This thesis handles the Distributed Real-T … This thesis handles the Distributed Real-Time Operating Systems thematic. In the industrial world, these systems are fundamental components since they distribute the tasks through a number of cells and operate in such a manner that, in normal working conditions, neither equipement nor human lives are placed in jeapordy.
The first intervention of this thesis focus on the developement of analytical conditions to ensure that the tasks fulfill their deadlines. If the system does not obey these mathematical conditions then it is unschedulable. In particular, the targeted system model focus on assynchronous periodic tasks together with sporadic tasks, using a fixed-priority preemptive scheduler.
The second intervention approaches the ARINC 653 specification, that comes from the aeronautical industry, and its implementation using COTS RTOS. In particular, the RTEMS OS is used as a case study. Two architectural solutions are presented where the fittest, regarding the target platform
restrictions, is chosen.
Lastly, the I/O devices are handled to make them fault tolerant using exclusively hardware mechanisms already present in the native CPU (system clock). If the number of events triggered by the external hardware is too high, the system can enter an overload state and the remaining tasks, which can be more critical, can miss their deadlines. This thesis proposes a mechism that temporarily inhibits event processing when it detects an overlad scenario (e.g. event rate too high, minimum inter-arrival time violated, etc). minimum inter-arrival time violated, etc).