An Aging-Aware Routing Algorithm for Improving Reliability in NoC-Based Multicore Systems.

Lu Yin,Yi Zhang,Junlong Zhou,Qing Chang,Jin Sun
DOI: https://doi.org/10.1109/HPCC/SmartCity/DSS.2019.00240
2019-01-01
Abstract:As the interconnection scale of network-on-chip (NoC)-based multicore systems becomes more complicated and the characteristic size of the transistor continues to shrink, aging-induced reliability issues has become the main concerns in today's NoC design paradigm. Considering the NBTI aging effect upon NoC routers, this paper proposes an aging-aware routing algorithm that aims to improve the reliability of NoC system while optimizing routing power consumption. The proposed method introduces an concept of aging factor (AF) to evaluate the stress level for each router. We further formulate a rigorous optimization model that imposes a constraint upon NoC router's utilization according to its AF metric. The optimization model is to determine an optimized routing solution such that the total power consumption for all routers is minimized under the AF constraints. To solve the formulated optimization problem, we develop an efficient routing algorithm based upon the search strategy of ant colony optimization. The routing solution obtained by the aging-aware algorithm is beneficial for improving the reliability of NoC system by ensuring lower utilization of stressed routers. Experimental results demonstrate that, compared with the non-aging-aware routing strategy, the proposed algorithm can significantly improve system reliability with an insignificant increase in total power consumption.
What problem does this paper attempt to address?