Solving AI problems with memristors: A case study for optimal

Abstract

This paper presents a novel circuit-level Cellular Automata (CA)-inspired computational scheme capable of executing computations within memory. The proposed computing structures exploit the threshold-based resistance switching behavior of memristors and of their multi-state composite components. Array-like circuit structures with memristors are designed and their ability to efficiently solve the classic "bin packing" problem is verified via a simulation-based validation using a published memristor device model. A fundamental memristive cell which implements a one-dimensional CA rule is described in detail and then employed in a sophisticated two-dimensional array able to execute the "First-Fit" (decreasing) bin packing algorithm.