Sequential redundancy identification using recursive learning

Abstract

A sequential redundancy identification procedure is presented. Based on uncontrollability analysis and recursive learning techniques, this procedure identifies c-cycle redundancies in large circuits, without simplifying assumptions or state transition information. The proposed procedure can identify redundant faults which require conflicting assignments on multiple lines. In this sense, it is a generalization of FIRES, a state-of-the-art redundancy identification algorithm. A modification of the proposed procedure is also presented for identifying untestable faults. Experimental results on ISCAS benchmarks demonstrate that these two procedures can efficiently identify a large portion of c-cycle redundant and untestable faults.