Task Recomputation Based Reliability Improvements In Heterogeneous Multi-Core Systems

Reliability issues in embedded systems are becoming limiting factors while the demand for high performance and computational complexity continue to increase. Especially, systems operating under harsh conditions are prone to generate erroneous results. As embedded systems already house large datasets, traditional techniques to improve reliability such as task redundancy can be very costly in terms of memory space consumption and hence may cause performance degradation. In this paper, we propose a task-recomputation based approach that utilizes idle time frames of computational resources in order to iteratively recompute tasks to increase reliability of overall design without incurring any additional execution latency, area or extra memory space. We present our experimental evaluation and compare our results with a state-of-the-art technique. The experimental results collected using both task graphs extracted from benchmarks and automatically-generated task graphs show the effectiveness of the proposed approach.