Materials 2021, 14, 5223. https://doi.org/10.3390/ma14185223 www.mdpi.com/journal/materials
Article
In-Memory-Computing Realization with a
Photodiode/Memristor Based Vision Sensor
Nikolaos Vasileiadis
1,2,
*, Vasileios Ntinas
2
, Georgios Ch. Sirakoulis
2
and Panagiotis Dimitrakis
1,
*
1
Institute of Nanoscience and Nanotechnology, National Center of Scientific Research “Demokritos”,
15341 Agia Paraskevi, Greece
2
Department of Electrical and Computer Engineering, Democritus University of Thrace (DUTh),
67100 Xanthi, Greece; vntinas@ee.duth.gr (V.N.); gsirak@ee.duth.gr (G.C.S.)
* Correspondence: n.vasiliadis@inn.demokritos.gr (N.V.); p.dimitrakis@inn.demokritos.gr (P.D.);
Tel.: +30-210-650-3118
Abstract: State-of-the-art IoT technologies request novel design solutions in edge computing, result-
ing in even more portable and energy-efficient hardware for in-the-field processing tasks. Vision
sensors, processors, and hardware accelerators are among the most demanding IoT applications.
Resistance switching (RS) two-terminal devices are suitable for resistive RAMs (RRAM), a promis-
ing technology to realize storage class memories. Furthermore, due to their memristive nature,
RRAMs are appropriate candidates for in-memory computing architectures. Recently, we demon-
strated a CMOS compatible silicon nitride (SiNx) MIS RS device with memristive properties. In this
paper, a report on a new photodiode-based vision sensor architecture with in-memory computing
capability, relying on memristive device, is disclosed. In this context, the resistance switching dy-
namics of our memristive device were measured and a data-fitted behavioral model was extracted.
SPICE simulations were made highlighting the in-memory computing capabilities of the proposed
photodiode-one memristor pixel vision sensor. Finally, an integration and manufacturing perspec-
tive was discussed.
Keywords: resistive random-access memory (RRAM); resistance switching; silicon nitride;
memristor; vision sensor; photodiode; crossbar; in-memory computing; edge computing; dot
product engine; IoT; SPICE
1. Introduction
During the last decade, it became apparent that created data are increasing rapidly,
requesting revolutionary solutions when memory and storage is concerned. These needs
are more demanding in case of Internet of Things (IoT) applications and the correspond-
ing IoT sensors that produce zettabytes of data nowadays. The most straightforward ap-
proach to tackle the uprising urgent issue is the local pre-processing of the unstructured
data generated by the IoT sensors in an edge-based sense [1–4]. Such a promising solution
will eventually minimize the requesting power consumption of the corresponding IoT
applications and at the same time advance the overall computing in terms of energy effi-
ciency. However, following conventional digital design approaches involving either spe-
cialized signal processors or more generic microcontrollers does not prove as promising
as expected and especially when power consumption is highly demanded [5]. The next
obvious step of utilizing a more specific-oriented CMOS-based design can be significantly
enhanced with the addition of novel nanoelectronic devices with memory abilities,
namely memristors, to be combined with the ΙοΤ sensors. To further investigate the prom-
ising aspects of such a hardware approach enabling also in-memory computation at IoT
sensors, special interest is on vision sensors as a fine candidate for edge computing. The
vision sensors, when integrated with such processing hardware, are enabled to provide
Citation: Vasileiadis, N.; Ntinas, V.;
Sirakoulis, G.C.; Dimitrakis, P.
In-Memory-Computing Realization
with a Photodiode/Memristor Based
Vision Sensor. Materials 2021, 14,
5223. https://doi.org/
10.3390/ma14185223
Academic Editor: Stephan Menzel
and Katarzyna Bejtka
Received: 28 June 2021
Accepted: 7 September 2021
Published: 10 September 2021
Publisher’s Note: MDPI stays neu-
tral with regard to jurisdictional
claims in published maps and institu-
tional affiliations.
Copyright: © 2021 by the authors. Li-
censee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and con-
ditions of the Creative Commons At-
tribution (CC BY) license (http://crea-
tivecommons.org/licenses/by/4.0/).