𝙏𝙖𝙨𝙠 𝙊𝙛𝙛𝙡𝙤𝙖𝙙𝙞𝙣𝙜 𝙛𝙤𝙧 𝙀𝙙𝙜𝙚 𝘾𝙤𝙢𝙥𝙪𝙩𝙞𝙣𝙜 𝘽𝙖𝙨𝙚𝙙 𝙤𝙣 𝘿𝙚𝙚𝙥 𝙍𝙚𝙞𝙣𝙛𝙤𝙧𝙘𝙚𝙢𝙚𝙣𝙩 𝙇𝙚𝙖𝙧𝙣𝙞𝙣𝙜
Why should you read this research article?
▶️ Proposes a
#DeepReinforcementLearning (
#DRL)-based
#TaskOffloading framework for
#EdgeComputing that explicitly models inter-task dependencies via
#DirectedAcyclicGraphs (
#DAG), addressing a critical gap in existing methods that treat tasks as independent.
▶️Formulates the offloading problem as a
#MarkovDecisionProcess (
#MDP) with energy consumption and latency as joint reward signals, enabling fine-grained offloading decisions that balance competing performance objectives in dynamic network environments.
▶️Introduces a
#Seq2Seq architecture with a
#BidirectionalGRU (
#BiGRU) encoder and attention mechanism, enabling the model to adaptively assign weights to different task actions and avoid falling into local optima during the offloading process.
▶️Demonstrates superior performance across various DAG topologies — at 20 Mbps, latency is 2.3% lower than the best comparison algorithm; at 8 Mbps with 45 task nodes,
#QoS is 6.08% higher — with strong convergence and stability throughout.
▶️Lays a clear foundation for future research in multi-user, multi-edge-server systems and 5G-enabled "end-edge-cloud" heterogeneous networks, where low latency, high reliability, and efficient resource management are critical.
Register a FREE account today at World Scientific to read this article. 𝐕𝐚𝐥𝐢𝐝 𝐭𝐢𝐥𝐥 𝟑𝟎 𝐒𝐞𝐩𝐭𝐞𝐦𝐛𝐞𝐫 𝟐𝟎𝟐𝟔!
Recommend this journal to your library today. We will be happy to work with your librarian to get you the necessary access — email us at mailto:cjlim
@wspc.com to kickstart this access today!
