{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,20]],"date-time":"2026-10-20T11:20:07Z","timestamp":1792495207959,"version":"4.5.0"},"reference-count":33,"publisher":"MDPI AG","issue":"11","license":[{"start":{"date-parts":[[2026,6,1]],"date-time":"2026-06-01T00:00:00Z","timestamp":1780272000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Molecules"],"abstract":"<jats:p>Resource-constrained edge processors deployed on unmanned aerial vehicles and wearable platforms require compact, drift-robust gas classification models for a range of environmental and security monitoring applications, including CBRN-motivated scenarios. Existing approaches rely on server-grade architectures incompatible with edge-board-scale deployment, or on classifiers that chemically degrade severely under long-term sensor drift. Each UCI gas class was mapped to a CBRN behavioral category based on physicochemical analogy (molecular functional group, vapor pressure, and metal-oxide semiconductor (MOS) cross-sensitivity pattern), following established precedent. Analyzed were Ammonia (NH3), Acetaldehyde (CH3CHO), Acetone ((CH3)2CO), Ethylene (C2H4), Ethanol (C2H5OH), Toluene (C6H5CH3). We propose herein an end-to-end pipeline integrating a novel 1-D convolutional neural network with depth-wise separable convolutions (LiteSensor-Net), INT8 post-training quantization, structured magnitude pruning, and a knowledge-distillation domain-adaptation module (KD\u2013DM) for sensor drift compensation. Using the UCI Gas Sensor Array Drift Dataset (13,910 measurements; 16 metal-oxide sensors; six analyte gases; a 36-month work span). LiteSensor-Net achieved accuracy = 92.63 \u00b1 2.02%, macro-F1 = 0.898, model size = 5.99 kB INT8 pruned, inference latency = 6.3 ms, RAM footprint = 31.7 kB, and energy per inference = 0.04 mJ (all metrics on Raspberry Pi 4B, ARM Cortex-A72). Under chronological forward-chaining evaluation, KD\u2013DM\u201320 achieved 47.91 \u00b1 18.79% mean accuracy over Batches 2\u201310, representing a +9.25 pp improvement over uncompensated NC (38.66%). A six-metric benchmark framework\u2014accuracy, macro-F1, model size, inference latency, RAM footprint, and energy per inference\u2014is introduced to standardize edge-AI gas classifier evaluation. The proposed pipeline provides an open-source, deployable foundation for edge-class gas classification systems, with CBRN detection as a motivating application. Full operational validation on certified chemical simulants remains as future work.<\/jats:p>","DOI":"10.3390\/molecules31111884","type":"journal-article","created":{"date-parts":[[2026,6,1]],"date-time":"2026-06-01T12:19:42Z","timestamp":1780316382000},"page":"1884","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":3,"title":["Drift-Robust Lightweight Deep Learning on Open Gas Sensor Benchmarks: A Reproducible Architecture Study with CBRN Applicability Mapping"],"prefix":"10.3390","volume":"31","author":[{"ORCID":"https:\/\/orcid.org\/0009-0004-7822-4861","authenticated-orcid":false,"given":"Soohwan","family":"Kim","sequence":"first","affiliation":[{"name":"CBRN Defense Research Institute, Seoul 06796, Republic of Korea"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Myeongsik","family":"Shin","sequence":"additional","affiliation":[{"name":"CBRN Defense Research Institute, Seoul 06796, Republic of Korea"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-8739-218X","authenticated-orcid":false,"given":"Ku","family":"Kang","sequence":"additional","affiliation":[{"name":"CBRN Defense Research Institute, Seoul 06796, Republic of Korea"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-5116-2064","authenticated-orcid":false,"given":"Doo-Hee","family":"Lee","sequence":"additional","affiliation":[{"name":"CBRN Defense Research Institute, Seoul 06796, Republic of Korea"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"David G.","family":"Churchill","sequence":"additional","affiliation":[{"name":"Department of Chemistry, Korea Advanced Institute of Science and Technology (KAIST), Daejeon 34141, Republic of Korea"},{"name":"Therapeutic Bioengineering Section, KAIST Institute for Health Science and Technology (KIHST), Daejeon 34141, Republic of Korea"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yoon Jeong","family":"Jang","sequence":"additional","affiliation":[{"name":"CBRN Defense Research Institute, Seoul 06796, Republic of Korea"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1968","published-online":{"date-parts":[[2026,6,1]]},"reference":[{"key":"ref_1","doi-asserted-by":"crossref","unstructured":"Gawlik-Kobyli\u0144ska, M., Gudzbeler, G., Szklarski, \u0141., Kopp, N., Koch-Eschweiler, H., and Urban, M. (2021). The EU\u2013SENSE System for Chemical Hazards Detection, Identification, and Monitoring. Appl. Sci., 11.","DOI":"10.3390\/app112110308"},{"key":"ref_2","unstructured":"NATO Science & Technology Organization (2020). Science & Technology Trends 2020\u20132040: Exploring the S&T Edge, NATO Science & Technology Organization. Available online: https:\/\/www.sto.nato.int."