Fault—tolerant k—out—of—n logic unit networks
Eiji Fujiwara, Kohji Matsuoka · Systems and Computers in Japan · 1988
Abstract For the automatic repair of a fault in the on‐line system, and for yield improvement in LSI fabrication, a system structure must be developed where spare parts are provided and used whenever necessary. This paper considers the general logical network, which is composed of such units as gate and memory. It proposes a new switching and connection structure in order to mask the effect of faults. Thek‐out‐of‐nunit group (called a block) is operated withkunits (k≤n) being used in the normal operation and the remaining (k‐n) units being used as spares. Thenswitching units, instead of thektraditional switching units cover the faults in the switching units. Assuming this structure, a connection which theoretically minimizes the number of connections between thek‐out‐ofnand thek‐out‐of‐mblocks (m, n≥k) is described, presenting a redundant construction which improves the reliability, with minimum redundancy. The proposed construction is applied to the yield improvement of the large‐sized LSI chips, and it is shown that the method can improve the yield to any value by increasing the number of blocks, and can realize a flexible and an easy design in the redundancy structure.