[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-seo-269390-105":3,"detail-sidebar-cat-0-en-105":81,"doc-detail-269390-en":131},{"code":4,"msg":5,"data":6},0,"ok",{"site_id":7,"language":8,"slug":9,"title":10,"keywords":11,"description":12,"schema_data":13,"social_meta":74,"head_meta":76,"extra_data":78,"updated_unix":80},105,"en","horizon-understanding-and-predicting-global-starlink-performance","Horizon - Understanding and Predicting Global Starlink Performance","","Starlink has deployed over 7,800 satellites, yet end-to-end performance prediction remains difficult due to rapid orbital dynamics, frequent terminal handovers, and weather-driven signal attenuation. Horizon is a global-scale machine learning system that combines 11 months of crowdsourced measurements from M-Lab and Cloudflare across 90+ countries with meteorological data and orbital propagation features. Evaluations on held-out temporal and standard splits reduce latency and throughput errors and outperform baselines, while location and atmospheric features quantify prediction drivers.",{"@graph":14,"@context":73},[15,34,56],{"@type":16,"itemListElement":17},"BreadcrumbList",[18,23,27,31],{"item":19,"name":20,"@type":21,"position":22},"https://docshare.wps.com","Home","ListItem",1,{"item":24,"name":25,"@type":21,"position":26},"https://docshare.wps.com/document/","Document",2,{"item":28,"name":29,"@type":21,"position":30},"https://docshare.wps.com/document/research-report/","Research & Report",3,{"item":32,"name":10,"@type":21,"position":33},"https://docshare.wps.com/document/horizon-understanding-and-predicting-global-starlink-performance/269390/",4,{"url":32,"name":10,"@type":35,"image":36,"author":41,"headline":10,"publisher":44,"fileFormat":47,"inLanguage":8,"description":12,"dateModified":48,"datePublished":49,"encodingFormat":47,"isAccessibleForFree":50,"interactionStatistic":51},"DigitalDocument",{"url":37,"@type":38,"width":39,"height":40},"https://docshare.wps.com/thumbnails/horizon-understanding-and-predicting-global-starlink-performance/269390.png","ImageObject",300,407,{"name":42,"@type":43},"Rizky","Person",{"url":19,"name":45,"@type":46},"DocShare","Organization","application/pdf","2026-11-04","2026-09-15",true,{"@type":52,"interactionType":53,"userInteractionCount":55},"InteractionCounter",{"@type":54},"ViewAction",16,{"@type":57,"mainEntity":58},"FAQPage",[59,65,69],{"name":60,"@type":61,"acceptedAnswer":62},"What makes Starlink performance difficult to predict at global scale?","Question",{"text":63,"@type":64},"Rapid orbital motion forces frequent handovers, ground routing changes over time, and weather causes additional signal attenuation, creating strong end-to-end variability.","Answer",{"name":66,"@type":61,"acceptedAnswer":67},"How does Horizon build globally generalizable performance models?",{"text":68,"@type":64},"Horizon uses crowdsourced measurement platforms to capture broad geographic diversity, then integrates meteorological data and satellite orbital propagation features with long-term measurements.",{"name":70,"@type":61,"acceptedAnswer":71},"What do the results show about latency and throughput prediction quality?",{"text":72,"@type":64},"On a held-out one-week temporal window, Horizon achieves mean absolute errors of 17.76 ms for latency and 25.63 Mbps for throughput, and it improves over baseline models on a standard split.","https://schema.org",{"og:url":32,"og:type":75,"og:title":10,"og:site_name":45,"og:description":12},"article",{"robots":77,"canonical":32},"index,follow",{"doc_id":79,"site_id":7},269390,1789451235,{"code":4,"msg":82,"data":83},"success",[84,88,92,96,101,106,111,115,120,123,127],{"id":22,"doc_module":4,"doc_module_name":25,"category_name":85,"show_sort_weight":86,"slug":87},"Story & Novel",90,"story-novel",{"id":26,"doc_module":4,"doc_module_name":25,"category_name":89,"show_sort_weight":90,"slug":91},"Literature",80,"literature",{"id":33,"doc_module":4,"doc_module_name":25,"category_name":93,"show_sort_weight":94,"slug":95},"Exam",70,"exam",{"id":97,"doc_module":4,"doc_module_name":25,"category_name":98,"show_sort_weight":99,"slug":100},5,"Comic",60,"comic",{"id":102,"doc_module":4,"doc_module_name":25,"category_name":103,"show_sort_weight":104,"slug":105},6,"Technology",50,"technology",{"id":107,"doc_module":4,"doc_module_name":25,"category_name":108,"show_sort_weight":109,"slug":110},7,"Healthcare",40,"healthcare",{"id":112,"doc_module":4,"doc_module_name":25,"category_name":29,"show_sort_weight":113,"slug":114},8,30,"research-report",{"id":116,"doc_module":4,"doc_module_name":25,"category_name":117,"show_sort_weight":118,"slug":119},9,"Religion & Spirituality",20,"religion-spirituality",{"id":118,"doc_module":4,"doc_module_name":25,"category_name":121,"show_sort_weight":118,"slug":122},"World Cup","world-cup",{"id":124,"doc_module":4,"doc_module_name":25,"category_name":125,"show_sort_weight":124,"slug":126},10,"Lifestyle","lifestyle",{"id":128,"doc_module":4,"doc_module_name":25,"category_name":129,"show_sort_weight":97,"slug":130},19,"General","general",{"code":4,"msg":82,"data":132},{"doc_id":79,"user_id":133,"nickname":42,"user_avatar":134,"doc_module":4,"category_id":112,"category_name":29,"doc_title":10,"doc_description":12,"doc_content":135,"file_id":136,"file_url":137,"file_type":138,"file_size":139,"view_count":55,"is_deleted":4,"is_public":22,"is_downloadable":22,"audit_status":22,"page_count":113,"language":140,"language_code":8,"site_id":7,"html_lang":8,"table_of_contents":141,"faqs":142,"seo_title":143,"seo_description":12,"update_tm":80,"read_time":144,"preview_page_count":4},962085564807,"https://ap-avatar.wpscdn.com/davatar_6f874abed73319feea01a86fa6f0fab8","Horizon: Understanding and Predicting Global Starlink Performance  \nCRISTIAN BENGHE∗ , TU Delft, The Netherlands VLAD GRAURE∗†, TU Delft, The Netherlands TANYA SHREEDHAR, TU Delft, The Netherlands NITINDER MOHAN, TU Delft, The Netherlands  \nStarlink has deployed over 7,800 satellites serving millions of subscribers, yet predicting its performance remains an open challenge. Rapid orbital dynamics, frequent handovers, and weather-induced signal attenuation create variability that existing models, built on a handful of instrumented terminals in limited regions, cannot capture at global scale. We present Horizon, the first global-scale machine learning system for predicting LEO satellite Internet performance. Our key insight is that crowdsourced measurement platforms, while noisier than controlled experiments, provide the geographic diversity necessary to build globally generalizable models. Horizon integrates 11 months of measurements from M-Lab and Cloudflare spanning 90+ countries with meteorological data and satellite orbital propagation features. On a fully held-out one-week temporal window, Horizon achieves mean absolute errors of 17.76 ms for latency and 25.63 Mbps for throughput; on a standard 80/20 split it outperforms all baselines, including adaptations of state-of-the-art architectures. Feature importance analysis reveals that geographic position dominates prediction, with latitude alone contributing 42–46%, while weather features account for 14–15%, quantifying the impact of atmospheric conditions on Ku/Ka-band links. Leave-one-location-out experiments confirm that Horizon generalizes to regions absent from training, enabling performance estimation where measurement infrastructure does not yet exist. Our dataset and pipeline are publicly available, providing a foundation for global LEO network performance  \nvisibility.  \nCCS Concepts: • Networks → Network performance modeling; Network performance analysis. Additional Key Words and Phrases: LEO Internet, Crowdsourced Telemetry, Performance Prediction, Starlink, Machine Learning  \nACM Reference Format:  \nCristian Benghe, Vlad Graure, Tanya Shreedhar, and Nitinder Mohan. 2026. Horizon: Understanding and Predicting Global Starlink Performance . Proc. ACM Meas. Anal. Comput. Syst. 10, 2, Article 41 (June 2026), 30 pages. [https://doi](https://doi.org/10.1145/3805639)[.](https://doi.org/10.1145/3805639)[org/10](https://doi.org/10.1145/3805639)[.](https://doi.org/10.1145/3805639)[1145/3805639](https://doi.org/10.1145/3805639)  \n1 Introduction  \nThe global demand for reliable Internet connectivity continues to grow, yet a large population of users in rural, remote, and underserved regions remain poorly connected because deploying terrestrial infrastructure is often economically infeasible at the “last mile” and across sparsely populated areas [24, 32] . Satellite Internet has long been viewed as a coverage solution, but traditional geostationary (GEO) systems incur inherently high round-trip delay due to the orbital altitude, yielding  \n∗ Both authors contributed equally to this research.  \n†Now employed at Google.  \nAuthors’ Contact Information: Cristian Benghe, [C](C.Benghe@student.tudelft.nl)[.](C.Benghe@student.tudelft.nl)[Benghe@student](C.Benghe@student.tudelft.nl)[.](C.Benghe@student.tudelft.nl)[tudelft](C.Benghe@student.tudelft.nl)[.](C.Benghe@student.tudelft.nl)[nl](C.Benghe@student.tudelft.nl), TU Delft, The Netherlands; Vlad Graure, [vladgraure@yahoo](vladgraure@yahoo.ca)[.](vladgraure@yahoo.ca)[ca](vladgraure@yahoo.ca), TU Delft, The Netherlands; Tanya Shreedhar, [t](t.shreedhar@tudelft.nl)[.](t.shreedhar@tudelft.nl)[shreedhar@tudelft](t.shreedhar@tudelft.nl)[.](t.shreedhar@tudelft.nl)[nl](t.shreedhar@tudelft.nl), TU Delft, The Netherlands; Nitinder Mohan, [N](N.Mohan@tudelft.nl)[.](N.Mohan@tudelft.nl)[Mohan@tudelft](N.Mohan@tudelft.nl)[.](N.Mohan@tudelft.nl)[nl](N.Mohan@tudelft.nl), TU Delft, The Netherlands.  \nThis work is licensed under a Creative Commons Attri","cbCaibVEJkPC9ryt","https://ap.wps.com/l/cbCaibVEJkPC9ryt","pdf",3873056,"English","# Introduction\n## Global connectivity needs and the role of satellite Internet\n## Why Low-Earth-Orbit performance is hard to predict","[{\"question\":\"What makes Starlink performance difficult to predict at global scale?\",\"answer\":\"Rapid orbital motion forces frequent handovers, ground routing changes over time, and weather causes additional signal attenuation, creating strong end-to-end variability.\"},{\"question\":\"How does Horizon build globally generalizable performance models?\",\"answer\":\"Horizon uses crowdsourced measurement platforms to capture broad geographic diversity, then integrates meteorological data and satellite orbital propagation features with long-term measurements.\"},{\"question\":\"What do the results show about latency and throughput prediction quality?\",\"answer\":\"On a held-out one-week temporal window, Horizon achieves mean absolute errors of 17.76 ms for latency and 25.63 Mbps for throughput, and it improves over baseline models on a standard split.\"}]","Horizon - Understanding and Predicting Global Starlink Performance | PDF",76]