{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,4,5]],"date-time":"2026-04-05T00:47:22Z","timestamp":1775350042265,"version":"3.50.1"},"reference-count":48,"publisher":"MDPI AG","issue":"4","license":[{"start":{"date-parts":[[2021,2,4]],"date-time":"2021-02-04T00:00:00Z","timestamp":1612396800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/501100014538","name":"Lembaga Pengelola Dana Pendidikan","doi-asserted-by":"publisher","award":["Ref: S-1027\/LPDP.4\/2019"],"award-info":[{"award-number":["Ref: S-1027\/LPDP.4\/2019"]}],"id":[{"id":"10.13039\/501100014538","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Sensors"],"abstract":"<jats:p>Accurate air quality monitoring requires processing of multi-dimensional, multi-location sensor data, which has previously been considered in centralised machine learning models. These are often unsuitable for resource-constrained edge devices. In this article, we address this challenge by: (1) designing a novel hybrid deep learning model for hourly PM2.5 pollutant prediction; (2) optimising the obtained model for edge devices; and (3) examining model performance running on the edge devices in terms of both accuracy and latency. The hybrid deep learning model in this work comprises a 1D Convolutional Neural Network (CNN) and a Long Short-Term Memory (LSTM) to predict hourly PM2.5 concentration. The results show that our proposed model outperforms other deep learning models, evaluated by calculating RMSE and MAE errors. The proposed model was optimised for edge devices, the Raspberry Pi 3 Model B+ (RPi3B+) and Raspberry Pi 4 Model B (RPi4B). This optimised model reduced file size to a quarter of the original, with further size reduction achieved by implementing different post-training quantisation. In total, 8272 hourly samples were continuously fed to the edge device, with the RPi4B executing the model twice as fast as the RPi3B+ in all quantisation modes. Full-integer quantisation produced the lowest execution time, with latencies of 2.19 s and 4.73 s for RPi4B and RPi3B+, respectively.<\/jats:p>","DOI":"10.3390\/s21041064","type":"journal-article","created":{"date-parts":[[2021,2,4]],"date-time":"2021-02-04T21:29:27Z","timestamp":1612474167000},"page":"1064","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":49,"title":["Optimising Deep Learning at the Edge for Accurate Hourly Air Quality Prediction"],"prefix":"10.3390","volume":"21","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-2486-253X","authenticated-orcid":false,"given":"I Nyoman Kusuma","family":"Wardana","sequence":"first","affiliation":[{"name":"School of Engineering, University of Warwick, Coventry CV4 7AL, UK"},{"name":"Department of Electrical Engineering, Politeknik Negeri Bali, Badung, Bali 80364, Indonesia"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-4706-0049","authenticated-orcid":false,"given":"Julian W.","family":"Gardner","sequence":"additional","affiliation":[{"name":"School of Engineering, University of Warwick, Coventry CV4 7AL, UK"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-0568-5048","authenticated-orcid":false,"given":"Suhaib A.","family":"Fahmy","sequence":"additional","affiliation":[{"name":"Computer, Electrical and Mathematical Sciences and Engineering, King Abdullah University of Science and Technology (KAUST), Thuwal 23955, Saudi Arabia"},{"name":"School of Engineering, University of Warwick, Coventry CV4 7AL, UK"}]}],"member":"1968","published-online":{"date-parts":[[2021,2,4]]},"reference":[{"key":"ref_1","doi-asserted-by":"crossref","first-page":"6900","DOI":"10.1109\/ACCESS.2017.2778504","article-title":"A Survey on the Edge Computing for the Internet of Things","volume":"6","author":"Yu","year":"2018","journal-title":"IEEE Access"},{"key":"ref_2","doi-asserted-by":"crossref","unstructured":"Ngu, A.H.H., Gutierrez, M., Metsis, V., Nepal, S., and Sheng, M.Z. (2016). IoT Middleware: A Survey on Issues and Enabling technologies. IEEE Internet Things J., 1.","DOI":"10.1109\/JIOT.2016.2615180"},{"key":"ref_3","doi-asserted-by":"crossref","first-page":"2483","DOI":"10.1109\/JIOT.2017.2767291","article-title":"Evaluating Critical Security Issues of the IoT World: Present and Future Challenges","volume":"5","author":"Frustaci","year":"2018","journal-title":"IEEE Internet Things J."