Programming Resistive States of Memristive Devices via Current Control

EasyChair Preprint 8586

Abstract

Arrays of memristors can be used in promising neural computing systems as a programmable resistance (analog multiplication ratio) when performing operations of analog scalar multiplication of vectors, discrete in time. To form the required resistance, the memristor must be subjected to a "programming" procedure. The procedure can be carried out in two different ways, related to the different direction of movement according to the volt-ampere characteristic of the memristor. The paper deals with the proposed memristor elements programming device based on a digital-analog converter and a variable resistor. The paper says about the advantages and disadvantages of various existing and possible programming methods. The developed device is universal and does not require additional analysis of the output voltage.

Keyphrases: Artificial Intelligence, Memristor, analog neural calculations, programmable resistance, programming scheme

@booklet{EasyChair:8586,
  author    = {Evgeny Bukvarev and Ksenya Fomina and Semen Shabalin and Sergey Shchanikov},
  title     = {Programming Resistive States of Memristive Devices via Current Control},
  howpublished = {EasyChair Preprint 8586},
  year      = {EasyChair, 2022}}