Bio-inspired fault-tolerance

In the last decade, there has been a considerable increase of interest in fault-tolerant computing due to dependability problems related to process scaling, embedded software, and ubiquitous computing. In this paper, we consider an approach to fault-tolerance which is inspired by gene regulatory networks of living cells. Living cells are capable of maintaining their functionality under a variety of genetic changes and external perturbations. They have natural self-healing, self-maintaining, self-replicating and self-assembling mechanisms. The fault-tolerance of living cells is due to the intrinsic robustness of attractors' landscapes of their gene regulatory networks. Previously, we introduced a technique which exploits the stability of attractors to achieve a fault-tolerant computation. In this paper, we evaluate this technique on the example of a gene regulatory network model of Arabidopsis thaliana and show that it can tolerate 70% single-point mutations in the outputs of the defining tables of gene functions.