},{"key":"ref_3","doi-asserted-by":"crossref","first-page":"177","DOI":"10.1080\/0163660X.2020.1770969","article-title":"Emerging Technologies and the Future of CBRN Terrorism","volume":"43","author":"Koblentz","year":"2020","journal-title":"Wash. Q."},{"key":"ref_4","doi-asserted-by":"crossref","unstructured":"Noh, D., and Oh, E. (2025). Chemical Detection Using Mobile Platforms and AI\u2013Based Data Processing Technologies. J. Sens. Actuator Netw., 14.","DOI":"10.3390\/jsan14010006"},{"key":"ref_5","doi-asserted-by":"crossref","unstructured":"Raza, W., Osman, A., Ferrini, F., and De Natale, F. (2021). Energy-Efficient Inference on the Edge Exploiting TinyML Capabilities for UAVs. Drones, 5.","DOI":"10.3390\/drones5040127"},{"key":"ref_6","doi-asserted-by":"crossref","first-page":"269","DOI":"10.1007\/s40820-024-01489-z","article-title":"Overcoming the Limits of Cross\u2013Sensitivity: Pattern Recognition Methods for Chemiresistive Gas Sensor Array","volume":"16","author":"Mei","year":"2024","journal-title":"Nano-Micro Lett."},{"key":"ref_7","doi-asserted-by":"crossref","first-page":"4285","DOI":"10.1039\/D4TC05220J","article-title":"Advances in Metal Oxide Semiconductor Gas Sensor Arrays Based on Machine Learning Algorithms","volume":"13","author":"Han","year":"2025","journal-title":"J. Mater. Chem. C"},{"key":"ref_8","doi-asserted-by":"crossref","first-page":"12969","DOI":"10.1109\/JSEN.2021.3076412","article-title":"Artificial Olfaction in the 21st Century","volume":"21","author":"Covington","year":"2021","journal-title":"IEEE Sens. J."},{"key":"ref_9","doi-asserted-by":"crossref","first-page":"96892","DOI":"10.1109\/ACCESS.2023.3294111","article-title":"A Comprehensive Survey on TinyML","volume":"11","author":"Abadade","year":"2023","journal-title":"IEEE Access"},{"key":"ref_10","doi-asserted-by":"crossref","first-page":"1595","DOI":"10.1016\/j.jksuci.2021.11.019","article-title":"A Review on TinyML: State\u2013of\u2013the\u2013Art and Prospects","volume":"34","author":"Ray","year":"2022","journal-title":"J. King Saud Univ. Comput. Inf. Sci."},{"key":"ref_11","doi-asserted-by":"crossref","unstructured":"Yaqoob, U., and Younis, M.I. (2021). Chemical Gas Sensors: Recent Developments, Challenges, and the Potential of Machine Learning\u2014A Review. Sensors, 21.","DOI":"10.3390\/s21082877"},{"key":"ref_12","doi-asserted-by":"crossref","first-page":"320","DOI":"10.1016\/j.snb.2012.01.074","article-title":"Chemical gas sensor drift compensation using classifier ensembles","volume":"166\u2013167","author":"Vergara","year":"2012","journal-title":"Sens. Actuators B Chem."},{"key":"ref_13","doi-asserted-by":"crossref","first-page":"131668","DOI":"10.1016\/j.snb.2022.131668","article-title":"Drift in a popular metal oxide sensor dataset reveals limitations for gas classification benchmarks","volume":"361","author":"Dennler","year":"2022","journal-title":"Sens. Actuators B Chem."},{"key":"ref_14","doi-asserted-by":"crossref","first-page":"26383","DOI":"10.1109\/ACCESS.2023.3257414","article-title":"A Study of Drift Effect in a Popular Metal Oxide Sensor and Gas Recognition Using Public Gas Datasets","volume":"11","author":"Chang","year":"2023","journal-title":"IEEE Access"},{"key":"ref_15","unstructured":"Howard, A.G., Sandler, M., Chu, G., Chen, L.C., Chen, B., Tan, M., Wang, W., Zhu, Y., Pang, R., and Vasudevan, V. (November, January 27). Searching for MobileNetV3. Proceedings of the IEEE\/CVF International Conference on Computer Vision (ICCV), Seoul, Republic of Korea."},{"key":"ref_16","unstructured":"TensorFlow Team (2026, May 24). TensorFlow Lite: Model Optimization Guide. Available online: https:\/\/www.tensorflow.org\/lite\/performance\/model_optimization."},{"key":"ref_17","doi-asserted-by":"crossref","unstructured":"Lin, J., and Zhan, X. (2026). Sensor\u2013Drift Compensation in Electronic\u2013Nose\u2013Based Gas Recognition Using Knowledge Distillation. Informatics, 13.","DOI":"10.3390\/informatics13010015"},{"key":"ref_18","doi-asserted-by":"crossref","first-page":"6564","DOI":"10.1109\/JSEN.2020.3041949","article-title":"Multi\u2013Classifier Tree with 327 Transient Features for Drift Compensation in Electronic Nose","volume":"21","author":"Rehman","year":"2021","journal-title":"IEEE Sens. J."