},{"key":"ref_4","doi-asserted-by":"crossref","first-page":"1655","DOI":"10.1109\/JPROC.2019.2921977","article-title":"Deep Learning With Edge Computing: A Review","volume":"107","author":"Chen","year":"2019","journal-title":"Proc. Inst. Electr. Electron. Eng. IEEE"},{"key":"ref_5","doi-asserted-by":"crossref","first-page":"637","DOI":"10.1109\/JIOT.2016.2579198","article-title":"Edge Computing: Vision and Challenges","volume":"3","author":"Shi","year":"2016","journal-title":"IEEE Internet Things J."},{"key":"ref_6","doi-asserted-by":"crossref","first-page":"4921","DOI":"10.1109\/JIOT.2019.2893866","article-title":"From Cloud Down to Things: An Overview of Machine Learning in Internet of Things","volume":"6","author":"Samie","year":"2019","journal-title":"IEEE Internet Things J."},{"key":"ref_7","doi-asserted-by":"crossref","first-page":"869","DOI":"10.1109\/COMST.2020.2970550","article-title":"Convergence of Edge Computing and Deep Learning: A Comprehensive Survey","volume":"22","author":"Wang","year":"2020","journal-title":"IEEE Commun. Surv. Tutorials"},{"key":"ref_8","doi-asserted-by":"crossref","first-page":"436","DOI":"10.1038\/nature14539","article-title":"Deep learning","volume":"521","author":"LeCun","year":"2015","journal-title":"Nature"},{"key":"ref_9","doi-asserted-by":"crossref","first-page":"223","DOI":"10.1109\/TSMC.2020.3042898","article-title":"Internet of Things as System of Systems: A Review of Methodologies, Frameworks, Platforms, and Tools","volume":"51","author":"Fortino","year":"2021","journal-title":"IEEE Trans. Syst. Man Cybern. Syst."},{"key":"ref_10","first-page":"90","article-title":"Modeling Opportunistic IoT Services in Open IoT Ecosystems","volume":"1867","author":"Fortino","year":"2017","journal-title":"CEUR Workshop Proc."},{"key":"ref_11","doi-asserted-by":"crossref","first-page":"55","DOI":"10.1109\/MSSC.2017.2745818","article-title":"Embedded Deep Neural Network Processing: Algorithmic and Processor Techniques Bring Deep Learning to IoT and Edge Devices","volume":"9","author":"Verhelst","year":"2017","journal-title":"IEEE Solid-State Circuits Mag."},{"key":"ref_12","first-page":"1929","article-title":"Dropout: A Simple Way to Prevent Neural Networks from Overfitting","volume":"15","author":"Srivastava","year":"2014","journal-title":"J. Mach. Learn. Res."},{"key":"ref_13","unstructured":"Hinton, G., Vinyals, O., and Dean, J. (2015). Distilling the Knowledge in a Neural Network. arXiv."},{"key":"ref_14","doi-asserted-by":"crossref","first-page":"103136","DOI":"10.1016\/j.micpro.2020.103136","article-title":"A fast and scalable architecture to run convolutional neural networks in low density FPGAs","volume":"77","author":"Duarte","year":"2020","journal-title":"Microprocess. Microsyst."},{"key":"ref_15","doi-asserted-by":"crossref","first-page":"43","DOI":"10.1016\/j.jclepro.2018.08.242","article-title":"Does air pollution affect public health and health inequality? Empirical evidence from China","volume":"203","author":"Yang","year":"2018","journal-title":"J. Clean. Prod."},{"key":"ref_16","doi-asserted-by":"crossref","first-page":"468","DOI":"10.1016\/j.ecolecon.2016.06.030","article-title":"Self-protection investment exacerbates air pollution exposure inequality in urban China","volume":"131","author":"Sun","year":"2017","journal-title":"Ecol. Econ."},{"key":"ref_17","doi-asserted-by":"crossref","first-page":"135771","DOI":"10.1016\/j.scitotenv.2019.135771","article-title":"Air quality prediction at new stations using spatially transferred bi-directional long short-term memory network","volume":"705","author":"Ma","year":"2020","journal-title":"Sci. Total Environ."},{"key":"ref_18","doi-asserted-by":"crossref","first-page":"101471","DOI":"10.1016\/j.scs.2019.101471","article-title":"An intelligent hybrid model for air pollutant concentrations forecasting: Case of Beijing in China","volume":"47","author":"Liu","year":"2019","journal-title":"Sustain. Cities Soc."},{"key":"ref_19","doi-asserted-by":"crossref","first-page":"252","DOI":"10.1016\/j.rse.2015.02.005","article-title":"Remote sensing of atmospheric fine particulate matter (PM2.5) mass concentration near the ground from satellite observation","volume":"160","author":"Zhang","year":"2015","journal-title":"Remote Sens. Environ."