Bio-inspired Anomaly Detection for Low-cost Gas Sensors

Junxiu Liu, Jim Harkin, Liam McDaid, Shvan Karim, Alan G. Millard, James Hilder, Simon Hickinbotham, Anju P. Johnson, Jon Timmis, David M. Halliday, Andy M. Tyrrel

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

This paper proposes a novel anomaly detection method for gas sensors using spiking neural network principles. The synapse models with excitatory/inhibitory responses and a single spiking neuron are employed to develop the bio-inspired anomaly detector for a single gas sensor. The approach can detect anomalies in the data, which is collected by the gas sensor by identifying rapid changes rather than a magnitude threshold. In particular, the false-positive detections due to the drifts of low-cost sensors are minimised using the proposed bio-inspired approach. Using the chemicals of surgical spirits and isobutanol as test substances, experiments were carried out to evaluate the proposed method. Results demonstrate that gas anomalies can be detected when the chemical substances are presented to the sensor. In addition, results show that the approach can detect under the presence of sensor drift. The proposed bio-inspired detector was implemented on FPGA hardware, which demonstrates relatively low resources. Compact and energy efficient CMOS-based implementations of the synapse are also available which supports the low-cost potential applications of this approach, e.g. use in safety with drones and ground robots in hazardous scene detection.

Original languageEnglish
Title of host publication18th International Conference on Nanotechnology, NANO 2018
PublisherIEEE Computer Society
Volume2018-July
ISBN (Electronic)9781538653364
DOIs
Publication statusPublished - 28 Jan 2019
Externally publishedYes
Event18th International Conference on Nanotechnology - Cork, Ireland
Duration: 23 Jul 201826 Jul 2018
Conference number: 18
http://ieeenano18.org/

Conference

Conference18th International Conference on Nanotechnology
Abbreviated titleNANO 2018
CountryIreland
CityCork
Period23/07/1826/07/18
Internet address

Fingerprint

Chemical sensors
anomalies
sensors
Sensors
gases
Detectors
Costs
synapses
spiking
Neurons
Field programmable gate arrays (FPGA)
Gases
Robots
Neural networks
Hardware
detectors
neurons
robots
safety
CMOS

Cite this

Liu, J., Harkin, J., McDaid, L., Karim, S., Millard, A. G., Hilder, J., ... Tyrrel, A. M. (2019). Bio-inspired Anomaly Detection for Low-cost Gas Sensors. In 18th International Conference on Nanotechnology, NANO 2018 (Vol. 2018-July). [8626301] IEEE Computer Society. https://doi.org/10.1109/NANO.2018.8626301
Liu, Junxiu ; Harkin, Jim ; McDaid, Liam ; Karim, Shvan ; Millard, Alan G. ; Hilder, James ; Hickinbotham, Simon ; Johnson, Anju P. ; Timmis, Jon ; Halliday, David M. ; Tyrrel, Andy M. / Bio-inspired Anomaly Detection for Low-cost Gas Sensors. 18th International Conference on Nanotechnology, NANO 2018. Vol. 2018-July IEEE Computer Society, 2019.
@inproceedings{34aa6e33e1c545ef8abaf190847ead7d,
title = "Bio-inspired Anomaly Detection for Low-cost Gas Sensors",
abstract = "This paper proposes a novel anomaly detection method for gas sensors using spiking neural network principles. The synapse models with excitatory/inhibitory responses and a single spiking neuron are employed to develop the bio-inspired anomaly detector for a single gas sensor. The approach can detect anomalies in the data, which is collected by the gas sensor by identifying rapid changes rather than a magnitude threshold. In particular, the false-positive detections due to the drifts of low-cost sensors are minimised using the proposed bio-inspired approach. Using the chemicals of surgical spirits and isobutanol as test substances, experiments were carried out to evaluate the proposed method. Results demonstrate that gas anomalies can be detected when the chemical substances are presented to the sensor. In addition, results show that the approach can detect under the presence of sensor drift. The proposed bio-inspired detector was implemented on FPGA hardware, which demonstrates relatively low resources. Compact and energy efficient CMOS-based implementations of the synapse are also available which supports the low-cost potential applications of this approach, e.g. use in safety with drones and ground robots in hazardous scene detection.",
author = "Junxiu Liu and Jim Harkin and Liam McDaid and Shvan Karim and Millard, {Alan G.} and James Hilder and Simon Hickinbotham and Johnson, {Anju P.} and Jon Timmis and Halliday, {David M.} and Tyrrel, {Andy M.}",
year = "2019",
month = "1",
day = "28",
doi = "10.1109/NANO.2018.8626301",
language = "English",
volume = "2018-July",
booktitle = "18th International Conference on Nanotechnology, NANO 2018",
publisher = "IEEE Computer Society",
address = "United States",

