Optimal group diagnosis procedures for VLSI/WSI array architectures

Addresses the problem of partitioning VLSI/WSI (wafer scale integration) array architectures into disjoint maximal diagnosis blocks (MDBs) and finding an optical group diagnosis policy for testing and locating faulty elements (modules) in these MDBs. The optimization criterion is to minimize the accumulated diagnosis cost in deriving a feasible reconfiguration solution. The technique for partitioning an array is based on the parallel partition approach. The problem of finding an optimal group diagnosis procedure is modeled as a (t+1)-ary decision tree, where t is the size of an MDB. Instead of dealing with a (t+1)-ary decision tree directly, the problem is further reduced to that of handling a binary decision tree or a block-walking representation. Properties related to the group diagnosis procedures and binary decision trees are derived. Simulation results are provided to further support the effectiveness of the proposed approach over other approaches