A deep reinforcement learning-based task offloading algorithm for cell-free architecture CFMADRL: Co-optimization of delay-energy sensitive tasks

A deep reinforcement learning-based task offloading algorithm for cell-free architecture CFMADRL: Co-optimization of delay-energy sensitive tasks | 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 A deep reinforcement learning-based task offloading algorithm for cell-free architecture CFMADRL: Co-optimization of delay-energy sensitive tasks Keke Li, Xiaochun Wu, Yichao Lou, Mingjun Qi This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6997149/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 17 You are reading this latest preprint version Abstract With the continuous proliferation of wireless devices, the growing pressure on wireless channels has resulted in degraded network quality for edge users and increased energy consumption of mobile devices. To enhance user experience and extend device battery life, this paper proposes a multi-agent deep reinforcement learning-based task offloading algorithm, named Cell-Free Multi-Agent Deep Reinforcement Learning (CFMADRL). Under a cell-free architecture, the proposed model introduces two representative types of tasks, namely delay-sensitive and energy-sensitive tasks, and designs a multidimensional task classification mechanism that integrates task complexity, device status, and delay/energy pressure metrics. To effectively handle heterogeneous tasks, a dual-agent collaborative framework is constructed, where each agent is dedicated to a specific optimization objective: minimizing task completion delay or reducing energy consumption, through global task offloading and resource scheduling. Furthermore, CFMADRL incorporates a user-driven access point (AP) cooperative offloading mechanism and a hierarchical optimization strategy. The system optimization problem is decomposed into two subproblems: computational resource allocation and task offloading/power control, which are addressed using convex optimization and multi-agent deep reinforcement learning, respectively. Simulation results validate that the proposed algorithm significantly outperforms existing benchmark methods in terms of reducing system delay and energy consumption, and demonstrates strong robustness and adaptability in dynamic edge computing environments. Task Offloading Cell-Free Architecture Delay-Sensitive Tasks Energy-Sensitive Tasks Deep Reinforcement Learning Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Editorial decision: Revision requested 18 Oct, 2025 Reviews received at journal 14 Oct, 2025 Reviews received at journal 12 Oct, 2025 Reviews received at journal 08 Oct, 2025 Reviews received at journal 07 Oct, 2025 Reviewers agreed at journal 29 Sep, 2025 Reviewers agreed at journal 29 Sep, 2025 Reviewers agreed at journal 27 Sep, 2025 Reviewers agreed at journal 27 Sep, 2025 Reviews received at journal 27 Sep, 2025 Reviewers agreed at journal 26 Sep, 2025 Reviewers agreed at journal 26 Sep, 2025 Reviewers agreed at journal 26 Sep, 2025 Reviewers invited by journal 25 Sep, 2025 Editor assigned by journal 01 Jul, 2025 Submission checks completed at journal 01 Jul, 2025 First submitted to journal 28 Jun, 2025 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-6997149","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":525062382,"identity":"e43732a9-1621-46c0-965d-01a58c4936f3","order_by":0,"name":"Keke Li","email":"","orcid":"","institution":"Zhejiang Gongshang University","correspondingAuthor":false,"prefix":"","firstName":"Keke","middleName":"","lastName":"Li","suffix":""},{"id":525062383,"identity":"77a0f376-5941-48a9-82e6-2dfcd3b7c1ab","order_by":1,"name":"Xiaochun Wu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABB0lEQVRIiWNgGAWjYJCCA0DMzMbffIBBAiJgQJQWdn6JYwnEawEBfsmGHLhK/FrkZ/c+PFzwi0Ha4MCZzx8sftUlNrA3b5NgqLmDU4vBneMGh2f2MRgbHO7dYCDZdzixgedYmQTDsWe4tUikMRzm7fmfbHDg7IYEyZ4DiQ0SOWYSjA2HcTtsBlgLQ/2GAzkPDkj2AB0m/wa/FoYbQC08PxiYgd5nbJD4wQy0hQe/FoM7x4C2NDAwAwPZmEGy4bBxG09asUXCMTwOm93G/JnnDzgqH3+W+FMn289+eOONDzV4HAaKPcY2CJtZEshgA7EScGuAaGH4A2EzfviDT+koGAWjYBSMVAAAz1FXtGic4SkAAAAASUVORK5CYII=","orcid":"","institution":"Zhejiang Gongshang University","correspondingAuthor":true,"prefix":"","firstName":"Xiaochun","middleName":"","lastName":"Wu","suffix":""},{"id":525062384,"identity":"cdf0faf3-464e-460f-82f8-462d195b6d59","order_by":2,"name":"Yichao Lou","email":"","orcid":"","institution":"Zhejiang Gongshang University","correspondingAuthor":false,"prefix":"","firstName":"Yichao","middleName":"","lastName":"Lou","suffix":""},{"id":525062385,"identity":"aa922bd8-5aac-48f0-9dfc-c208b4da51e8","order_by":3,"name":"Mingjun Qi","email":"","orcid":"","institution":"Zhejiang Gongshang University","correspondingAuthor":false,"prefix":"","firstName":"Mingjun","middleName":"","lastName":"Qi","suffix":""}],"badges":[],"createdAt":"2025-06-28 10:23:12","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6997149/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6997149/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":93167185,"identity":"5153ab47-e7db-43dc-964e-c7c352144d20","added_by":"auto","created_at":"2025-10-09 18:16:03","extension":"json","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":6338,"visible":true,"origin":"","legend":"","description":"","filename":"bbf0bcf3f8444b46b65934405d9ce474.json","url":"https://assets-eu.researchsquare.com/files/rs-6997149/v1/5dc50f9dc08eed016ae578ad.json"},{"id":93168100,"identity":"55e57beb-c78e-462a-a90a-c16f341d4e0a","added_by":"auto","created_at":"2025-10-09 18:24:04","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1035569,"visible":true,"origin":"","legend":"","description":"","filename":"snarticletemplate.