Using a flexible real-time scheduling strategy in a distributed embedded application

The increasing complexity of current real-time applications demands systems to make good use of the available computational resources. Guaranteeing timing behavior using traditional real-time scheduling policies is turning to be unfeasible, as these policies are based in pessimistic worst-case assumptions that pre-allocate resources even though they will be rarely used. The Time-Aware Fault-Tolerant (TAFT) scheduler presents a flexible scheduling strategy that allows the construction of schedules maximizing the CPU utilization. This work presents the use of the TAFT scheduler with a computational intensive distributed embedded application, in order to keep a high ratio of successful tasks completion and so guaranteeing its computational progress. More specifically, the paper sketches a "TAFT-based" fusion algorithm for laser-scanner data and derives the requirements that the execution of this algorithm has on the CPU-scheduling.