Reliability-Aware Task Scheduling using Clustered Replication for Multi-core Real-Time systems
Alireza Namazi, Meisam Abdollahi, Saeed Safari, Siamak Mohammadi, Masoud Daneshtalab · 2016
This paper proposes a model for a new reliability-aware task scheduling method for hard real-time multi-core systems. The proposed method is based on a novel clustered replication which maintains the desired reliability threshold, minimizing both inter-core communication and redundancy overhead in multi-core network-on-chip based platforms. Both single and multiple errors are considered in this method. Experimental results show that the proposed method can schedule hard real-time tasks with relatively lower latency and better communication overhead in comparison with the conventional replication method. The proposed method can reduce the execution latency and communication volume about 29.4% and 30.1% in comparison with conventional replication, respectively. Results show that the redundancy increase of the proposed method is about 60% less than conventional replication and depends on the defined reliability threshold of the system.