},{"key":"ref_20","doi-asserted-by":"crossref","unstructured":"Hamra, G.B., Guha, N., Cohen, A., Laden, F., Raaschou-Nielsen, O., Samet, J.M., Vineis, P., Forastiere, F., Saldiva, P., and Yorifuji, T. (2014). Outdoor Particulate Matter Exposure and Lung Cancer: A Systematic Review and Meta-Analysis. Environ. Health Perspect.","DOI":"10.1289\/ehp.1408092"},{"key":"ref_21","doi-asserted-by":"crossref","first-page":"1127","DOI":"10.1001\/jama.295.10.1127","article-title":"Fine Particulate Air Pollution and Hospital Admission for Cardiovascular and Respiratory Diseases","volume":"295","author":"Dominici","year":"2006","journal-title":"JAMA"},{"key":"ref_22","doi-asserted-by":"crossref","first-page":"367","DOI":"10.1289\/ehp.1103898","article-title":"Reducing Personal Exposure to Particulate Air Pollution Improves Cardiovascular Health in Patients with Coronary Heart Disease","volume":"120","author":"Langrish","year":"2012","journal-title":"Environ. Health Perspect."},{"key":"ref_23","doi-asserted-by":"crossref","first-page":"127453","DOI":"10.1016\/j.chemosphere.2020.127453","article-title":"Atmospheric particulate matter and potentially hazardous compounds around residential\/road side soil in an urban area","volume":"259","author":"Islam","year":"2020","journal-title":"Chemosphere"},{"key":"ref_24","doi-asserted-by":"crossref","first-page":"108662","DOI":"10.1016\/j.abb.2020.108662","article-title":"Effects of air pollution particles (ultrafine and fine particulate matter) on mitochondrial function and oxidative stress \u2013 Implications for cardiovascular and neurodegenerative diseases","volume":"696","author":"Daiber","year":"2020","journal-title":"Arch. Biochem. Biophys."},{"key":"ref_25","doi-asserted-by":"crossref","first-page":"8049504","DOI":"10.1155\/2020\/8049504","article-title":"A Machine Learning Approach to Predict Air Quality in California","volume":"2020","author":"Castelli","year":"2020","journal-title":"Complexity"},{"key":"ref_26","doi-asserted-by":"crossref","first-page":"101019","DOI":"10.1016\/j.ecoinf.2019.101019","article-title":"Predicting air quality with deep learning LSTM: Towards comprehensive models","volume":"55","author":"Navares","year":"2020","journal-title":"Ecol. Inform."},{"key":"ref_27","doi-asserted-by":"crossref","first-page":"997","DOI":"10.1016\/j.envpol.2017.08.114","article-title":"Long short-term memory neural network for air pollutant concentration predictions: Method development and evaluation","volume":"231","author":"Li","year":"2017","journal-title":"Environ. Pollut."},{"key":"ref_28","doi-asserted-by":"crossref","unstructured":"Xayasouk, T., Lee, H., and Lee, G. (2020). Air Pollution Prediction Using Long Short-Term Memory (LSTM) and Deep Autoencoder (DAE) Models. Sustainability, 12.","DOI":"10.3390\/su12062570"},{"key":"ref_29","doi-asserted-by":"crossref","first-page":"2021","DOI":"10.1016\/j.aej.2020.12.009","article-title":"Spatiotemporal prediction of air quality based on LSTM neural network","volume":"60","author":"Seng","year":"2021","journal-title":"Alex. Eng. J."},{"key":"ref_30","unstructured":"Xu, J., Chen, L., Lv, M., Zhan, C., Chen, S., and Chang, J. (2021). HighAir: A Hierarchical Graph Neural Network-Based Air Quality Forecasting Method. arXiv."},{"key":"ref_31","doi-asserted-by":"crossref","first-page":"117729","DOI":"10.1016\/j.jclepro.2019.117729","article-title":"A temporal-spatial interpolation and extrapolation method based on geographic Long Short-Term Memory neural network for PM2.5","volume":"237","author":"Ma","year":"2019","journal-title":"J. Clean. Prod."},{"key":"ref_32","doi-asserted-by":"crossref","first-page":"286","DOI":"10.1016\/j.chemosphere.2019.01.121","article-title":"An ensemble long short-term memory neural network for hourly PM2.5 concentration forecasting","volume":"222","author":"Bai","year":"2019","journal-title":"Chemosphere"},{"key":"ref_33","doi-asserted-by":"crossref","first-page":"486","DOI":"10.1016\/j.chemosphere.2018.12.128","article-title":"Long short-term memory - Fully connected (LSTM-FC) neural network for PM2.5 concentration prediction","volume":"220","author":"Zhao","year":"2019","journal-title":"Chemosphere"},{"key":"ref_34","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1016\/j.scitotenv.2019.01.333","article-title":"A hybrid model for spatiotemporal forecasting of PM2.5 based on graph convolutional neural network and long short-term memory","volume":"664","author":"Qi","year":"2019","journal-title":"Sci. Total Environ."