[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-121600-en":3,"doc-seo-121600-105":29,"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":20,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":11,"language":21,"language_code":22,"site_id":23,"html_lang":22,"table_of_contents":24,"faqs":25,"seo_title":26,"seo_description":14,"update_tm":27,"read_time":28},121600,7971461741311,"Ophelia","https://ap-avatar.wpscdn.com/avatar/74000253aff267980c6?x-image-process=image/resize,m_fixed,w_180,h_180&k=1779345379180704826",8,"Research & Report","Onboard Machine Learning for Satellite Edge Computing - The SPAICE Project Use Case","This work tackles the challenge of deploying flexible, AI-driven payload capabilities directly onboard next-generation satellites. In the SPAICE project, it presents the design and hardware deployment of hardware-optimized machine learning models for flexible payload operation and adaptive beamforming. The models are restructured to lower memory and parameter overhead, then quantized and compiled for the Versal ACAPAI platform, while Cross-Layer Equalization and Fast Fine-Tuning reduce quantization loss and preserve near-floating-point accuracy. Experiments show markedly faster inference than workstation execution, enabling real-time, reconfigurable payload operation with high computational efficiency.","Onboard Machine Learning for Satellite Edge Computing: The SPAICE Project Use Case  \nLuis M. Garcs-Socarrs†, Raudel Cuiman†, Flor Ortiz†, Juan A. Vsquez-Peralvo†, Jorge L. Gonzlez-Rios†,  \nMouhamad Chehaitly†, Arkadii Kazanskii†, Sahar Malmir†, Amirhossein Nik†, Jan Thoemel†, Sumit Kumar♭ , Marcele Kuhfuss†, Swetha Varadajulu†, Eva Lagunas†, Juan C. M. Duncan†, Jorge Querol†, Symeon Chatzinotas†  \n† Interdisciplinary Centre for Security, Reliability and Trust (SnT), University of Luxembourg, Luxembourg ♭ Luxembourg Institute of Science and Technology (LIST)  \nAbstract—This work addresses the challenge of implementing an artificial intelligence-driven flexible payload onboard for nextgeneration satellites. Within the SPAICE project, we present the design and hardware deployment of hardware-optimized machine learning models for flexible payload and adaptive beamforming. The models are restructured to reduce memory and parameter overhead, then quantized and compiled for the Versal ACAPAI platform. Optimization strategies, including Cross-Layer Equalization and Fast Fine-Tuning, mitigate quantization losses while maintaining near-floating-point accuracy. Experimental results demonstrate significantly faster inference than workstation implementations, confirming the feasibility of deploying advanced machine learning models onboard satellites for realtime, reconfigurable payload operation with high computational efficiency.  \nIndex Terms—Onboard satellite processing, Flexible payload, Adaptive beamforming, Machine learning, Convolutional neural networks, Versal ACAP  \nI. INTRODUCTION  \nRecent advances in satellite communications have highlighted the growing demand for greater flexibility and autonomy in payload operations. Conventional satellite systems depend heavily on ground-based processing, which introduces latency, limits adaptability, and increases operational costs. As mission requirements become more dynamic, onboard processing has emerged as a promising approach to enable real-time decision-making under strict resource constraints.  \nRapid advancement of machine learning (ML) and artificial intelligence (AI) has led to their integration into a wide range of applications, including space-based systems, autonomous navigation, and communication networks. In the case of Low Earth Orbit (LEO) satellites, the limited visibility windows to ground stations make fast and autonomous decision-making essential. Moreover, functional splits in next-generation satellite communication architectures require a flexible distribution of signal processing tasks between the space and ground segments. In this context, efficient onboard ML integration enables real-time adaptation to traffic and channel dynamics, maximizing resource utilization while ensuring service reliability.  \nThis work has been supported by the European Space Agency (ESA), which funded it under Contract No. 4000134522/21/NL/FGL,”Satellite Signal Processing Techniques using a Commercial Off-The-Shelf AI Chipset (SPAICE) .”Please note that the views of the authors of this paper do not necessarily reflect the views of ESA  \nHowever, deploying ML models in resource-constrained environments, such as onboard satellites and edge computing platforms, faces several challenges. These include computational limitations, memory constraints, power efficiency, and the need for real-time inference under dynamic conditions. Unlike traditional ML deployment on high-performance computing systems, where abundant resources enable complex model execution, embedded hardware platforms require substantial model optimization, quantization, and hardware-specific adaptations to achieve an optimal balance between performance and efficiency.  \nMachine learning for onboard satellite studies have focused initially on limited use cases and a narrow range of methods, restricting the generality of the results. Furthermore, the short evaluation periods have hindered comprehensive assessment and comparison agai","cbCaidyX7Y5AJJDI","https://ap.wps.com/l/cbCaidyX7Y5AJJDI","pdf",3102622,1,"English","en",105,"# Introduction\n## Motivation for onboard ML and AI\n## Challenges in resource-constrained deployment\n## Proposed offline training and onboard inference workflow\n## Related work and context","[{\"question\":\"What problem does the SPAICE project address in satellite edge computing?\",\"answer\":\"It addresses how to implement flexible AI-driven payload functions onboard next-generation satellites under strict resource constraints, enabling real-time adaptation with efficient computation.\"},{\"question\":\"How are the machine learning models prepared for onboard deployment?\",\"answer\":\"The models are restructured to reduce memory and parameter overhead, then quantized and compiled for the Versal ACAPAI platform to fit embedded hardware requirements.\"},{\"question\":\"Which optimization methods help maintain accuracy after quantization?\",\"answer\":\"Cross-Layer Equalization and Fast Fine-Tuning mitigate quantization losses while keeping accuracy close to floating-point performance.\"}]","Onboard Machine Learning for Satellite Edge Computing - The SPAICE Project Use Case | PDF",1785736422,20,{"code":4,"msg":30,"data":31},"ok",{"site_id":23,"language":22,"slug":32,"title":13,"keywords":33,"description":14,"schema_data":34,"social_meta":85,"head_meta":87,"extra_data":89,"updated_unix":27},"onboard-machine-learning-for-satellite-edge-computing-the-spaice-project-use-case","",{"@graph":35,"@context":84},[36,53,67],{"@type":37,"itemListElement":38},"BreadcrumbList",[39,43,47,50],{"item":40,"name":41,"@type":42,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":44,"name":45,"@type":42,"position":46},"https://docshare.wps.com/document/","Document",2,{"item":48,"name":12,"@type":42,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/onboard-machine-learning-for-satellite-edge-computing-the-spaice-project-use-case/121600/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":22,"description":14,"dateModified":61,"datePublished":61,"encodingFormat":60,"isAccessibleForFree":62,"interactionStatistic":63},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":40,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-08-03",true,{"@type":64,"interactionType":65,"userInteractionCount":20},"InteractionCounter",{"@type":66},"ViewAction",{"@type":68,"mainEntity":69},"FAQPage",[70,76,80],{"name":71,"@type":72,"acceptedAnswer":73},"What problem does the SPAICE project address in satellite edge computing?","Question",{"text":74,"@type":75},"It addresses how to implement flexible AI-driven payload functions onboard next-generation satellites under strict resource constraints, enabling real-time adaptation with efficient computation.","Answer",{"name":77,"@type":72,"acceptedAnswer":78},"How are the machine learning models prepared for onboard deployment?",{"text":79,"@type":75},"The models are restructured to reduce memory and parameter overhead, then quantized and compiled for the Versal ACAPAI platform to fit embedded hardware requirements.",{"name":81,"@type":72,"acceptedAnswer":82},"Which optimization methods help maintain accuracy after quantization?",{"text":83,"@type":75},"Cross-Layer Equalization and Fast Fine-Tuning mitigate quantization losses while keeping accuracy close to floating-point performance.","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":23},{"code":4,"msg":5,"data":91},[92,96,100,104,109,114,119,122,126,129,133],{"id":20,"doc_module":4,"doc_module_name":45,"category_name":93,"show_sort_weight":94,"slug":95},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":97,"show_sort_weight":98,"slug":99},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":45,"category_name":101,"show_sort_weight":102,"slug":103},"Exam",70,"exam",{"id":105,"doc_module":4,"doc_module_name":45,"category_name":106,"show_sort_weight":107,"slug":108},5,"Comic",60,"comic",{"id":110,"doc_module":4,"doc_module_name":45,"category_name":111,"show_sort_weight":112,"slug":113},6,"Technology",50,"technology",{"id":115,"doc_module":4,"doc_module_name":45,"category_name":116,"show_sort_weight":117,"slug":118},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":120,"slug":121},30,"research-report",{"id":123,"doc_module":4,"doc_module_name":45,"category_name":124,"show_sort_weight":28,"slug":125},9,"Religion & Spirituality","religion-spirituality",{"id":28,"doc_module":4,"doc_module_name":45,"category_name":127,"show_sort_weight":28,"slug":128},"World Cup","world-cup",{"id":130,"doc_module":4,"doc_module_name":45,"category_name":131,"show_sort_weight":130,"slug":132},10,"Lifestyle","lifestyle",{"id":134,"doc_module":4,"doc_module_name":45,"category_name":135,"show_sort_weight":105,"slug":136},19,"General","general"]