Automated minimization of concurrent online checkers for Network-on-Chips

The paper introduces automated minimization of a set of concurrent online checkers for Network-on-Chips (NoCs) under given fault detection quality constraints. The proposed framework allows accurate and complete evaluation of the fault detection capabilities of checkers, which in turn enables finding seamless trade-offs between the overhead area of the checkers and the fault detection quality. The features of the automated minimization approach include formal proof for the absence or presence of true misses in checkers and a minimal fault detection latency. The minimization technique is based on a divide-and-conquer approach of partitioning the checkers´ fault table into independent clusters. The checkers within the cluster are weighted and the set of checkers is minimized based on a heuristic method. Experiments on the control part (routing and arbitration) of an NoC router show that 100% fault coverage with very low overhead area will be achieved by the proposed minimization approach.