Statistical Analysis for Memristor Crossbar Memories

Rawan Naous, et al., "Statistical Analysis for Memristor Crossbar Memories." Int. Journal of Unconventional Computing 12, 2016, 251.

The inherent memory capabilities bound with compactness and scalability features pose the memristor as an ideal rival for conventional nonvolatile memories. Its use within a crossbar structure offers high-density storage and tight dimensionality. A fitting parameter in the midst of the trend towards nanoscale integration, where gateless memories are on the rise in an attempt to achieve further space savings. Nonetheless, the gate sacrifice comes at the expense of the fidelity and accuracy of the readout values. In which the sneak path phenomenon distorts the data and acts as a hindering factor to reliable detection and higher density attainment. In this work, a novel approach is adopted to accommodate the sneak path and counter its effect on the memory reading. In contrast to the alternative techniques, where spatial and temporal solutions are applied to alleviate the distortion limitation and set a dynamic threshold, statistical measures benefit from the prior read data within the array. It builds upon the noise reduction and estimation principles, mainly borrowing concepts of coding and detection theory to enhance the access time and accuracy of the reading process.