[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-120569-en":3,"doc-seo-120569-105":30,"detail-sidebar-cat-0-en-105":90},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},120569,687197207919,"Theodora","https://ap-avatar.wpscdn.com/avatar/a000253d6f5f7c60be?x-image-process=image/resize,m_fixed,w_180,h_180&k=1779446848396160552",8,"Research & Report","Extending Machine Learning Based RF Coverage Predictions to 3D","Advances in fast mmWave signal power prediction enable machine-learning models to deliver accurate power estimates with real-time simulation speed. The work emphasizes improved training-data pre-processing to reduce noise in ray-tracing outputs, thereby lowering mean absolute error. It further develops methods for extending from 2D to full 3D coverage prediction by varying transmitter height. A UNet-based CNN with encoder-decoder design and regression on volumetric inputs is trained and evaluated across controlled outdoor scenarios.","Extending Machine Learning Based RF Coverage  \nPredictions to 3D  \nMuyao Chen, Mathieu Châteauvert, Jonathan Ethier  \nCommunications Research Centre Canada  \nOttawa, Ontario, Canada  \n[alice.chen@ised-isde.gc.ca](alice.chen@ised-isde.gc.ca), [mathieu.chateauvert@ised-isde.gc.ca](mathieu.chateauvert@ised-isde.gc.ca), [jonathan.ethier@ised-isde.gc.ca](jonathan.ethier@ised-isde.gc.ca)  \nAbstract—This paper discusses recent advancements made in the fast prediction of signal power in mmWave communications environments. Using machine learning (ML) it is possible to train models that provide power estimates that are both accurate and with real-time simulation speeds. Work involving improved training data pre-processing as well as 3D predictions with arbitrary transmitter height is discussed.  \nKeywords—propagation, machine learning, mmWave, CNN, 3D  \nI. INTRODUCTION  \nThe accurate prediction of communication system metrics is vital for the efficient and robust deployment of wireless networks. It is well established that deep learning techniques, including machine learning, can enhance the tools that operators and regulators use to analyze these deployments, via faster predictions or deeper insights into trends in existing data [1] .  \nOne area of interest is the prediction of radio frequency (RF) power in communications environments. This type of prediction provides access to quality-of-service metrics as well as insight into interference levels between adjacent deployments. It has been shown that convolutional neural networks (CNNs) can provide accurate power predictions when trained with simulation data from traditional RF power simulation tools [2] . CNN-based modeling work was done in [3] and [4] where raytracing simulation software [5,6] was used to generate the training data, forming the basis of prediction models. This offered more accurate simulations but presented additional challenges due to the complexity of the predicted power distributions. The work was shown to successfully provide accurate predictions relative to their ray-tracing tool counterparts, though with significant simulation speed improvements.  \nThe work in this paper is a continuation of [3,4] with a new focus exploring more efficient pre-processing techniques for training data and constructing models with arbitrary transmitter height placements leading to full 3D prediction capabilities.  \nII. RECENT PRE-PROCESSING TECHNIQUE ADVANCEMENT  \nRay tracing simulation outputs can often be noisy. In order to address the noise issue, we apply an algorithm [7] that yields a locally time-averaged result of the rays impinging on each simulation point in the scene, resulting in physically based data smoothing. The small variations in the ray-tracing output power would increase the prediction error as the ML model cannot learn how to predict noisy variations. By decreasing the noise in the training and test sets, the mean absolute error (MAE) was  \nreduced from 1.42 to 0.55 dB. These test sets have power maps with 1 meter resolution per pixel, 32x32 scene size, 28 GHz operating frequency, 720k training samples and 180k test samples. As a soundness check, simulations were performed in scenes with no buildings (i.e., empty space with terrain) and the MAE of predictions were as low as 0.13 dB, approaching zero error. This is as one would expect since ML ought to learn simple scenes lacking non-line-of-sight with ease. Similar improvements were observed for larger scenes.  \nFig. 1. Impact of the local time-averaging smoothing algorithm.  \nIII. EXTENDING THE MODEL TO 3D PREDICTIONS  \nThis section is primarily focused on investigating the possibility of extending the current 2D RF simulation tool to 3D, motivated by the need to vary the transmitter height to achieve a more flexible prediction tool. In this study, an ML approach has been developed to optimize outdoor wireless coverage using pixel-wise regression models and CNNs. The 3D prediction model offers an advantage over ","cbCaii0I7ARYjjP2","https://ap.wps.com/l/cbCaii0I7ARYjjP2","pdf",310301,1,2,"English","en",105,"# Introduction\n## Recent pre-processing technique advancement\n# Extending the model to 3D predictions\n## Data acquisition and preparation\n## Power simulation scenarios\n## Deep learning model training and test","[{\"question\":\"Why is fast signal power prediction important in mmWave communications?\",\"answer\":\"Fast prediction supports efficient and robust wireless network deployment by enabling quicker analysis of deployment metrics and interference insights.\"},{\"question\":\"How does the pre-processing technique improve prediction accuracy?\",\"answer\":\"A locally time-averaging smoothing algorithm reduces noise in ray-tracing outputs, preventing the ML model from being trained on noisy power variations and lowering mean absolute error.\"},{\"question\":\"What changes when extending RF coverage predictions from 2D to 3D?\",\"answer\":\"3D inputs are built by concatenating multiple 2D planes at different elevations, enabling the transmitter position to vary in 3D while maintaining fast prediction speed.\"}]","Extending Machine Learning Based RF Coverage Predictions to 3D | PDF",1785730693,5,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":85,"head_meta":87,"extra_data":89,"updated_unix":28},"extending-machine-learning-based-rf-coverage-predictions-to-3d","",{"@graph":36,"@context":84},[37,53,67],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,47,50],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":21},"https://docshare.wps.com/document/","Document",{"item":48,"name":12,"@type":43,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":43,"position":52},"https://docshare.wps.com/document/extending-machine-learning-based-rf-coverage-predictions-to-3d/120569/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":23,"description":14,"dateModified":61,"datePublished":61,"encodingFormat":60,"isAccessibleForFree":62,"interactionStatistic":63},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":41,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-08-03",true,{"@type":64,"interactionType":65,"userInteractionCount":4},"InteractionCounter",{"@type":66},"ViewAction",{"@type":68,"mainEntity":69},"FAQPage",[70,76,80],{"name":71,"@type":72,"acceptedAnswer":73},"Why is fast signal power prediction important in mmWave communications?","Question",{"text":74,"@type":75},"Fast prediction supports efficient and robust wireless network deployment by enabling quicker analysis of deployment metrics and interference insights.","Answer",{"name":77,"@type":72,"acceptedAnswer":78},"How does the pre-processing technique improve prediction accuracy?",{"text":79,"@type":75},"A locally time-averaging smoothing algorithm reduces noise in ray-tracing outputs, preventing the ML model from being trained on noisy power variations and lowering mean absolute error.",{"name":81,"@type":72,"acceptedAnswer":82},"What changes when extending RF coverage predictions from 2D to 3D?",{"text":83,"@type":75},"3D inputs are built by concatenating multiple 2D planes at different elevations, enabling the transmitter position to vary in 3D while maintaining fast prediction speed.","https://schema.org",{"og:url":51,"og:type":86,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":88,"canonical":51},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":91},[92,96,100,104,108,113,118,121,126,129,133],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":93,"show_sort_weight":94,"slug":95},"Story & Novel",90,"story-novel",{"id":21,"doc_module":4,"doc_module_name":46,"category_name":97,"show_sort_weight":98,"slug":99},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":46,"category_name":101,"show_sort_weight":102,"slug":103},"Exam",70,"exam",{"id":29,"doc_module":4,"doc_module_name":46,"category_name":105,"show_sort_weight":106,"slug":107},"Comic",60,"comic",{"id":109,"doc_module":4,"doc_module_name":46,"category_name":110,"show_sort_weight":111,"slug":112},6,"Technology",50,"technology",{"id":114,"doc_module":4,"doc_module_name":46,"category_name":115,"show_sort_weight":116,"slug":117},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":119,"slug":120},30,"research-report",{"id":122,"doc_module":4,"doc_module_name":46,"category_name":123,"show_sort_weight":124,"slug":125},9,"Religion & Spirituality",20,"religion-spirituality",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":127,"show_sort_weight":124,"slug":128},"World Cup","world-cup",{"id":130,"doc_module":4,"doc_module_name":46,"category_name":131,"show_sort_weight":130,"slug":132},10,"Lifestyle","lifestyle",{"id":134,"doc_module":4,"doc_module_name":46,"category_name":135,"show_sort_weight":29,"slug":136},19,"General","general"]