We address the problem of synthesizing real-time embedded controllers, taking into account constraints deriving from the implementation platform. Specifically, we address the problem of controlling a set of independent systems by using a shared single CPU processor board. Assuming a time-triggered model of computation for control processes and a real-time preemptive scheduling policy, the problem of optimizing robustness of the controlled systems is formulated. Decision variables of the optimization problem are the processes' activation periods and the feedback gains. The analytical formulation of the problem enables an efficient numerical solution based on a branch and bound scheme. The resulting design is directly implementable without performance degradations that may otherwise arise due to scheduling and execution delays.
1Postdoctoral Researcher, ReTiS Lab-Scuola Superiore S. Anna, Italy
2Professor, Università di Pisa, Italy