Neuromemristive Circuits for Edge Computing: A Review

Abstract

The volume, veracity, variability and velocity of data produced from the ever increasing network of sensors connected to Internet pose challenges for power management, scalability and sustainability of cloud computing infrastructure. Increasing the data processing capability of edge computing devices at lower power requirements can reduce several overheads for cloud computing solutions. This paper provides the review of neuromorphic CMOS-memristive architectures that can be integrated into edge computing devices. We discuss why the neuromorphic architectures are useful for edge devices and show the advantages, drawbacks and open problems in the field of neuro-memristive circuits for edge computing.

Learning in Memristive Neural Network Architectures Using Analog Backpropagation Circuits

“…Three main brain inspired learning architectures that we consider in this work are neural networks [5], [16], [17], [20], HTM [18] and LSTM [19].…”

“…As large scale crossbars usually suffer from leakage currents, the most widely used architecture for the crossbar synapses is 1 transistor 1 memristor 1T 1M [20], [59], [60]. Different variants of transistors and selector devices are used in the literature for the crossbar architecture for improving the crossbar performance.…”

2D Layered Materials for Memristive and Neuromorphic Applications

“…[13,77,78] The edge computing could alleviate burdens of data transmission between edge devices and large data center and accelerate data processing in large data centers. [13,77,78] The edge computing could alleviate burdens of data transmission between edge devices and large data center and accelerate data processing in large data centers.…”

“…[13,77,78] The edge computing could alleviate burdens of data transmission between edge devices and large data center and accelerate data processing in large data centers. [13,14] Nevertheless, the challenges in power con-sumption, device reliability, and high-density integration hinder the development of TMOs based memristive devices. [13,14] Nevertheless, the challenges in power con-sumption, device reliability, and high-density integration hinder the development of TMOs based memristive devices.…”

“…These features make them ideal candidates for applications in memory devices, [3,5,9] in-memory computing, [3,[10][11][12] and edge computing. [13,14] The working principle of the TMOs-based memristive devices relies on ion drift or diffusion, which resembles motion of ions in the biological neurons and synapses. The ionic motions exhibit dynamic behaviors.…”