Fault Detection of Bloom Filters for Defect Maps

Bloom filters can be used as a data structure for defect maps in nanoscale memory. Unlike most other applications of Bloom filters, both false positive and false negative induced by a fault cause a fatal error in the memory system. In this paper, we present a technique for detecting faults in Bloom filters for defect maps. Spare hashing units and a simple coding technique for bit vectors are employed to detect faults during normal operation. Parallel write/read is also proposed to detect faults with high probability even without spare hashing units.