TinyML-Based Real-Time Drift Compensation for Gas Sensors Using Spectral–Temporal Neural Networks

The implementation of low-cost sensitive and selective gas sensors for monitoring fruit ripening and quality strongly depends on their long-term stability. Gas sensor drift undermines the long-term reliability of low-cost sensing platforms, particularly in precision agriculture. We present a real-ti...

Full description

Saved in:
Bibliographic Details
Main Authors: Adir Krayden, M. Avraham, H. Ashkar, T. Blank, S. Stolyarova, Yael Nemirovsky
Format: Article
Language:English
Published: MDPI AG 2025-06-01
Series:Chemosensors
Subjects:
Online Access:https://www.mdpi.com/2227-9040/13/7/223
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849733590154215424
author Adir Krayden
M. Avraham
H. Ashkar
T. Blank
S. Stolyarova
Yael Nemirovsky
author_facet Adir Krayden
M. Avraham
H. Ashkar
T. Blank
S. Stolyarova
Yael Nemirovsky
author_sort Adir Krayden
collection DOAJ
description The implementation of low-cost sensitive and selective gas sensors for monitoring fruit ripening and quality strongly depends on their long-term stability. Gas sensor drift undermines the long-term reliability of low-cost sensing platforms, particularly in precision agriculture. We present a real-time drift compensation framework based on a lightweight Temporal Convolutional Neural Network (TCNN) combined with a Hadamard spectral transform. The model operates causally on incoming sensor data, achieving a mean absolute error below 1 mV on long-term recordings (equivalent to <1 particle per million (ppm) gas concentration). Through quantization, we compress the model by over 70%, without sacrificing accuracy. Demonstrated on a combustion-type gas sensor system (dubbed GMOS) for ethylene monitoring, our approach enables continuous, drift-corrected operation without the need for recalibration or dependence on cloud-based services, offering a generalizable solution for embedded environmental sensing—in food transportation containers, cold storage facilities, de-greening rooms and directly in the field.
format Article
id doaj-art-6785156e2bf64ac7800b23451c5b31d3
institution DOAJ
issn 2227-9040
language English
publishDate 2025-06-01
publisher MDPI AG
record_format Article
series Chemosensors
spelling doaj-art-6785156e2bf64ac7800b23451c5b31d32025-08-20T03:08:00ZengMDPI AGChemosensors2227-90402025-06-0113722310.3390/chemosensors13070223TinyML-Based Real-Time Drift Compensation for Gas Sensors Using Spectral–Temporal Neural NetworksAdir Krayden0M. Avraham1H. Ashkar2T. Blank3S. Stolyarova4Yael Nemirovsky5Electrical and Computer Engineering Department, Technion—Israel Institute of Technology, Haifa 3200003, IsraelElectrical and Computer Engineering Department, Technion—Israel Institute of Technology, Haifa 3200003, IsraelElectrical and Computer Engineering Department, Technion—Israel Institute of Technology, Haifa 3200003, IsraelElectrical and Computer Engineering Department, Technion—Israel Institute of Technology, Haifa 3200003, IsraelTodos Technologies, Kinneret 12 Street, Airport City 7019900, IsraelElectrical and Computer Engineering Department, Technion—Israel Institute of Technology, Haifa 3200003, IsraelThe implementation of low-cost sensitive and selective gas sensors for monitoring fruit ripening and quality strongly depends on their long-term stability. Gas sensor drift undermines the long-term reliability of low-cost sensing platforms, particularly in precision agriculture. We present a real-time drift compensation framework based on a lightweight Temporal Convolutional Neural Network (TCNN) combined with a Hadamard spectral transform. The model operates causally on incoming sensor data, achieving a mean absolute error below 1 mV on long-term recordings (equivalent to <1 particle per million (ppm) gas concentration). Through quantization, we compress the model by over 70%, without sacrificing accuracy. Demonstrated on a combustion-type gas sensor system (dubbed GMOS) for ethylene monitoring, our approach enables continuous, drift-corrected operation without the need for recalibration or dependence on cloud-based services, offering a generalizable solution for embedded environmental sensing—in food transportation containers, cold storage facilities, de-greening rooms and directly in the field.https://www.mdpi.com/2227-9040/13/7/223gas sensor driftTinyMLconvolutional neural networksspectral filteringhadamard transformreal-time signal processing
spellingShingle Adir Krayden
M. Avraham
H. Ashkar
T. Blank
S. Stolyarova
Yael Nemirovsky
TinyML-Based Real-Time Drift Compensation for Gas Sensors Using Spectral–Temporal Neural Networks
Chemosensors
gas sensor drift
TinyML
convolutional neural networks
spectral filtering
hadamard transform
real-time signal processing
title TinyML-Based Real-Time Drift Compensation for Gas Sensors Using Spectral–Temporal Neural Networks
title_full TinyML-Based Real-Time Drift Compensation for Gas Sensors Using Spectral–Temporal Neural Networks
title_fullStr TinyML-Based Real-Time Drift Compensation for Gas Sensors Using Spectral–Temporal Neural Networks
title_full_unstemmed TinyML-Based Real-Time Drift Compensation for Gas Sensors Using Spectral–Temporal Neural Networks
title_short TinyML-Based Real-Time Drift Compensation for Gas Sensors Using Spectral–Temporal Neural Networks
title_sort tinyml based real time drift compensation for gas sensors using spectral temporal neural networks
topic gas sensor drift
TinyML
convolutional neural networks
spectral filtering
hadamard transform
real-time signal processing
url https://www.mdpi.com/2227-9040/13/7/223
work_keys_str_mv AT adirkrayden tinymlbasedrealtimedriftcompensationforgassensorsusingspectraltemporalneuralnetworks
AT mavraham tinymlbasedrealtimedriftcompensationforgassensorsusingspectraltemporalneuralnetworks
AT hashkar tinymlbasedrealtimedriftcompensationforgassensorsusingspectraltemporalneuralnetworks
AT tblank tinymlbasedrealtimedriftcompensationforgassensorsusingspectraltemporalneuralnetworks
AT sstolyarova tinymlbasedrealtimedriftcompensationforgassensorsusingspectraltemporalneuralnetworks
AT yaelnemirovsky tinymlbasedrealtimedriftcompensationforgassensorsusingspectraltemporalneuralnetworks