Physics-Informed Deep Reinforcement Learningfor Dynamic 5G NR Network Slice Orchestrationin AGV Teleoperation

Physics-Informed Deep Reinforcement Learningfor Dynamic 5G NR Network Slice Orchestrationin AGV Teleoperation | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Physics-Informed Deep Reinforcement Learningfor Dynamic 5G NR Network Slice Orchestrationin AGV Teleoperation Mohammed Basingab, Muhammad Uzair This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9558189/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Teleoperating Automated Guided Vehicles (AGVs) over 5G networks demandssimultaneous ultra-low latency for control commands and sustained high through-put for camera feeds, yet wireless channel dynamics and mobility-induced han-dovers routinely violate both requirements. Standard deep reinforcement learning(DRL) approaches suffer from slow convergence and physically inconsistentresource allocation, because they ignore established propagation models, vehic-ular kinematics, and channel estimation theory. We propose Physics-InformedDRL (PI-DRL), a framework that embeds three domain-specific differential con-straints, namely 3GPP TR 38.901 channel propagation, vehicular kinematics,and wheel-slip dynamics, as differentiable penalty terms into a Proximal Pol-icy Optimisation objective, regularising the policy search to the manifold ofphysically realisable solutions. The agent couples a three-layer PI-LSTM featureextractor with multi-head self-attention to jointly optimise Physical ResourceBlock allocation, Bandwidth Part switching, and predictive handover across four5G NR slices. Evaluated in an OMNeT++ 6.0.3/Simu5G/INET co-simulationplatform across five geographic scenarios, PI-DRL reduces mean control-planelatency by 49 to 59% over static slicing (7.2 to 15.8 ms) while maintaining 99.998to 99.999% reliability, improves video throughput by up to 80.5%, converges37% faster than data-driven LSTM baselines, and achieves a 10.8-percentage-point improvement in handover success rate over reactive A3-RSRP. Ablationexperiments confirm that each physics loss contributes independently and syner-gistically, with the combined three-loss model yielding 15.7% higher reward thanstandard PPO. These results establish physics-guided reinforcement learning as a principled methodology for 5G resource orchestration in safety-critical vehicular teleoperation. 5G network slicing Deep reinforcement learning Physics-informed neural networks AGV teleoperation URLLC Predictive handover Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9558189","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":632034608,"identity":"b8db721d-aee5-4af1-b950-66eca3a9c5d8","order_by":0,"name":"Mohammed Basingab","email":"","orcid":"","institution":"King Abdulaziz University","correspondingAuthor":false,"prefix":"","firstName":"Mohammed","middleName":"","lastName":"Basingab","suffix":""},{"id":632034612,"identity":"27d126f2-6487-437b-9295-d2a29a21fa6c","order_by":1,"name":"Muhammad Uzair","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABD0lEQVRIiWNgGAWjYFACHjDJ2AAkDjAw2Mjxg7gJBcRrSTOWBLESDIjUAgSHEjccANF4tOjOyD346WaOneyG84cfHq7ccyBx8/nViR8eGDDI84sdwKrF7EZesnTutmTjDQeOGRw88+yO8bYbbzdLAB1mOHN2Ag4tOQZALcyJGw42GBxsOPBMdtuNsxtAWhIMbuPUYvw7d1t94obD7B+AWg4zbp5xdvMPAlrMgLYcTtxwjAdky2HFDfy92/DbcuaNmXXutuPGM8/wFAC1pBlL3ODdZpFgIIHbL8dzjG/nbquW7Tt/fPPHhgPAqOw/u/nmjwobeX5p7FqwAAmwSglilYMA/wFSVI+CUTAKRsEIAAA4pnEC2N+z4gAAAABJRU5ErkJggg==","orcid":"","institution":"","correspondingAuthor":true,"prefix":"","firstName":"Muhammad","middleName":"","lastName":"Uzair","suffix":""}],"badges":[],"createdAt":"2026-04-28 20:53:15","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9558189/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9558189/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":108183026,"identity":"832c0cc2-d772-4325-baf9-c1f9dd11c701","added_by":"auto","created_at":"2026-04-30 08:59:45","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":3035880,"visible":true,"origin":"","legend":"","description":"","filename":"WPCManuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9558189/v1_covered_e6bb2828-8793-4715-8d3b-2e3f738077d8.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Physics-Informed Deep Reinforcement Learningfor Dynamic 5G NR Network Slice Orchestrationin AGV Teleoperation","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"5G network slicing, Deep reinforcement learning, Physics-informed neural networks, AGV teleoperation, URLLC, Predictive handover","lastPublishedDoi":"10.21203/rs.3.rs-9558189/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9558189/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eTeleoperating Automated Guided Vehicles (AGVs) over 5G networks demandssimultaneous ultra-low latency for control commands and sustained high through-put for camera feeds, yet wireless channel dynamics and mobility-induced han-dovers routinely violate both requirements. Standard deep reinforcement learning(DRL) approaches suffer from slow convergence and physically inconsistentresource allocation, because they ignore established propagation models, vehic-ular kinematics, and channel estimation theory. We propose Physics-InformedDRL (PI-DRL), a framework that embeds three domain-specific differential con-straints, namely 3GPP TR 38.901 channel propagation, vehicular kinematics,and wheel-slip dynamics, as differentiable penalty terms into a Proximal Pol-icy Optimisation objective, regularising the policy search to the manifold ofphysically realisable solutions. The agent couples a three-layer PI-LSTM featureextractor with multi-head self-attention to jointly optimise Physical ResourceBlock allocation, Bandwidth Part switching, and predictive handover across four5G NR slices. Evaluated in an OMNeT++ 6.0.3/Simu5G/INET co-simulationplatform across five geographic scenarios, PI-DRL reduces mean control-planelatency by 49 to 59% over static slicing (7.2 to 15.8 ms) while maintaining 99.998to 99.999% reliability, improves video throughput by up to 80.5%, converges37% faster than data-driven LSTM baselines, and achieves a 10.8-percentage-point improvement in handover success rate over reactive A3-RSRP. Ablationexperiments confirm that each physics loss contributes independently and syner-gistically, with the combined three-loss model yielding 15.7% higher reward thanstandard PPO. These results establish physics-guided reinforcement learning as a principled methodology for 5G resource orchestration in safety-critical vehicular teleoperation.\u003c/p\u003e","manuscriptTitle":"Physics-Informed Deep Reinforcement Learningfor Dynamic 5G NR Network Slice Orchestrationin AGV Teleoperation","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-30 02:36:10","doi":"10.21203/rs.3.rs-9558189/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"9a1531a3-1ad6-4a27-82ee-b39ccdfcbdc4","owner":[],"postedDate":"April 30th, 2026","published":true,"recentEditorialEvents":[{"type":"checksComplete","content":"","date":"2026-04-30T06:43:41+00:00","index":"","fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-04-30T02:36:10+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-30 02:36:10","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9558189","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9558189","identity":"rs-9558189","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}