Refine
Year of publication
Document Type
Is part of the Bibliography
- no (35)
Keywords
- AUTOSAR (2)
- Architecture Description Language (1)
- Goodness-of-fit (1)
- Model-Driven Development (1)
- Multi-Core (1)
- Multi-Objective Optimization (1)
- Multicore (1)
- Real-Time Systems (1)
- Real-time systems (1)
- Simulation (1)
Institute
Begutachtungsstatus
- peer-reviewed (4)
Partly Proportionate fair (Partly-Pfair) scheduling, which allows task migration at runtime and assigns each task processing time with regard to its weight, makes it possible to build highly efficient embedded multi-core systems. Due to its non-work-conserving behavior, which might leave the CPU idle even when tasks are ready to execute, tasks finish only shortly before their deadlines are reached. Benefits are lower task jitter, but additional workload, e.g. through interrupts, can lead to deadline violations. In this paper we present a work-conserving extension of Partly-Pfair scheduling, called PERfair scheduling and the algorithm P-ERfair-PD2 which applies Pfair modifications used for Partly-Pfair on the concept of ERfairness and PD2 policies. With a simulation based schedulability examination we show for multiple time base (MTB) task sets that P-ERfair- PD2 has the same performance as Partly-Pfair-PD2. Additionally, we show that P-ERfair- PD2 has a much higher robustness against perturbations, and therefore it is well suited for embedded domains, especially for the Automotive domain.
Eingebettete Systeme unterliegen neben den funktionalen Anforderungen besonders nichtfunktionalen Qualitätsanforderungen wie Effizienz, Zuverlässigkeit und Echtzeitfähigkeit. Mit steigendem Bedarf an Rechenkapazität können bisherige Konzepte zur Leistungssteigerung von Singlecore-Systemen jedoch nicht mehr eingesetzt werden - der Umstieg auf Multicore-Systeme wird erforderlich. Im zweiten Teil dieser Arbeit wird ein simulationsbasierter Ansatz zum Vergleich von Multicore-Scheduling-Algorithmen vorgestellt, mit dem Algorithmen für Multicore-Systeme mit voller Migration und dynamischer Task-Priorität untersucht werden. Wir erweitern diesen Ansatz um ein Verfahren zur Untersuchung einer Tasksetmenge mit stochastisch beschriebenen Eigenschaften und vergleichen ihn mit den im Teil 1 beschriebenen Algorithmen BinPacking-EDF und P-ERfair-PD² für eine Gruppe von Automotive Powertrain Systemen.
Proportionate fair (Pfair) scheduling, which allows task migration at runtime and assigns each task processing time with regard to its weight, is one of the most efficient group of SMP multiprocessor scheduling algorithms known up to now. Drawbacks are tight requirements to the task system, namely the restriction to periodic task systems with synchronized task activation, quantized task execution time, and implicit task deadline. Most likely, a typical embedded real-time system does not fulfill these requirements. In this paper we address violations of these requirements. For heterogeneous task systems, we define the multiple time base (MTB) task system, which is a less pessimistic model than sporadic task systems and is used for automotive systems. We apply the concept of Pfair scheduling to MTB task systems, called partly proportionate fair (Partly-Pfair) scheduling. The restrictions on MTB task systems required for Partly-Pfair ness are weaker than restrictions on periodic task systems required for Pfair ness. In a simulation based study we examined the performance of Partly-Pfair-PD and found it capable to schedule feasible MTB task sets causing a load of up to 100% of the system capacity.
The shift from single-core to multi-core processors in real-time embedded systems leads to communication based effects on timing such as inter-core communication delays and blocking times. Moreover, the complexity of the scheduling problem increases when multi-core processors are used. In priority-based-scheduling, a fixed priority assignment is used in order to enable predictable behavior of the system. Predictability means that the system has to be analyzable which allows the detection of problems coming from scheduling decisions. For fixed priority scheduling in multi-core real-time embedded systems, a proper task priority assignment has to be done in a way that the system has minimal effects on timing. In this paper, we present an approach for finding near-optimal solutions for task priority assignment and the preemption/cooperation problem. A genetic algorithm is hereby used to create priority assignment solutions. A timing simulator is used for evaluation of each solution regarding real-time properties, memory consumption and communication overhead. In a case study we demonstrate that the proposed approach performs better than well known and single-core optimal heuristics for relatively complex systems.
We present in this paper a new lock-based resource sharing protocol PWLP (Preemptable Waiting Locking Protocol) for embedded multi-core processors. It is based on the busy-wait model and works with non-preemptive critical sections while task may be preempted by tasks with a higher priority when waiting for resources. Our protocol can be applied in partitioned as well as global scheduling scenarios, in which task-fix, job-fix or dynamically assigned priorities may be used. Furthermore, the PWLP permits nested requests to shared resources. Finally, we present a case study based on event-based simulations in which the FMLP (Flexible Multiprocessor Locking Protocol) and the proposed PWLP are compared.