},{"key":"ref_19","doi-asserted-by":"crossref","first-page":"2422","DOI":"10.1002\/qre.3354","article-title":"Sensor Drift Compensation for Gas Mixture Classification in Batch Experiments","volume":"39","author":"Oh","year":"2023","journal-title":"Qual. Reliab. Eng. Int."},{"key":"ref_20","unstructured":"Howard, A.G., Zhu, M., Chen, B., Kalenichenko, D., Wang, W., Weyand, T., Andreetto, M., and Adam, H. (2017). MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications. arXiv."},{"key":"ref_21","doi-asserted-by":"crossref","unstructured":"Li, J., Xie, G., Dai, L., Yang, M., and Su, Y. (2025). Detection and Pattern Recognition of Chemical Warfare Agents by MOS\u2013Based MEMS Gas Sensor Array. Sensors, 25.","DOI":"10.3390\/s25082633"},{"key":"ref_22","doi-asserted-by":"crossref","unstructured":"Witkiewicz, Z., Jasek, K., and Grabka, M. (2023). Semiconductor Gas Sensors for Detecting Chemical Warfare Agents and Their Simulants. Sensors, 23.","DOI":"10.3390\/s23063272"},{"key":"ref_23","unstructured":"Vergara, A. (2012). Gas Sensor Array Drift Dataset. UCI Machine Learning Repository, Center for Machine Learning and Intelligent Systems."},{"key":"ref_24","first-page":"1","article-title":"Decontamination Solutions and Techniques for Chemical, Biological, Radiological and Nuclear (CBRN) Incidents","volume":"20","author":"Karakurt","year":"2024","journal-title":"J. Def. Sci."},{"key":"ref_25","doi-asserted-by":"crossref","unstructured":"Sandler, M., Howard, A., Zhu, M., Zhmoginov, A., and Chen, L.C. (2018, January 18\u201323). MobileNetV2: Inverted Residuals and Linear Bottlenecks. Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Salt Lake City, UT, USA.","DOI":"10.1109\/CVPR.2018.00474"},{"key":"ref_26","unstructured":"Zhu, M.H., and Gupta, S. (2018). To Prune, or Not to Prune: Exploring the Efficacy of Pruning for Model Compression. arXiv."},{"key":"ref_27","unstructured":"Hinton, G., Vinyals, O., and Dean, J. (2015). Distilling the Knowledge in a Neural Network. arXiv."},{"key":"ref_28","unstructured":"Courty, B., Schmidt, V., Coutarel, M., Feld, B., Lecourt, J., Connell, L., Amine, S. (2024). mlco2\/Codecarbon: V2.4.1, Zenodo."},{"key":"ref_29","doi-asserted-by":"crossref","first-page":"2906","DOI":"10.1021\/acssensors.4c03655","article-title":"Gas Sensor Drift Compensation Using Semi\u2013Supervised Ensemble Classifiers with Multi\u2013Level Features and Center Loss","volume":"10","author":"Jiang","year":"2025","journal-title":"ACS Sens."},{"key":"ref_30","doi-asserted-by":"crossref","unstructured":"Reddi, V.J., Cheng, C., Kanter, D., Mattson, P., Schmuelling, G., Wu, C.-J., Anderson, B., Breughe, M., Charlebois, M., and Chou, W. (2020, January 30). MLPerf Inference Benchmark. Proceedings of the ACM\/IEEE 47th Annual International Symposium on Computer Architecture (ISCA), Valencia, Spain.","DOI":"10.1109\/ISCA45697.2020.00045"},{"key":"ref_31","unstructured":"Banbury, C., Reddi, V.J., Torelli, P., Holleman, J., Jeffries, N., Kir\u00e1ly, C., Montino, P., Kanter, D., Ahmed, S.A., and Pau, D. (2021). MLPerf Tiny Benchmark. arXiv."},{"key":"ref_32","doi-asserted-by":"crossref","first-page":"494","DOI":"10.1561\/2200000101","article-title":"A Gentle Introduction to Conformal Prediction and Distribution-Free Uncertainty Quantification","volume":"16","author":"Angelopoulos","year":"2023","journal-title":"Found. Trends Mach. Learn."},{"key":"ref_33","doi-asserted-by":"crossref","first-page":"1628","DOI":"10.1038\/s41597-025-05993-8","article-title":"Long\u2013term drift behavior in metal oxide gas sensor arrays: A one\u2013year dataset from an electronic nose","volume":"12","author":"Eimler","year":"2025","journal-title":"Sci. Data"}],"container-title":["Molecules"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/1420-3049\/31\/11\/1884\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,6,1]],"date-time":"2026-06-01T12:20:15Z","timestamp":1780316415000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/1420-3049\/31\/11\/1884"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,6,1]]},"references-count":33,"journal-issue":{"issue":"11","published-online":{"date-parts":[[2026,6]]}},"alternative-id":["molecules31111884"],"URL":"https:\/\/doi.org\/10.3390\/molecules31111884","relation":{},"ISSN":["1420-3049"],"issn-type":[{"value":"1420-3049","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,6,1]]}}}