},{"key":"ref_35","unstructured":"Zhang, Q., Lam, J.C., Li, V.O., and Han, Y. (2020). Deep-AIR: A Hybrid CNN-LSTM Framework forFine-Grained Air Pollution Forecast. arXiv."},{"key":"ref_36","doi-asserted-by":"crossref","first-page":"20050","DOI":"10.1109\/ACCESS.2019.2897028","article-title":"A Novel Combined Prediction Scheme Based on CNN and LSTM for Urban PM2.5 Concentration","volume":"7","author":"Qin","year":"2019","journal-title":"IEEE Access"},{"key":"ref_37","doi-asserted-by":"crossref","first-page":"26933","DOI":"10.1109\/ACCESS.2020.2971348","article-title":"A Hybrid CNN-LSTM Model for Forecasting Particulate Matter (PM2.5)","volume":"8","author":"Li","year":"2020","journal-title":"IEEE Access"},{"key":"ref_38","doi-asserted-by":"crossref","first-page":"76690","DOI":"10.1109\/ACCESS.2019.2921578","article-title":"Air Pollution Forecasting Using a Deep Learning Model Based on 1D Convnets and Bidirectional GRU","volume":"7","author":"Tao","year":"2019","journal-title":"IEEE Access"},{"key":"ref_39","unstructured":"Chen, H.C., Putra, K.T., and Chun-WeiLin, J. (2021). A Novel Prediction Approach for Exploring PM2.5 Spatiotemporal Propagation Based on Convolutional Recursive Neural Networks. arXiv."},{"key":"ref_40","unstructured":"Banner, R., Nahshan, Y., and Soudry, D. (2019, January 8\u201314). Post training 4-bit quantization of convolutional networks for rapid-deployment. Proceedings of the Advances in Neural Information Processing Systems, Vancouver, BC, Canada."},{"key":"ref_41","unstructured":"Wu, H., Judd, P., Zhang, X., Isaev, M., and Micikevicius, P. (2020). Integer Quantization for Deep Learning Inference: Principles and Empirical Evaluation. arXiv."},{"key":"ref_42","doi-asserted-by":"crossref","first-page":"194","DOI":"10.1016\/j.neucom.2020.12.035","article-title":"Fully integer-based quantization for mobile convolutional neural network inference","volume":"432","author":"Peng","year":"2021","journal-title":"Neurocomputing"},{"key":"ref_43","unstructured":"Li, J., and Alvarez, R. (2021). On the quantization of recurrent neural networks. arXiv."},{"key":"ref_44","unstructured":"Abadi, M., Agarwal, A., Barham, P., Brevdo, E., Chen, Z., Citro, C., Corrado, G.S., Davis, A., Dean, J., and Devin, M. (2016). TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems. arXiv."},{"key":"ref_45","doi-asserted-by":"crossref","first-page":"1735","DOI":"10.1162\/neco.1997.9.8.1735","article-title":"Long Short-Term Memory","volume":"9","author":"Hochreiter","year":"1997","journal-title":"Neural Comput."},{"key":"ref_46","unstructured":"(2021, January 22). Post-Training Quantization. Available online: https:\/\/www.tensorflow.org\/lite\/performance\/post_training_quantization."},{"key":"ref_47","first-page":"20170457","article-title":"Cautionary tales on air-quality improvement in Beijing","volume":"473","author":"Zhang","year":"2017","journal-title":"Proc. R. Soc. A Math. Phys. Eng. Sci."},{"key":"ref_48","doi-asserted-by":"crossref","first-page":"995","DOI":"10.1016\/j.future.2017.11.013","article-title":"Raspberry Pi assisted face recognition framework for enhanced law-enforcement services in smart cities","volume":"108","author":"Sajjad","year":"2017","journal-title":"Future Gener. Comput. Syst."}],"container-title":["Sensors"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/1424-8220\/21\/4\/1064\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,10,11]],"date-time":"2025-10-11T05:20:06Z","timestamp":1760160006000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/1424-8220\/21\/4\/1064"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2021,2,4]]},"references-count":48,"journal-issue":{"issue":"4","published-online":{"date-parts":[[2021,2]]}},"alternative-id":["s21041064"],"URL":"https:\/\/doi.org\/10.3390\/s21041064","relation":{},"ISSN":["1424-8220"],"issn-type":[{"value":"1424-8220","type":"electronic"}],"subject":[],"published":{"date-parts":[[2021,2,4]]}}}