Dependency-Aware Adaptive Task Offloading inEdge-AI IoT Networks Using Transformer-DDQN

Dependency-Aware Adaptive Task Offloading inEdge-AI IoT Networks Using Transformer-DDQN | 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 Dependency-Aware Adaptive Task Offloading inEdge-AI IoT Networks Using Transformer-DDQN Ihsan Ullah, Qaisar Ali, Tejaswini Nagaraj This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8180895/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 19 You are reading this latest preprint version Abstract Efficient task scheduling and dependency management are essential for optimizing task offloading in Edge-AI-enabled IoT edge and cloud computing environments, where low latency, high computational efficiency, and energy awareness are critical. This paper proposes a deep reinforcement learning framework that integrates Graph Convolutional Networks (GCNs), Transformer networks, and a Double Deep Q-Network (DDQN) to handle complex task dependencies and dynamic workloads in real time. The GCN module models and prioritizes structural inter-task dependencies, while the Transformer captures temporal and contextual correlations across task sequences using multi-head self-attention, extracting long-range spatio-temporal features that represent workload variations and inter-task dependencies. These features are fed into the DDQN, which learns adaptive offloading and scheduling policies based on current network conditions, resource availability, and task urgency. The DDQN mitigates overestimation bias, enhancing convergence stability and decision precision. Experimental results demonstrate that the proposed GCN-Transformer-DDQN framework significantly reduces end-to-end latency, improves energy efficiency, and increases task throughput compared to existing methods. This architecture provides a scalable, dependency-aware solution for real-time task orchestration in Edge-AI-driven IoT environments. Edge AI Edge-Cloud Computing Task Offloading Dependency-Aware Scheduling Graph Convolutional Networks Transformer Double Deep Q-Network IoT Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Reviewers agreed at journal 19 May, 2026 Reviews received at journal 18 May, 2026 Reviews received at journal 18 May, 2026 Reviewers agreed at journal 17 May, 2026 Reviewers agreed at journal 17 May, 2026 Reviews received at journal 15 May, 2026 Reviewers agreed at journal 14 May, 2026 Reviewers agreed at journal 14 May, 2026 Reviewers agreed at journal 14 May, 2026 Reviewers agreed at journal 14 May, 2026 Reviewers agreed at journal 13 May, 2026 Reviewers agreed at journal 12 May, 2026 Reviewers agreed at journal 12 May, 2026 Reviewers agreed at journal 12 May, 2026 Reviewers agreed at journal 12 May, 2026 Reviewers invited by journal 12 May, 2026 Editor assigned by journal 12 May, 2026 Submission checks completed at journal 24 Nov, 2025 First submitted to journal 22 Nov, 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-8180895","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":639488831,"identity":"37c0b202-19ec-4a5f-b431-a8c8da7318b8","order_by":0,"name":"Ihsan Ullah","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA60lEQVRIiWNgGAWjYBACNgkQWYEqaECEljMMDDyoWhLw6AFpYWwjRQufdPvDz4Xz6uTtJXKfPWBsq0tsYG/eJsH44zBuh8mcMZaeue2wYY9EurkBYxtbYgPPsTIJhgQ8WiRyGKR5tx1g7JFIY5NgbONJbJDIMQNquY1HS/rj37xz6uyhWiQSG+TfENKSYCbN28CcCNViALSFh5CWHDNrnmOHk3vOPANqP5dg3MaTVmyRkPYfpxb5GemPb/PU1Nm2twNt+VBWJ9vPfnjjjQ82aTi1oIJENqC9IEYCkRqA4A/xSkfBKBgFo2DkAACZPkaJUZZ8qgAAAABJRU5ErkJggg==","orcid":"","institution":"Yuan Ze University","correspondingAuthor":true,"prefix":"","firstName":"Ihsan","middleName":"","lastName":"Ullah","suffix":""},{"id":639488832,"identity":"b3546179-1578-46c2-8039-500d6a37da1e","order_by":1,"name":"Qaisar Ali","email":"","orcid":"","institution":"Yuan Ze University","correspondingAuthor":false,"prefix":"","firstName":"Qaisar","middleName":"","lastName":"Ali","suffix":""},{"id":639488833,"identity":"5c1ebb8e-242e-4488-b9cb-1f68c9b859a7","order_by":2,"name":"Tejaswini Nagaraj","email":"","orcid":"","institution":"Yuan Ze University","correspondingAuthor":false,"prefix":"","firstName":"Tejaswini","middleName":"","lastName":"Nagaraj","suffix":""}],"badges":[],"createdAt":"2025-11-22 14:08:11","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8180895/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8180895/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":109248977,"identity":"9a99dcc9-3fd1-4a0e-baf9-a75ba94f5816","added_by":"auto","created_at":"2026-05-14 08:41:16","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":661041,"visible":true,"origin":"","legend":"","description":"","filename":"DependencyAwareAdaptiveTaskOffloadinginEdgeAIIoTNetworksUsingTransformerDDQNTejoJSC.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8180895/v1_covered_0f045b6f-6ba7-4ca0-987c-db981b4c1f1c.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Dependency-Aware Adaptive Task Offloading inEdge-AI IoT Networks Using Transformer-DDQN","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"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":"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":"Edge AI, Edge-Cloud Computing, Task Offloading, Dependency-Aware Scheduling, Graph Convolutional Networks, Transformer, Double Deep Q-Network, IoT","lastPublishedDoi":"10.21203/rs.3.rs-8180895/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8180895/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eEfficient task scheduling and dependency management are essential for optimizing task offloading in Edge-AI-enabled IoT edge and cloud computing environments, where low latency, high computational efficiency, and energy awareness are critical. This paper proposes a deep reinforcement learning framework that integrates Graph Convolutional Networks (GCNs), Transformer networks, and