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6997149/v1_covered_6d5769b8-7ec0-449d-ba35-7c630ba071a8.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"A deep reinforcement learning-based task offloading algorithm for cell-free architecture CFMADRL: Co-optimization of delay-energy sensitive tasks","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"the-journal-of-supercomputing","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"Learn more about [The Journal of Supercomputing](https://www.springer.com/journal/11227)","snPcode":"11227","submissionUrl":"https://submission.nature.com/new-submission/11227/3","title":"The Journal of Supercomputing","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"Task Offloading, Cell-Free Architecture, Delay-Sensitive Tasks, Energy-Sensitive Tasks, Deep Reinforcement Learning","lastPublishedDoi":"10.21203/rs.3.rs-6997149/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6997149/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"With the continuous proliferation of wireless devices, the growing pressure on wireless channels has resulted in degraded network quality for edge users and increased energy consumption of mobile devices. To enhance user experience and extend device battery life, this paper proposes a multi-agent deep reinforcement learning-based task offloading algorithm, named Cell-Free Multi-Agent Deep Reinforcement Learning (CFMADRL).\nUnder a cell-free architecture, the proposed model introduces two representative types of tasks, namely delay-sensitive and energy-sensitive tasks, and designs a multidimensional task classification mechanism that integrates task complexity, device status, and delay/energy pressure metrics.\nTo effectively handle heterogeneous tasks, a dual-agent collaborative framework is constructed, where each agent is dedicated to a specific optimization objective: minimizing task completion delay or reducing energy consumption, through global task offloading and resource scheduling.\nFurthermore, CFMADRL incorporates a user-driven access point (AP) cooperative offloading mechanism and a hierarchical optimization strategy. The system optimization problem is decomposed into two subproblems: computational resource allocation and task offloading/power control, which are addressed using convex optimization and multi-agent deep reinforcement learning, respectively.\nSimulation results validate that the proposed algorithm significantly outperforms existing benchmark methods in terms of reducing system delay and energy consumption, and demonstrates strong robustness and adaptability in dynamic edge computing environments. ","manuscriptTitle":"A deep reinforcement learning-based task offloading algorithm for cell-free architecture CFMADRL: Co-optimization of delay-energy sensitive tasks","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-10-09 18:15:59","doi":"10.21203/rs.3.rs-6997149/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-10-19T01:16:23+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-10-14T13:35:34+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-10-12T09:30:24+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-10-09T00:52:50+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-10-08T00:41:23+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"267403195224056907110302863822779492252","date":"2025-09-29T15:18:52+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"293642935746640742722460159161628976764","date":"2025-09-29T05:22:35+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"221521326111222050786364942446355678075","date":"2025-09-28T03:55:30+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"27410670895391461222678615756194491160","date":"2025-09-28T02:33:51+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-09-27T22:07:18+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"237338476928934580792580042604081064587","date":"2025-09-26T17:34:28+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"269083824959143669894616684368089201341","date":"2025-09-26T15:11:19+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"147223436061517321898869480477417337612","date":"2025-09-26T08:45:54+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-09-26T02:04:50+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-07-01T07:11:08+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-07-01T07:10:16+00:00","index":"","fulltext":""},{"type":"submitted","content":"The Journal of Supercomputing","date":"2025-06-28T10:13:22+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"the-journal-of-supercomputing","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"Learn more about [The Journal of Supercomputing](https://www.springer.com/journal/11227)","snPcode":"11227","submissionUrl":"https://submission.nature.com/new-submission/11227/3","title":"The Journal of Supercomputing","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"cda3ffab-d210-41f3-b064-bda7261b528b","owner":[],"postedDate":"October 9th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2025-11-03T18:38:12+00:00","versionOfRecord":[],"versionCreatedAt":"2025-10-09 18:15:59","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6997149","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6997149","identity":"rs-6997149","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}