Information Density in Multi-Layer Resistive Memories

TL;DR

This paper analyzes the information capacity of multi-layer resistive memories, addressing the sneak-path problem and deriving bounds for information density with and without peripheral circuitry.

Contribution

It provides exact and asymptotic bounds on the storage capacity of multi-layer resistive memories, introducing an encoding scheme that approaches capacity.

Findings

Information density remains non-zero in the limit for extreme aspect ratios.

Derived exact and asymptotic bounds for single- and multi-layer arrays.

Presented an encoding scheme that asymptotically achieves capacity.

Abstract

Resistive memories store information in a crossbar arrangement of two-terminal devices that can be programmed to patterns of high or low resistance. While extremely compact, this technology suffers from the "sneak-path" problem: certain information patterns cannot be recovered, as multiple low resistances in parallel make a high resistance indistinguishable from a low resistance. In this paper, a multi-layer device is considered, and the number of bits it can store is derived exactly and asymptotic bounds are developed. The information density of a series of isolated arrays with extreme aspect ratios is derived in the single- and multi-layer cases with and without peripheral selection circuitry. This density is shown to be non-zero in the limit, unlike that of the arrays with moderate aspect ratios previously considered. A simple encoding scheme that achieves capacity asymptotically is presented.