Efficient Signaling for Passive Memristive Crossbars to Prepare them for Spiking Neuromorphic Computing

Abstract

Memristive devices act as promising emerging devices to emulate the functionalities of the synapses and neurons. The memristive crossbars provide accelerated computing for various IoT and edge computing applications. Brain-inspired computing can be exploited to overcome the von-Neuman computers bottleneck to increase energy efficiency and accelerate processing, specifically in conceptual computing. All neuro-inspired computing methods are set in two platforms, including conventional and spiking neural networks. Due to the challenge in providing efficient signaling for the spiking computing process, this manuscript focuses on developing a signaling strategy and methodology for conceptual processing over the passive memristive crossbars. The efficient signaling enables the passive memristive crossbars to be utilized for energy-efficient computing, providing high dense circuit implementation.