a Double Deep Q-Network (DDQN) to handle complex task dependencies and dynamic workloads in real time. The GCN module models and prioritizes structural inter-task dependencies, while the Transformer captures temporal and contextual correlations across task sequences using multi-head self-attention, extracting long-range spatio-temporal features that represent workload variations and inter-task dependencies. These features are fed into the DDQN, which learns adaptive offloading and scheduling policies based on current network conditions, resource availability, and task urgency. The DDQN mitigates overestimation bias, enhancing convergence stability and decision precision. Experimental results demonstrate that the proposed GCN-Transformer-DDQN framework significantly reduces end-to-end latency, improves energy efficiency, and increases task throughput compared to existing methods. This architecture provides a scalable, dependency-aware solution for real-time task orchestration in Edge-AI-driven IoT environments.\u003c/p\u003e","manuscriptTitle":"Dependency-Aware Adaptive Task Offloading inEdge-AI IoT Networks Using Transformer-DDQN","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-05-14 08:40:04","doi":"10.21203/rs.3.rs-8180895/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewerAgreed","content":"91222494127284367607109135331521634477","date":"2026-05-19T06:53:19+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-05-18T23:49:00+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-05-18T05:41:26+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"224134691684624829296380550931717713541","date":"2026-05-18T00:56:07+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"238379911074875828848611575905064910471","date":"2026-05-17T05:52:01+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-05-15T09:49:51+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"79741428189781189933738420616232324318","date":"2026-05-15T02:38:12+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"157007111815106736326322236557094471508","date":"2026-05-15T01:02:54+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"284859318207119268340469301627988993396","date":"2026-05-14T22:18:01+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"208214108175307311228537968117056420206","date":"2026-05-14T17:11:47+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"43682868463034875357555850756619023031","date":"2026-05-13T05:16:00+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"173000200771402305599130139422880194074","date":"2026-05-13T00:06:41+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"231750977509206671099297156125608667129","date":"2026-05-12T17:56:09+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"3737357402759769273515142756738196167","date":"2026-05-12T17:05:04+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"258282505261603676091082645892295670968","date":"2026-05-12T16:51:23+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-05-12T16:43:37+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-05-12T16:21:17+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-11-24T09:26:42+00:00","index":"","fulltext":""},{"type":"submitted","content":"The Journal of Supercomputing","date":"2025-11-22T13:55:21+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":"d630ba01-b643-48c6-934b-25ba095c0756","owner":[],"postedDate":"May 14th, 2026","published":true,"recentEditorialEvents":[{"type":"reviewerAgreed","content":"91222494127284367607109135331521634477","date":"2026-05-19T06:53:19+00:00","index":147,"fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-05-18T23:49:00+00:00","index":146,"fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-05-18T05:41:26+00:00","index":145,"fulltext":""},{"type":"reviewerAgreed","content":"224134691684624829296380550931717713541","date":"2026-05-18T00:56:07+00:00","index":142,"fulltext":""},{"type":"reviewerAgreed","content":"238379911074875828848611575905064910471","date":"2026-05-17T05:52:01+00:00","index":140,"fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-05-15T09:49:51+00:00","index":139,"fulltext":""},{"type":"reviewerAgreed","content":"79741428189781189933738420616232324318","date":"2026-05-15T02:38:12+00:00","index":138,"fulltext":""},{"type":"reviewerAgreed","content":"157007111815106736326322236557094471508","date":"2026-05-15T01:02:54+00:00","index":137,"fulltext":""},{"type":"reviewerAgreed","content":"284859318207119268340469301627988993396","date":"2026-05-14T22:18:01+00:00","index":134,"fulltext":""},{"type":"reviewerAgreed","content":"208214108175307311228537968117056420206","date":"2026-05-14T17:11:47+00:00","index":131,"fulltext":""},{"type":"reviewerAgreed","content":"43682868463034875357555850756619023031","date":"2026-05-13T05:16:00+00:00","index":126,"fulltext":""},{"type":"reviewerAgreed","content":"173000200771402305599130139422880194074","date":"2026-05-13T00:06:41+00:00","index":122,"fulltext":""},{"type":"reviewerAgreed","content":"231750977509206671099297156125608667129","date":"2026-05-12T17:56:09+00:00","index":121,"fulltext":""},{"type":"reviewerAgreed","content":"3737357402759769273515142756738196167","date":"2026-05-12T17:05:04+00:00","index":119,"fulltext":""},{"type":"reviewerAgreed","content":"258282505261603676091082645892295670968","date":"2026-05-12T16:51:23+00:00","index":118,"fulltext":""},{"type":"reviewersInvited","content":"112","date":"2026-05-12T16:43:37+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-05-12T16:21:17+00:00","index":"","fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-05-14T08:40:04+00:00","versionOfRecord":[],"versionCreatedAt":"2026-05-14 08:40:04","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8180895","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8180895","identity":"rs-8180895","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}