FPGA-Based Fault-Injection and Data Acquisition of Self-Repairing Spiking Neural Network Hardware

Shvan Haji Karim, Jim Harkin, LJ McDaid, Bryan Gardiner, Junxiu Liu, David M. Halliday, Andy M. Tyrrell, Jon Timmis, Alan G. Millard, Anju Johnson

Research output: Contribution to conferencePaper

1 Citation (Scopus)

Abstract

Spiking Astrocyte-neuron Networks (SANNs) model the adaptive/repair feature of the human brain. They integrate astrocyte cells with spiking neurons to facilitate a distributed and fine-grained self-repair capability at the synapse level. SANNs are more complex with the addition of astrocyte cells and require longer simulation times, as they are dynamic over much longer time-scales than traditional neural networks. Therefore, dedicated FPGA accelerators offer reductions in simulation times. To support the acceleration of SANNs, the capability of fault injection to synapses and monitoring significant levels of neuron and astrocyte data for off-chip transmission to PC-based analysis, are required. This paper presents an FPGA-based monitoring platform (FMP) for injecting faults and capturing and analyzing data acquired from the SANN FPGA accelerator, Astrobyte. The FMP uses custom logic and a NIOS II based system to control fault injection and data monitoring on the FPGA. Results show accurate accelerated simulations of fault injection scenarios using FMP with speedups up to 65 times greater compared with equivalent Matlab implementations.
LanguageEnglish
Pages1-5
Number of pages5
DOIs
Publication statusPublished - 27 May 2018
EventIEEE International Symposium on Circuits and Systems - Florence, Italy
Duration: 27 May 201830 May 2018

Conference

ConferenceIEEE International Symposium on Circuits and Systems
Abbreviated titleISCAS
CountryItaly
CityFlorence
Period27/05/1830/05/18

Fingerprint

Field programmable gate arrays (FPGA)
Data acquisition
Neurons
Neural networks
Hardware
Monitoring
Particle accelerators
Repair
Astrocytes
Brain

Keywords

  • FPGA acceleration
  • Astrocytes
  • Data Acquisition
  • Spiking neural network
  • Self repair
  • Fault injection

Cite this

Haji Karim, S., Harkin, J., McDaid, LJ., Gardiner, B., Liu, J., Halliday, D. M., ... Johnson, A. (2018). FPGA-Based Fault-Injection and Data Acquisition of Self-Repairing Spiking Neural Network Hardware. 1-5. Paper presented at IEEE International Symposium on Circuits and Systems, Florence, Italy. https://doi.org/10.1109/ISCAS.2018.8351512
Haji Karim, Shvan ; Harkin, Jim ; McDaid, LJ ; Gardiner, Bryan ; Liu, Junxiu ; Halliday, David M. ; Tyrrell, Andy M. ; Timmis, Jon ; Millard, Alan G. ; Johnson, Anju. / FPGA-Based Fault-Injection and Data Acquisition of Self-Repairing Spiking Neural Network Hardware. Paper presented at IEEE International Symposium on Circuits and Systems, Florence, Italy.5 p.
@conference{f6d2ef6b2cb44a22bd3eba4ee45a81da,
title = "FPGA-Based Fault-Injection and Data Acquisition of Self-Repairing Spiking Neural Network Hardware",
abstract = "Spiking Astrocyte-neuron Networks (SANNs) model the adaptive/repair feature of the human brain. They integrate astrocyte cells with spiking neurons to facilitate a distributed and fine-grained self-repair capability at the synapse level. SANNs are more complex with the addition of astrocyte cells and require longer simulation times, as they are dynamic over much longer time-scales than traditional neural networks. Therefore, dedicated FPGA accelerators offer reductions in simulation times. To support the acceleration of SANNs, the capability of fault injection to synapses and monitoring significant levels of neuron and astrocyte data for off-chip transmission to PC-based analysis, are required. This paper presents an FPGA-based monitoring platform (FMP) for injecting faults and capturing and analyzing data acquired from the SANN FPGA accelerator, Astrobyte. The FMP uses custom logic and a NIOS II based system to control fault injection and data monitoring on the FPGA. Results show accurate accelerated simulations of fault injection scenarios using FMP with speedups up to 65 times greater compared with equivalent Matlab implementations.",
keywords = "FPGA acceleration, Astrocytes, Data Acquisition, Spiking neural network, Self repair, Fault injection",
author = "{Haji Karim}, Shvan and Jim Harkin and LJ McDaid and Bryan Gardiner and Junxiu Liu and Halliday, {David M.} and Tyrrell, {Andy M.} and Jon Timmis and Millard, {Alan G.} and Anju Johnson",
year = "2018",
month = "5",
day = "27",
doi = "10.1109/ISCAS.2018.8351512",
language = "English",
pages = "1--5",
note = "IEEE International Symposium on Circuits and Systems, ISCAS ; Conference date: 27-05-2018 Through 30-05-2018",

