Citation Context

...ven if optimization is considered, it is very limited and considered in isolation, and thus is not reflected at all levels of the design process, including mapping, scheduling and energy minimization =-=[11]-=-. Regarding energy minimization, the most common approach that allows energy/performance trade-offs during run-time of the application is dynamic voltage scaling (DVS) [20]. DVS aims to reduce the dyn...

Citation Context

...ven in the presence of faults. Fault tolerance can be addressed with hardware architecture solutions, such as TTA [17], or software-based solutions such as re-execution, replication and checkpointing =-=[25]-=-. In this paper we do not address the issue of fault-tolerance (which is orthogonal to our problem), and we assume that the designer has developed the applications such that they provide the required ...

Citation Context

...Embedded Technology and Applications Symposium 1080-1812/10 $26.00 © 2010 IEEE DOI 10.1109/RTAS.2010.31 89 even in the presence of faults. Fault-tolerance has been addressed separately for hard [14], =-=[15]-=- and soft [16] real-time systems. In [17] we have shown how to handle permanent faults in mixed hard/soft systems through task migration, such that hard deadlines are still satisfied and the soft task...

Citation Context

...k While there is an extensive literature on fault tolerance for distributed systems [15], [16], our approach focuses on embedded control applications, and it is closer to the works by Izosimov et al. =-=[17]-=-–[19]. They address the problem of transient faults in the processing elements, assuming a model that combines time-triggered communication protocol [20] and cyclic static scheduling of the processors...

Citation Context

...h real-time systems robustness in presence of faults. Some of them propose diverse approaches for robust scheduling algorithms [11]-[13], others are focused on the task reexecution and/or replication =-=[17]-=- and finally, another group of researchers proposes additional application tasks in order to check the application tasks workspace memory [16]. As a general characteristic of existing works, they prop...

Citation Context

... in the presence of faults. Fault tolerance can be addressed with hardware architecture solutions, such as TTA [1], or software-based solutions such as re-execution, replication and checkpointing. In =-=[10]-=- we have shown how checkpointing and active replication can be combined in an optimized implementation that leads to a schedulable fault-tolerant application without increasing the amount of employed ...

Citation Context

...l to the period of the soft task. Figure 1: Optimal vs Greedy mapping quirements of A can no longer be satisfied. Transient faults are tolerated using equidistant checkpointing with rollback recovery =-=[15]-=-, which uses time redundancy to tolerate transient faults. Permanent faults are tolerated using task migration, i.e., starting the safety-critical tasks on other processors. The principle of checkpoin...

Citation Context

...sks multiple times and perform voting of the results [24]. Another common technique is to insert checkpoints into the software and rollback the execution from a safe state in case faults are detected =-=[18, 26]-=-. For real-time applications, temporal redundancy must be used with utmost care, since the overhead in time may lead to deadline violations. The schedulability issue in the context of temporal redunda...