}

Liu, J, Harkin, J, McDaid, L, Karim, S, Millard, AG, Hilder, J, Hickinbotham, S, Johnson, AP, Timmis, J, Halliday, DM & Tyrrel, AM 2019, Bio-inspired Anomaly Detection for Low-cost Gas Sensors. in 18th International Conference on Nanotechnology, NANO 2018. vol. 2018-July, 8626301, IEEE Computer Society, 18th International Conference on Nanotechnology, Cork, Ireland, 23/07/18. https://doi.org/10.1109/NANO.2018.8626301

Bio-inspired Anomaly Detection for Low-cost Gas Sensors. / Liu, Junxiu; Harkin, Jim; McDaid, Liam; Karim, Shvan; Millard, Alan G.; Hilder, James; Hickinbotham, Simon; Johnson, Anju P.; Timmis, Jon; Halliday, David M.; Tyrrel, Andy M.

18th International Conference on Nanotechnology, NANO 2018. Vol. 2018-July IEEE Computer Society, 2019. 8626301.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Bio-inspired Anomaly Detection for Low-cost Gas Sensors

AU - Liu, Junxiu

AU - Harkin, Jim

AU - McDaid, Liam

AU - Karim, Shvan

AU - Millard, Alan G.

AU - Hilder, James

AU - Hickinbotham, Simon

AU - Johnson, Anju P.

AU - Timmis, Jon

AU - Halliday, David M.

AU - Tyrrel, Andy M.

PY - 2019/1/28

Y1 - 2019/1/28

N2 - This paper proposes a novel anomaly detection method for gas sensors using spiking neural network principles. The synapse models with excitatory/inhibitory responses and a single spiking neuron are employed to develop the bio-inspired anomaly detector for a single gas sensor. The approach can detect anomalies in the data, which is collected by the gas sensor by identifying rapid changes rather than a magnitude threshold. In particular, the false-positive detections due to the drifts of low-cost sensors are minimised using the proposed bio-inspired approach. Using the chemicals of surgical spirits and isobutanol as test substances, experiments were carried out to evaluate the proposed method. Results demonstrate that gas anomalies can be detected when the chemical substances are presented to the sensor. In addition, results show that the approach can detect under the presence of sensor drift. The proposed bio-inspired detector was implemented on FPGA hardware, which demonstrates relatively low resources. Compact and energy efficient CMOS-based implementations of the synapse are also available which supports the low-cost potential applications of this approach, e.g. use in safety with drones and ground robots in hazardous scene detection.

AB - This paper proposes a novel anomaly detection method for gas sensors using spiking neural network principles. The synapse models with excitatory/inhibitory responses and a single spiking neuron are employed to develop the bio-inspired anomaly detector for a single gas sensor. The approach can detect anomalies in the data, which is collected by the gas sensor by identifying rapid changes rather than a magnitude threshold. In particular, the false-positive detections due to the drifts of low-cost sensors are minimised using the proposed bio-inspired approach. Using the chemicals of surgical spirits and isobutanol as test substances, experiments were carried out to evaluate the proposed method. Results demonstrate that gas anomalies can be detected when the chemical substances are presented to the sensor. In addition, results show that the approach can detect under the presence of sensor drift. The proposed bio-inspired detector was implemented on FPGA hardware, which demonstrates relatively low resources. Compact and energy efficient CMOS-based implementations of the synapse are also available which supports the low-cost potential applications of this approach, e.g. use in safety with drones and ground robots in hazardous scene detection.

UR - http://www.scopus.com/inward/record.url?scp=85062266036&partnerID=8YFLogxK

U2 - 10.1109/NANO.2018.8626301

DO - 10.1109/NANO.2018.8626301

M3 - Conference contribution

VL - 2018-July

BT - 18th International Conference on Nanotechnology, NANO 2018

PB - IEEE Computer Society

ER -

Liu J, Harkin J, McDaid L, Karim S, Millard AG, Hilder J et al. Bio-inspired Anomaly Detection for Low-cost Gas Sensors. In 18th International Conference on Nanotechnology, NANO 2018. Vol. 2018-July. IEEE Computer Society. 2019. 8626301 https://doi.org/10.1109/NANO.2018.8626301