Abstract
Control systems and real-time computing have for a long time been associated in control systems, with the aim of controlling a process and bringing it towards a state which complies with the objectives specified by the user. The process is often a phys- ical device, for instance mechanical (rolling mills), electromechanical (DVD player, robots), thermal (internal combustion engines), chemical (reactors), hydraulic (energy production). . . These can be complex processes associating several of these technolo- gies, for instance in terrestrial, aerial or underwater vehicles. They can also be com- puting components (scheduling of tasks, network gateway, website management) or electronic (power supply of a chip, phase locked loop), or even simulated components (avatar control in a virtual world, "hardware-in-the-loop" real-time simulators). The increasing complexity of these systems requires reviewing their properties in order to better integrate the control system design and implementation constraints executed on computing systems. The first section will recall the main properties of closed loop control as well as the constraints and limitations induced by their digital implementation. Section 1.2 examines how the control task scheduling constraints can be relaxed by exploiting these properties. The design of scheduling controllers can be performed in the formalism of sampled systems (Section 1.3), but also in that of weakly strong real-time scheduling (Section 1.4). Finally, an example of designing and implementing the scheduling control of tasks in a video decoder will be detailed in Section 1.5.