}

Haji Karim, S, Harkin, J, McDaid, LJ, Gardiner, B, Liu, J, Halliday, DM, Tyrrell, AM, Timmis, J, Millard, AG & Johnson, A 2018, 'FPGA-Based Fault-Injection and Data Acquisition of Self-Repairing Spiking Neural Network Hardware' Paper presented at IEEE International Symposium on Circuits and Systems, Florence, Italy, 27/05/18 - 30/05/18, pp. 1-5. https://doi.org/10.1109/ISCAS.2018.8351512

FPGA-Based Fault-Injection and Data Acquisition of Self-Repairing Spiking Neural Network Hardware. / Haji Karim, Shvan; Harkin, Jim; McDaid, LJ; Gardiner, Bryan; Liu, Junxiu; Halliday, David M.; Tyrrell, Andy M.; Timmis, Jon; Millard, Alan G.; Johnson, Anju.

2018. 1-5 Paper presented at IEEE International Symposium on Circuits and Systems, Florence, Italy.

Research output: Contribution to conferencePaper

TY - CONF

T1 - FPGA-Based Fault-Injection and Data Acquisition of Self-Repairing Spiking Neural Network Hardware

AU - Haji Karim, Shvan

AU - Harkin, Jim

AU - McDaid, LJ

AU - Gardiner, Bryan

AU - Liu, Junxiu

AU - Halliday, David M.

AU - Tyrrell, Andy M.

AU - Timmis, Jon

AU - Millard, Alan G.

AU - Johnson, Anju

PY - 2018/5/27

Y1 - 2018/5/27

N2 - Spiking Astrocyte-neuron Networks (SANNs) model the adaptive/repair feature of the human brain. They integrate astrocyte cells with spiking neurons to facilitate a distributed and fine-grained self-repair capability at the synapse level. SANNs are more complex with the addition of astrocyte cells and require longer simulation times, as they are dynamic over much longer time-scales than traditional neural networks. Therefore, dedicated FPGA accelerators offer reductions in simulation times. To support the acceleration of SANNs, the capability of fault injection to synapses and monitoring significant levels of neuron and astrocyte data for off-chip transmission to PC-based analysis, are required. This paper presents an FPGA-based monitoring platform (FMP) for injecting faults and capturing and analyzing data acquired from the SANN FPGA accelerator, Astrobyte. The FMP uses custom logic and a NIOS II based system to control fault injection and data monitoring on the FPGA. Results show accurate accelerated simulations of fault injection scenarios using FMP with speedups up to 65 times greater compared with equivalent Matlab implementations.

AB - Spiking Astrocyte-neuron Networks (SANNs) model the adaptive/repair feature of the human brain. They integrate astrocyte cells with spiking neurons to facilitate a distributed and fine-grained self-repair capability at the synapse level. SANNs are more complex with the addition of astrocyte cells and require longer simulation times, as they are dynamic over much longer time-scales than traditional neural networks. Therefore, dedicated FPGA accelerators offer reductions in simulation times. To support the acceleration of SANNs, the capability of fault injection to synapses and monitoring significant levels of neuron and astrocyte data for off-chip transmission to PC-based analysis, are required. This paper presents an FPGA-based monitoring platform (FMP) for injecting faults and capturing and analyzing data acquired from the SANN FPGA accelerator, Astrobyte. The FMP uses custom logic and a NIOS II based system to control fault injection and data monitoring on the FPGA. Results show accurate accelerated simulations of fault injection scenarios using FMP with speedups up to 65 times greater compared with equivalent Matlab implementations.

KW - FPGA acceleration

KW - Astrocytes

KW - Data Acquisition

KW - Spiking neural network

KW - Self repair

KW - Fault injection

U2 - 10.1109/ISCAS.2018.8351512

DO - 10.1109/ISCAS.2018.8351512

M3 - Paper

SP - 1

EP - 5

ER -