Constructing spare sharing networks for reliability enhancement of scalable systems

Future systems based on nano-scale devices will be highly susceptible to operational faults. When hardware redundancy is used to boost reliability, through the repair-based approaches, redundancy is most effectively utilized if any spare unit can be made available to replace any faulty unit. However, such fully connected network of spare sharing demands dreadfully expensive interconnection when system size and complexity scale up. Consequently, while spare sharing provides flexibility, the sharing must be limited due to interconnection constraints. In this paper, we study the reliability issue of systems with shared spare units, under limited interconnection constraints. Particularly, we focus on the construction of the “spare sharing network” to boost system reliability. The proposed methodology connects the vulnerable units to the exploitable existing spares, thus strengthening the entire system without adding extra spares. Simulation results confirm that the proposed methodology achieves a significant reliability boost at very small cost of additional interconnects.