Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Merged acquisition-processing system based on a photoelectrical neural network
KTH, School of Engineering Sciences in Chemistry, Biotechnology and Health (CBH).ORCID iD: 0000-0002-1768-1071
Fudan University, SIST, State Key Laboratory of ASIC and System.
KTH, School of Electrical Engineering and Computer Science (EECS).
Fudan University, SIST, State Key Laboratory of ASIC and System.
Show others and affiliations
(English)Manuscript (preprint) (Other academic)
Abstract [en]

In brain-inspired computing, one of the most attractive fields of study currently, new issues are emerging despite its booming development. From the systems perspective, the performances of the existing systems are hindered by inadequate hardware support, particularly the unavoidable data acquisition and transmission between the sensor module and the data processing module. In this work, we break this bottleneck by proposing a photoelectrical neural network (PNN) that merges the new sensing function into the processing network. Benefitting from its high-parallel structure and minimized hardware consumption, a novel merged acquisition-processing (MAP) system with very high efficiency and very low cost has been developed. As the key component of the MAP system, a dual-mode photoelectrical synapse (DMPS) is carefully designed and developed. It has advantages in terms of both function and performance as compared to the existing artificial synapses, which make it the best candidate for the proposed system. This work will initiate an entirely new field of unconventional neuromorphic systems.

National Category
Other Electrical Engineering, Electronic Engineering, Information Engineering
Identifiers
URN: urn:nbn:se:kth:diva-235739OAI: oai:DiVA.org:kth-235739DiVA, id: diva2:1253030
Note

QC 20181004

Available from: 2018-10-03 Created: 2018-10-03 Last updated: 2018-10-10Bibliographically approved
In thesis
1. Flexible Electrical and Photoelectrical Artificial Synapses for Neuromorphic Systems
Open this publication in new window or tab >>Flexible Electrical and Photoelectrical Artificial Synapses for Neuromorphic Systems
2018 (English)Doctoral thesis, comprehensive summary (Other academic)
Abstract [en]

Over the past decade, the field of personal electronic systems has trended toward mobile and wearable devices. However, the capabilities of existing electronic systems are overwhelmed by the computing demands at the wearable sensing stage. Two main bottlenecks are encountered. The first bottleneck is located within the computing module, between the processing units and the memory, and is known as the von-Neumann bottleneck. The second bottleneck is located between the sensing module and the computing module of the system.

Inspired by neuromorphic computing, an architecture of the sensitive neuromorphic network (SNN) is developed as a candidate for overcoming both bottlenecks. Suitable building blocks, especially in flexible form, must be developed. In this work, starting from the demand analysis and followed by prototype development, performance optimization, and feasibility testing, two kinds of critical devices were developed for fabricating a photosensitive neuromorphic network (PSNN).

A high-performance flexible electrical artificial synapse that is based on the electron-trapping mechanism was developed. In addition to the basic memristive features, multiple kinds of synaptic plasticity were also demonstrated, which enriched the collection of possible applications. Furthermore, optimization on multiple performance metrics was easily performed using the intrinsic features and structure of the device.

A new photoelectrical artificial synapse was also realized by successfully combining light signal sensing and processing in a single synapse. A flexible dual-mode photoelectrical synapse, which fulfilled the requirements of the designed PSNN working protocol, was demonstrated. The device showed gate-tunable photomemristive features, thereby enabling its application as a photoelectrical artificial synapse.

Using the newly developed devices and the proposed network architecture, this work successfully initiated a new area of research, namely, the sensitive neuromorphic network, and provided a valid solution that addresses the current limitations of existing wearable electronic systems.

Place, publisher, year, edition, pages
KTH Royal Institute of Technology, 2018. p. 58
Series
TRITA-CBH-FOU ; 2018:46
Keywords
Flexible electronics, Neuromorphic network, Memristor, Electron trapping, Electrical artificial synapse, Photoelectrical artificial synapse
National Category
Other Electrical Engineering, Electronic Engineering, Information Engineering
Research subject
Applied Medical Technology
Identifiers
urn:nbn:se:kth:diva-235742 (URN)978-91-7729-979-0 (ISBN)
Public defence
2018-10-31, T2, Hälsovägen 11C, Huddinge, 09:00 (English)
Opponent
Supervisors
Note

QC 20181008

Available from: 2018-10-08 Created: 2018-10-08 Last updated: 2018-10-08Bibliographically approved

Open Access in DiVA

No full text in DiVA

Search in DiVA

By author/editor
Yang, KunlongHuan, YuxiangSeoane, Fernando
By organisation
School of Engineering Sciences in Chemistry, Biotechnology and Health (CBH)School of Electrical Engineering and Computer Science (EECS)
Other Electrical Engineering, Electronic Engineering, Information Engineering

Search outside of DiVA

GoogleGoogle Scholar

urn-nbn

Altmetric score

urn-nbn
Total: 55 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf