{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,4]],"date-time":"2026-05-04T11:44:19Z","timestamp":1777895059716,"version":"3.51.4"},"reference-count":46,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2025,10,24]],"date-time":"2025-10-24T00:00:00Z","timestamp":1761264000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Commun. Netw."],"abstract":"<jats:sec>\n                    <jats:title>Introduction<\/jats:title>\n                    <jats:p>Beam-level traffic forecasting plays a vital role in the optimization of 5G networks by enabling proactive resource allocation and congestion control. However, the task is complicated by inherent data sparsity and the presence of multi-scale temporal dynamics, making accurate predictions difficult to achieve using conventional models.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>To address these challenges, we propose a Gated Recurrent Unit (GRU)-based Multi-Task Learning (MTL) framework, enhanced by a weighted ensemble approach. We systematically evaluate the performance of six forecasting models\u2014Linear Regression, DLinear, XGBoost, Echo State Network (ESN), Long Short-Term Memory (LSTM), and GRU-MTL\u2014across three input sequence lengths (168-h, 24-h, and 8-h) using real-world beam-level data from the ITU AI for Good initiative.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>Experimental findings reveal that the GRU-MTL model significantly outperforms traditional baselines, achieving a Mean Absolute Error (MAE) of 0.2136 on 168-h sequences compared to LSTM\u2019s 0.3223. Long sequences (168-h) reduce MAE by 56% relative to short 8-h windows, effectively mitigating the effects of sparsity. Furthermore, an ensemble of top-performing models (MTL, XGBoost, and Linear Regression) yields additional gains, reducing MAE to 0.2105\u2014a 1.45% improvement over MTL alone.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Discussion<\/jats:title>\n                    <jats:p>These results highlight the importance of long-term temporal context and model diversity for robust traffic prediction in sparse environments. The proposed framework offers practical guidelines: 168-h forecasting windows are optimal for weekly planning, and model ensembling enhances generalization across varying beam activity levels. This study contributes a scalable and accurate solution for spatio-temporal traffic forecasting in next-generation wireless networks.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.3389\/frcmn.2025.1658461","type":"journal-article","created":{"date-parts":[[2025,10,24]],"date-time":"2025-10-24T05:28:20Z","timestamp":1761283700000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":1,"title":["Spatio-temporal beam-level traffic forecasting in 5G wireless systems using multi-task learning"],"prefix":"10.3389","volume":"6","author":[{"given":"Israel","family":"Tommy","sequence":"first","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Taoreed","family":"Akinola","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Xiangfang","family":"Li","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Lijun","family":"Qian","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"1965","published-online":{"date-parts":[[2025,10,24]]},"reference":[{"key":"B1","unstructured":"3gpp release 17 description\n          \n          \n          2022"},{"key":"B2","doi-asserted-by":"publisher","first-page":"1617","DOI":"10.1109\/comst.2016.2532458","article-title":"Next generation 5g wireless networks: a comprehensive survey","volume":"18","author":"Agiwal","year":"2016","journal-title":"IEEE Commun. Surv. and Tutorials"},{"key":"B3","unstructured":"Spatio-temporal beam-level traffic forecasting challenge by itu\n          \n          \n          2024"},{"key":"B4","doi-asserted-by":"publisher","first-page":"148191","DOI":"10.1109\/ACCESS.2021.3124812","article-title":"6G wireless communications networks: a comprehensive survey","volume":"9","author":"Alsabah","year":"2012","journal-title":"IEEE Access"},{"key":"B5","first-page":"3312","article-title":"Outage-based beamforming for robust 6g millimeter-wave communication","volume":"70","author":"Alrabeiah","year":"2022","journal-title":"IEEE Trans. Commun."},{"key":"B6","volume-title":"Time series analysis: forecasting and control","author":"Box","year":"2015"},{"key":"B7","doi-asserted-by":"crossref","DOI":"10.1007\/b97391","volume-title":"Introduction to time series and forecasting","author":"Brockwell","year":"2002"},{"key":"B8","doi-asserted-by":"crossref","first-page":"785","DOI":"10.1145\/2939672.2939785","article-title":"Xgboost: a scalable tree boosting system","volume-title":"Proceedings of the 22nd acm sigkdd international conference on knowledge discovery and data mining","author":"Chen","year":"2016"},{"key":"B9","doi-asserted-by":"publisher","first-page":"3776","DOI":"10.3390\/s20133776","article-title":"Multitask learning and gcn-based taxi demand prediction for a traffic road network","volume":"20","author":"Chen","year":"2020","journal-title":"Sensors"},{"key":"B10","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1406.1078","article-title":"Learning phrase representations using rnn encoder-decoder for statistical machine translation","author":"Cho","year":"2014","journal-title":"arXiv Prepr"},{"key":"B11","unstructured":"Cisco annual internet report (2018\u20132023)\n          \n          \n          2020"},{"key":"B12","doi-asserted-by":"crossref","first-page":"160","DOI":"10.1145\/1390156.1390177","article-title":"A unified architecture for natural language processing: deep neural networks with multitask learning","volume-title":"Proceedings of the 25th international conference on machine learning","author":"Collobert","year":"2008"},{"key":"B13","doi-asserted-by":"crossref","first-page":"109","DOI":"10.1145\/1014052.1014067","article-title":"Regularized multi\u2013task learning","volume-title":"Proceedings of the tenth ACM SIGKDD international conference on knowledge discovery and data mining","author":"Evgeniou","year":"2004"},{"key":"B14","doi-asserted-by":"crossref","first-page":"324","DOI":"10.1109\/YAC.2016.7804912","article-title":"Using lstm and gru neural network methods for traffic flow prediction","volume-title":"2016 31st youth academic annual conference of Chinese association of automation (YAC)","author":"Fu","year":"2016"},{"key":"B15","doi-asserted-by":"publisher","first-page":"33","DOI":"10.1016\/j.neunet.2018.08.002","article-title":"Design of deep echo state networks","volume":"108","author":"Gallicchio","year":"2018","journal-title":"Neural Netw."},{"key":"B16","doi-asserted-by":"publisher","first-page":"637","DOI":"10.1016\/j.ijforecast.2006.03.005","article-title":"Exponential smoothing: the state of the art\u2014part ii","volume":"22","author":"Gardner","year":"2006","journal-title":"Int. J. Forecast."},{"key":"B17","doi-asserted-by":"publisher","first-page":"55","DOI":"10.1109\/mcom.001.1900411","article-title":"Toward 6g networks: use cases and technologies","volume":"58","author":"Giordani","year":"2020","journal-title":"IEEE Commun. Mag."},{"key":"B18","doi-asserted-by":"publisher","first-page":"2222","DOI":"10.1109\/TNNLS.2016.2582924","article-title":"Lstm: a search space odyssey","volume":"28","author":"Greff","year":"2016","journal-title":"IEEE Trans. neural Netw. Learn. Syst."},{"key":"B19","volume-title":"Forecasting: principles and practice","author":"Hyndman","year":"2018"},{"key":"B20","volume-title":"Imt vision \u2013 framework and overall objectives of the future development of imt for 2020 and beyond","year":"2020"},{"key":"B21","first-page":"95","article-title":"Modeling long-and short-term temporal patterns with deep neural networks","volume-title":"The 41st international ACM SIGIR conference on research and development in information retrieval","author":"Lai","year":"2018"},{"key":"B22","doi-asserted-by":"publisher","first-page":"1","DOI":"10.2307\/1269547","article-title":"Zero-inflated poisson regression, with an application to defects in manufacturing","volume":"34","author":"Lambert","year":"1992","journal-title":"Technometrics"},{"key":"B23","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1098\/rsta.2020.0209","article-title":"Time-series forecasting with deep learning","volume":"379","author":"Lim","year":"2021","journal-title":"Philosophical Trans. Math. Phys. Eng. Sci."},{"key":"B24","doi-asserted-by":"publisher","first-page":"2518","DOI":"10.1038\/s41598-024-51933-8","article-title":"Method for multi-task learning fusion network traffic classification to address small sample labels","volume":"14","author":"Liu","year":"2024","journal-title":"Sci. Rep."},{"key":"B25","doi-asserted-by":"publisher","first-page":"10 362","DOI":"10.1109\/tvt.2020.3005724","article-title":"Multi-task learning at the mobile edge: an effective way to combine traffic classification and prediction","volume":"69","author":"Rago","year":"2020","journal-title":"IEEE Trans. Veh. Technol."},{"key":"B26","doi-asserted-by":"crossref","first-page":"187","DOI":"10.1109\/ICMLA.2018.00035","article-title":"Network traffic prediction using recurrent neural networks","volume-title":"2018 17th IEEE international conference on machine learning and applications (ICMLA)","author":"Ramakrishnan","year":"2018"},{"key":"B27","doi-asserted-by":"publisher","first-page":"335","DOI":"10.1109\/access.2013.2260813","article-title":"Millimeter wave Mobile communications for 5g cellular: it will work","volume":"1","author":"Rappaport","year":"2013","journal-title":"IEEE Access"},{"key":"B28","first-page":"1","article-title":"Wireless communications and applications above 100 ghz: opportunities and challenges for 6g and beyond","volume-title":"IEEE international conference on communications (ICC)","author":"Rappaport","year":"2019"},{"key":"B29","doi-asserted-by":"publisher","first-page":"4997","DOI":"10.3390\/s23114997","article-title":"A novel traffic prediction method using machine learning for energy efficiency in service provider networks","volume":"23","author":"Rau","year":"2023","journal-title":"Sensors"},{"key":"B30","first-page":"56","article-title":"5g beamforming and its impact on wireless system design","volume":"19","author":"Rohde","year":"2018","journal-title":"IEEE Microw. Mag."},{"key":"B31","article-title":"An overview of multi-task learning in deep neural networks","author":"Ruder","year":"2017"},{"key":"B32","first-page":"1394","article-title":"A comparison of arima and lstm in forecasting time series","volume-title":"17th IEEE international conference on machine learning and applications (ICMLA)","author":"Siami-Namini","year":"2018"},{"key":"B33","doi-asserted-by":"publisher","first-page":"3722","DOI":"10.3390\/en13143722","article-title":"Forecasting hierarchical time series in power generation","volume":"13","author":"Silveira Gontijo","year":"2020","journal-title":"Energies"},{"key":"B34","doi-asserted-by":"publisher","first-page":"441","DOI":"10.15918\/j.jbit1004\u20100579.2022.065","article-title":"Spatio-temporal cellular network traffic prediction using multi-task deep learning for ai-enabled 6g","volume":"31","author":"Sun","year":"2022","journal-title":"J. Beijing Inst. Technol."},{"key":"B35","doi-asserted-by":"publisher","first-page":"606","DOI":"10.1016\/j.ejor.2011.05.018","article-title":"Intermittent demand: linking forecasting to inventory obsolescence","volume":"214","author":"Teunter","year":"2011","journal-title":"Eur. J. Operational Res."},{"key":"B36","doi-asserted-by":"crossref","first-page":"285","DOI":"10.1109\/CSE.2011.58","article-title":"Internet traffic classification using machine learning: a token-based approach","volume-title":"2011 14th IEEE international conference on computational science and engineering","author":"Wang","year":"2011"},{"key":"B37","doi-asserted-by":"publisher","first-page":"7475","DOI":"10.1109\/tii.2022.3141743","article-title":"A multitask learning-based network traffic prediction approach for sdn-enabled industrial internet of things","volume":"18","author":"Wang","year":"2022","journal-title":"IEEE Trans. Industrial Inf."},{"key":"B38","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2504.03792","article-title":"Dp-let: an efficient spatio-temporal network traffic prediction framework","author":"Wang","year":"2025","journal-title":"arXiv Prepr"},{"key":"B39","first-page":"76 112","article-title":"A survey on machine learning-based traffic prediction in cellular networks","volume":"8","author":"Wu","year":"2020","journal-title":"IEEE Access"},{"key":"B40","doi-asserted-by":"publisher","first-page":"5543","DOI":"10.3390\/s24175543","article-title":"Multi-scale spatio-temporal attention networks for network-scale traffic learning and forecasting","volume":"24","author":"Wu","year":"2024","journal-title":"Sensors"},{"key":"B41","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1709.04875","article-title":"Spatio-temporal graph convolutional networks: a deep learning framework for traffic forecasting","author":"Yu","year":"2017","journal-title":"arXiv Prepr"},{"key":"B42","doi-asserted-by":"publisher","first-page":"11 121","DOI":"10.1609\/aaai.v37i9.26317","article-title":"Are transformers effective for time series forecasting?","volume":"37","author":"Zeng","year":"2023","journal-title":"Proc. AAAI Conf. Artif. Intell."},{"key":"B43","doi-asserted-by":"publisher","first-page":"5586","DOI":"10.1109\/tkde.2021.3070203","article-title":"A survey on multi-task learning","volume":"34","author":"Zhang","year":"2021","journal-title":"IEEE Trans. Knowl. data Eng."},{"key":"B44","doi-asserted-by":"crossref","DOI":"10.1609\/aaai.v31i1.10735","article-title":"Deep spatio-temporal residual networks for citywide crowd flows prediction","author":"Zhang","year":"2017"},{"key":"B45","first-page":"1234","article-title":"Artificial intelligence-driven network traffic forecasting: a survey","volume":"25","author":"Zhang","year":"2023","journal-title":"IEEE Commun. Surv. and Tutorials"},{"key":"B46","doi-asserted-by":"publisher","first-page":"1727","DOI":"10.1109\/tnsm.2024.3522115","article-title":"An adaptive ensemble learning paradigm with spatial-temporal feature extraction for wireless traffic prediction","volume":"22","author":"Zhu","year":"2024","journal-title":"IEEE Trans. Netw. Serv. Manag."}],"container-title":["Frontiers in Communications and Networks"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/frcmn.2025.1658461\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,10,24]],"date-time":"2025-10-24T05:28:25Z","timestamp":1761283705000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/frcmn.2025.1658461\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,10,24]]},"references-count":46,"alternative-id":["10.3389\/frcmn.2025.1658461"],"URL":"https:\/\/doi.org\/10.3389\/frcmn.2025.1658461","relation":{},"ISSN":["2673-530X"],"issn-type":[{"value":"2673-530X","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,10,24]]},"article-number":"1658461"}}