularly Esterel and its formal verification tools. Esterel could not have reached maturity without the examples he provided, of a size and complexity far greater than those we imagined in the laboratory.