A Weighted Sensing Scheme for ReRAM-Based Cross-Point Memory Array

Abstract

In recent years, the design of cross-point array based on resistive random access memory (ReRAM) has been widely investigated because it offers extremely high storage density and low power consumption. However, the sneak-path leakage in such a resistive network is inevitable and grows fast as the array size increases, which severely constrains the scalability of cross-point array. This work proposes a novel weighted sensing scheme which suppresses the sneak-path leakage via precise voltage control of unselected memory cells. Furthermore, different weights are assigned to ReRAM cells, which can be read out at once using one single sensing component. The proposed weighted sensing scheme makes the implementation of large cross-point arrays possible meanwhile successfully reduces the area cost of peripheral circuitry. The effectiveness of the proposed design has been validated and studied via thorough circuit simulations.